# Taboola.com > Performance advertising at scale on the Open Web, AI bidding and optimization, predictive audiences, generative AI creative, display, vertical, carousel, and motion ads formats. Website: https://www.taboola.com --- ## Pages ### Taboola News URL: https://www.taboola.com/taboola-news Last Modified: 2025-02-06 14:56:29 --- ### Our Story URL: https://www.taboola.com/about/our-story Last Modified: 2025-05-27 04:04:08 --- ### Press Center URL: https://www.taboola.com/taboola-story-hub Last Modified: 2022-08-11 19:36:01 --- ### About URL: https://www.taboola.com/about Last Modified: 2020-06-22 16:09:49 --- ### Demo Page URL: https://www.taboola.com/demo Last Modified: 2023-02-19 09:37:55 --- ### Branding URL: https://www.taboola.com/branding Last Modified: 2024-06-04 05:39:58 --- ### Demo Preview URL: https://www.taboola.com/demo-preview Last Modified: 2021-02-24 01:20:45 --- ### Newsroom URL: https://www.taboola.com/newsroom Last Modified: 2022-06-14 08:42:55 --- ### Native Advertising URL: https://www.taboola.com/native-advertising Last Modified: 2025-02-20 05:21:46 --- ### Publishers URL: https://www.taboola.com/publishers Last Modified: 2021-05-30 07:49:45 --- ### Resources URL: https://www.taboola.com/resources Last Modified: 2020-07-13 04:31:11 --- ### Contact URL: https://www.taboola.com/contact Last Modified: 2024-07-11 07:05:09 --- ### Thank You URL: https://www.taboola.com/thank-you-contact Last Modified: 2021-02-24 01:34:40 --- ### Board URL: https://www.taboola.com/about/board Last Modified: 2025-06-04 03:33:52 --- ### Report URL: https://www.taboola.com/report Last Modified: 2025-08-06 03:25:20 --- ### Video Portfolio URL: https://www.taboola.com/video-portfolio Last Modified: 2020-06-02 11:34:36 --- ### Ebooks URL: https://www.taboola.com/resources/e-book Last Modified: 2020-07-19 11:51:57 --- ### Best Practices URL: https://www.taboola.com/resources/best-practices Last Modified: 2020-07-19 11:50:48 --- ### Taboola Webinars URL: https://www.taboola.com/resources/webinars Last Modified: 2023-07-12 06:08:18 --- ### All Press Releases URL: https://www.taboola.com/all-press-releases Last Modified: 2022-08-15 06:58:38 --- ### All News Items URL: https://www.taboola.com/all-news-items Last Modified: 2020-07-27 09:16:17 --- ### wibbitz URL: https://www.taboola.com/wibbitz Last Modified: 2021-03-16 11:50:26 <!doctype html> --- ### amp documentation URL: https://www.taboola.com/amp-documentation Last Modified: 2020-08-30 19:47:45 --- ### Policies URL: https://www.taboola.com/policies Last Modified: 2020-10-25 08:21:12 --- ### Awareness URL: https://www.taboola.com/advertise/awareness Last Modified: 2024-06-03 02:35:56 --- ### Consideration URL: https://www.taboola.com/advertise/consideration Last Modified: 2022-01-30 08:23:14 --- ### Conversions URL: https://www.taboola.com/advertise/conversions Last Modified: 2022-01-27 14:21:26 --- ### Careers URL: https://www.taboola.com/careers Last Modified: 2025-12-08 08:12:40 --- ### Teams URL: https://www.taboola.com/careers/teams Last Modified: 2021-02-08 14:06:10 --- ### Discover Our Moments URL: https://www.taboola.com/careers/discover Last Modified: 2022-09-08 15:01:13 --- ### Explore All Jobs URL: https://www.taboola.com/careers/jobs Last Modified: 2021-02-08 14:04:45 --- ### Guilding Principles URL: https://www.taboola.com/careers/principles Last Modified: 2021-02-08 14:05:02 --- ### Our Locations URL: https://www.taboola.com/careers/our-locations Last Modified: 2021-02-08 14:06:00 --- ### Offer Book URL: https://www.taboola.com/careers/offer-book Last Modified: 2023-01-24 12:55:39 --- ### Offer Book APAC URL: https://www.taboola.com/careers/offer-book-apac Last Modified: 2021-04-29 08:06:40 --- ### Offer Book EMEA URL: https://www.taboola.com/careers/offer-book-emea Last Modified: 2021-04-29 08:07:07 --- ### Taboola High Impact URL: https://www.taboola.com/advertise/high-impact-awareness Last Modified: 2023-08-03 07:56:40 --- ### Taboola Product interview URL: https://www.taboola.com/taboola-product-interview Last Modified: 2023-01-10 08:11:03 --- ### Social Responsibility URL: https://www.taboola.com/about/social-responsibility Last Modified: 2024-07-10 07:55:33 --- ### Taboola R&D interview URL: https://www.taboola.com/taboola-rnd-interview Last Modified: 2022-03-28 07:45:38 --- ### Professional Service Recruiting Process URL: https://www.taboola.com/professional-service-recruiting-process Last Modified: 2022-12-28 12:28:47 --- ### Creative Shop URL: https://www.taboola.com/creative-shop Last Modified: 2024-09-19 04:09:19 --- ### Demo URL: https://www.taboola.com/demo/ Last Modified: 2025-02-26 07:55:16 --- ### Demo Preview URL: https://www.taboola.com/demo-preview/ Last Modified: 2025-02-26 07:55:16 --- ### Advertiser URL: https://www.taboola.com/advertise/ Last Modified: 2026-05-14 11:39:02 --- ### Publishers URL: https://www.taboola.com/publishers/ Last Modified: 2026-04-09 07:05:38 --- ### Creative Shop URL: https://www.taboola.com/creative-shop/ Last Modified: 2026-07-29 13:45:56 --- ### Glossary URL: https://www.taboola.com/glossary/ Last Modified: 2026-02-02 12:48:11 --- ### Taboola Story Hub URL: https://www.taboola.com/taboola-story-hub/ Last Modified: 2025-08-21 07:36:19 --- ### Contact URL: https://www.taboola.com/contact/ Last Modified: 2026-02-04 10:38:13 --- ### Management URL: https://www.taboola.com/management/ Last Modified: 2026-08-06 15:26:30 --- ### Realize Pro URL: https://www.taboola.com/realize-pro/ Last Modified: 2025-11-12 12:27:29 --- ### Referral URL: https://www.taboola.com/referral/ Last Modified: 2026-02-09 09:51:10 --- ### Social Responsibility URL: https://www.taboola.com/social-responsibility/ Last Modified: 2026-01-14 14:25:42 --- ### DeeperDive URL: https://www.taboola.com/deeperdive/ Last Modified: 2026-06-16 11:48:15 --- ### Newsroom URL: https://www.taboola.com/newsroom/ Last Modified: 2026-06-03 13:51:06 --- ## Recorded Webinars ### Taboola's New Products for Advertisers URL: https://www.taboola.com/recorded-webinars/taboolas-new-products-for-advertisers Last Modified: 2024-08-01 11:10:11 --- ### Campaign Strategies to Sleigh the Holidays URL: https://www.taboola.com/recorded-webinars/upcoming-webinar-campaign-strategies-to-sleigh-the-holidays Last Modified: 2024-08-01 11:05:46 --- ### The Ultimate Q&A Session: How TO Scale with Native Ads URL: https://www.taboola.com/recorded-webinars/the-ultimate-qa-session-how-to-scale-with-native-ads Last Modified: 2024-08-01 11:06:52 --- ### Post-Click Strategies for Performance Campaigns URL: https://www.taboola.com/recorded-webinars/post-click-strategies-for-performance-campaigns Last Modified: 2024-08-01 11:07:40 --- ### The Ultimate Q&A Session: Maximize Native Ad Performance Across the Funnel URL: https://www.taboola.com/recorded-webinars/the-ultimate-qa-session-maximize-native-ad-performance-across-the-funnel Last Modified: 2024-08-01 11:08:32 --- ### GA4 Masterclass: What Marketers Need to Know Now URL: https://www.taboola.com/recorded-webinars/ga4-masterclass-what-marketers-need-to-know-now Last Modified: 2024-08-01 11:09:30 --- ### Maximiere den ROI: Wie du Social-Media-Kampagnen in Native Anzeigen umwandelst URL: https://www.taboola.com/recorded-webinars/maximiere-den-roi-wie-du-social-media-kampagnen-in-native-anzeigen-umwandelst Last Modified: 2024-08-01 11:15:09 --- ### Maximize ROI: Repurpose Social Campaigns for Native Ads URL: https://www.taboola.com/recorded-webinars/maximize-roi-repurpose-social-campaigns-for-native-ads Last Modified: 2024-08-01 11:11:05 --- ### Core Web Vitals + Cookieless World URL: https://www.taboola.com/recorded-webinars/core-web-vitals-cookieless-world Last Modified: 2024-08-01 11:17:42 --- ### Context Ist Der Sahnebecher URL: https://www.taboola.com/recorded-webinars/context-ist-der-sahnebecher Last Modified: 2024-08-01 11:17:17 --- ### The Need for Diverse Advertising URL: https://www.taboola.com/recorded-webinars/the-need-for-diverse-advertising Last Modified: 2024-08-01 11:19:04 --- ### Navigating Shifts in Ecommerce URL: https://www.taboola.com/recorded-webinars/navigating-shifts-in-ecommerce Last Modified: 2024-08-01 11:19:55 --- ### Moments of Next: Why, When, and Where They Happen URL: https://www.taboola.com/recorded-webinars/moments-of-next-why-when-and-where-they-happen Last Modified: 2021-10-20 20:06:05 --- ### Introducing the Moment of Next URL: https://www.taboola.com/recorded-webinars/introducing-the-moment-of-next Last Modified: 2021-10-20 20:05:50 --- ### Optimizing Your Video Creatives for Content Discovery URL: https://www.taboola.com/recorded-webinars/optimizing-your-video-creatives-for-content-discovery Last Modified: 2024-08-01 11:41:48 --- ### How Brands Communicate in Times of Crisis URL: https://www.taboola.com/recorded-webinars/how-brands-communicate-in-times-of-crisis Last Modified: 2021-10-20 20:05:39 --- ### Developing a Winning Content Marketing Strategy–Quickly URL: https://www.taboola.com/recorded-webinars/developing-a-winning-content-marketing-strategy%e2%80%93quickly Last Modified: 2024-08-01 11:42:50 --- ### Shift Your Auto Marketing Into Overdrive Auto Webinar URL: https://www.taboola.com/recorded-webinars/shift-your-auto-marketing-into-overdrive-auto-webinar Last Modified: 2021-10-20 20:08:27 --- ### Marketing at the Speed of Fashion and Beauty Webinar URL: https://www.taboola.com/recorded-webinars/marketing-at-the-speed-of-fashion-and-beauty-webinar Last Modified: 2021-10-20 20:05:58 --- ### What Creatives Drive the Best Ad Performance? URL: https://www.taboola.com/recorded-webinars/what-creatives-drive-the-best-ad-performance Last Modified: 2021-10-20 20:09:08 --- ### Tips to Increase Native Ad Performance URL: https://www.taboola.com/recorded-webinars/tips-to-increase-native-ad-performance Last Modified: 2021-10-20 20:09:03 --- ### What Makes Good Content? URL: https://www.taboola.com/recorded-webinars/what-makes-good-content Last Modified: 2021-10-20 20:09:12 --- ## Case Study ### Mercedes-Benz & OMD Taiwan URL: https://www.taboola.com/resources/case-studies/mercedes-benz-omd-taiwan Last Modified: 2026-06-15 03:51:55 --- ### Adnimation URL: https://www.taboola.com/resources/case-studies/adnimation Last Modified: 2026-06-08 06:37:57 --- ### Akbank & Optdcom URL: https://www.taboola.com/resources/case-studies/akbank-optdcom Last Modified: 2026-05-26 06:01:18 --- ### Embracon URL: https://www.taboola.com/resources/case-studies/embracon Last Modified: 2026-04-23 06:48:48 --- ### El Corte Inglés URL: https://www.taboola.com/resources/case-studies/el-corte-ingles Last Modified: 2026-03-16 13:09:00 --- ### MM New Media URL: https://www.taboola.com/resources/case-studies/mm-new-media Last Modified: 2026-03-05 07:59:46 --- ### Olight URL: https://www.taboola.com/resources/case-studies/olight Last Modified: 2026-02-23 08:26:53 --- ### El HuffPost URL: https://www.taboola.com/resources/case-studies/el-huffpost Last Modified: 2026-02-17 08:59:41 --- ### Good Neighbors URL: https://www.taboola.com/resources/case-studies/good-neighbors Last Modified: 2026-03-05 08:06:49 --- ### Mitsubishi & Cadastra URL: https://www.taboola.com/resources/case-studies/mitsubishi-cadastra Last Modified: 2026-01-22 09:48:07 --- ### Ypê & Zmes URL: https://www.taboola.com/resources/case-studies/ype-zmes Last Modified: 2026-01-02 04:37:25 --- ### The Independent URL: https://www.taboola.com/resources/case-studies/the-independent Last Modified: 2025-12-23 16:30:51 --- ### Manorama Online URL: https://www.taboola.com/resources/case-studies/manorama-online Last Modified: 2025-12-17 09:02:23 --- ### Cruise Critic URL: https://www.taboola.com/resources/case-studies/cruise-critic Last Modified: 2025-12-12 17:13:29 --- ### Ströer URL: https://www.taboola.com/resources/case-studies/stroer Last Modified: 2025-10-28 06:54:07 --- ### Minor Hotels URL: https://www.taboola.com/resources/case-studies/minor-hotels Last Modified: 2025-10-14 04:04:17 --- ### USA TODAY URL: https://www.taboola.com/resources/case-studies/usa-today Last Modified: 2025-10-07 16:50:56 --- ### Verisure Chile URL: https://www.taboola.com/resources/case-studies/verisure-chile Last Modified: 2025-09-17 11:29:47 --- ### Livguard URL: https://www.taboola.com/resources/case-studies/livguard Last Modified: 2025-09-23 09:25:26 --- ### Ashiana Housing Ltd. URL: https://www.taboola.com/resources/case-studies/ashiana-housing-ltd Last Modified: 2025-09-02 09:42:28 --- ### Lankasri URL: https://www.taboola.com/resources/case-studies/lankasri Last Modified: 2025-08-13 08:41:09 --- ### REV Media Group URL: https://www.taboola.com/resources/case-studies/rev-media-group Last Modified: 2025-08-11 14:46:00 --- ### Philips URL: https://www.taboola.com/resources/case-studies/philips-home-appliances Last Modified: 2025-08-11 14:46:25 --- ### Unilever URL: https://www.taboola.com/resources/case-studies/unilever-project-agora Last Modified: 2025-08-11 14:46:42 --- ### Chery - Hebrew URL: https://www.taboola.com/resources/case-studies/chery-hebrew Last Modified: 2025-06-20 01:43:37 --- ### Meitav - Hebrew URL: https://www.taboola.com/resources/case-studies/meitav-hebrew Last Modified: 2025-06-20 01:07:32 --- ### NYDJ & iQuanti URL: https://www.taboola.com/resources/case-studies/nydj-iquanti Last Modified: 2025-08-15 03:53:32 --- ### El Nacional URL: https://www.taboola.com/resources/case-studies/el-nacional.cat Last Modified: 2025-06-13 10:32:35 --- ### Ziwo URL: https://www.taboola.com/resources/case-studies/ziwo Last Modified: 2025-06-11 03:25:55 --- ### Meitav URL: https://www.taboola.com/resources/case-studies/meitav Last Modified: 2025-06-10 14:48:37 --- ### Chery URL: https://www.taboola.com/resources/case-studies/chery Last Modified: 2025-06-10 11:24:07 --- ### AIDA & Initiative URL: https://www.taboola.com/resources/case-studies/aida-initiative Last Modified: 2025-08-11 14:32:17 --- ### PortAventura World URL: https://www.taboola.com/resources/case-studies/portaventura-world Last Modified: 2025-06-17 10:17:11 --- ### Madame Coco & Ingage URL: https://www.taboola.com/resources/case-studies/madame-coco Last Modified: 2025-06-11 05:21:57 --- ### Medios Deportivos URL: https://www.taboola.com/resources/case-studies/medios-deportivos Last Modified: 2025-05-14 10:28:48 --- ### The Indian Express URL: https://www.taboola.com/resources/case-studies/the-indian-express Last Modified: 2025-05-14 10:29:17 --- ### Nation Media Group URL: https://www.taboola.com/resources/case-studies/nation-media-group Last Modified: 2025-05-14 10:29:37 --- ### Gazeta São Paulo URL: https://www.taboola.com/resources/case-studies/gazeta-sao-paulo Last Modified: 2025-04-16 17:15:59 --- ### GMF URL: https://www.taboola.com/resources/case-studies/gmf Last Modified: 2025-03-24 13:02:30 --- ### Verisure URL: https://www.taboola.com/resources/case-studies/verisure Last Modified: 2025-03-11 16:07:56 --- ### Blancheporte & Schwartz Consulting URL: https://www.taboola.com/resources/case-studies/blancheporte-schwartz-consulting Last Modified: 2025-03-06 11:02:59 --- ### Apartments.com URL: https://www.taboola.com/resources/case-studies/apartments.com Last Modified: 2025-02-27 12:12:27 --- ### H2Bet URL: https://www.taboola.com/resources/case-studies/h2bet Last Modified: 2025-04-07 14:23:01 --- ### Leckerschmecker.me URL: https://www.taboola.com/resources/case-studies/leckerschmecker.me Last Modified: 2025-01-13 17:01:54 --- ### Peugeot & Publicis Media URL: https://www.taboola.com/resources/case-studies/peugeot-publicis-media Last Modified: 2025-08-13 08:36:09 --- ### Vivara & Media.Monks URL: https://www.taboola.com/resources/case-studies/vivara-media.monks Last Modified: 2025-01-02 10:54:39 --- ### Open English URL: https://www.taboola.com/resources/case-studies/open-english Last Modified: 2024-12-17 10:11:42 --- ### Crucial & Spark Foundry URL: https://www.taboola.com/resources/case-studies/crucial-spark-foundry Last Modified: 2024-11-20 10:44:02 --- ### Amplifon URL: https://www.taboola.com/resources/case-studies/amplifon Last Modified: 2024-11-12 14:27:53 --- ### Nutrientes VidaLabs URL: https://www.taboola.com/resources/case-studies/nutrientes-vidalabs Last Modified: 2024-11-11 12:58:07 --- ### HuffPost Korea URL: https://www.taboola.com/resources/case-studies/huffpost-kr Last Modified: 2024-11-08 12:33:57 --- ### Lidl Hellas & Project Agora URL: https://www.taboola.com/resources/case-studies/lidl-hellas-project-agora Last Modified: 2024-11-07 12:39:03 --- ### abrdn Investments & Starcom URL: https://www.taboola.com/resources/case-studies/abrdn-starcom Last Modified: 2024-11-05 12:27:07 --- ### Wyborkierowcow.pl URL: https://www.taboola.com/resources/case-studies/wyborkierowcow.pl Last Modified: 2024-11-04 14:21:56 --- ### Piramal Realty & Realatte Ventures URL: https://www.taboola.com/resources/case-studies/piramal-realty-realatte-ventures Last Modified: 2024-10-03 10:27:26 --- ### ballnews media URL: https://www.taboola.com/resources/case-studies/ballnews-media Last Modified: 2024-09-13 13:11:27 --- ### Kavak URL: https://www.taboola.com/resources/case-studies/kavak Last Modified: 2024-09-13 12:19:28 --- ### Grupo Crónica URL: https://www.taboola.com/resources/case-studies/grupo-cronica Last Modified: 2024-09-25 15:05:01 --- ### Motor Culture Australia URL: https://www.taboola.com/resources/case-studies/motor-culture-australia Last Modified: 2024-08-27 12:45:48 --- ### Seattle Times Media Solutions URL: https://www.taboola.com/resources/case-studies/seattle-times-media-solutions Last Modified: 2024-07-16 13:54:04 --- ### Citroën & Publicis URL: https://www.taboola.com/resources/case-studies/citroen-publicis Last Modified: 2024-07-10 12:29:24 --- ### 7NEWS URL: https://www.taboola.com/resources/case-studies/7news Last Modified: 2024-07-01 18:09:13 --- ### Health IQ Communications URL: https://www.taboola.com/resources/case-studies/health-iq-communications Last Modified: 2024-07-08 11:23:37 --- ### The Independent URL: https://www.taboola.com/resources/case-studies/the-independent-uk Last Modified: 2024-06-04 10:28:56 --- ### The Independent URL: https://www.taboola.com/resources/case-studies/the-independent-us Last Modified: 2024-06-04 10:31:16 --- ### Yellow Cake Media URL: https://www.taboola.com/resources/case-studies/yellow-cake-media Last Modified: 2024-05-23 11:30:12 --- ### Bedrop URL: https://www.taboola.com/resources/case-studies/bedrop Last Modified: 2024-05-21 10:54:58 --- ### MacLucan & Panda Security URL: https://www.taboola.com/resources/case-studies/maclucan-panda-security Last Modified: 2024-05-17 08:15:13 --- ### Gray Television URL: https://www.taboola.com/resources/case-studies/gray-television Last Modified: 2024-05-08 10:21:25 --- ### Yahoo & Travel Company URL: https://www.taboola.com/resources/case-studies/yahoo-travel-company Last Modified: 2024-04-30 14:30:07 --- ### Yahoo & Tech Company URL: https://www.taboola.com/resources/case-studies/yahoo-and-tech-company Last Modified: 2024-04-24 15:49:43 --- ### Enpal URL: https://www.taboola.com/resources/case-studies/enpal Last Modified: 2024-05-10 11:41:11 --- ### Ezer Mizion & Digital Vibe - Hebrew URL: https://www.taboola.com/resources/case-studies/ezer-mizion-digital-vibe-hebrew Last Modified: 2024-03-20 06:17:36 --- ### Hyundai URL: https://www.taboola.com/resources/case-studies/hyundai Last Modified: 2024-03-19 12:39:12 --- ### Ezer Mizion & Digital Vibe URL: https://www.taboola.com/resources/case-studies/ezer-mizion-digital-vibe Last Modified: 2024-03-14 11:52:12 --- ### Yahoo & Financial Company URL: https://www.taboola.com/resources/case-studies/yahoo-financial-company-2 Last Modified: 2024-04-24 10:57:48 --- ### One Zero Bank - Hebrew URL: https://www.taboola.com/resources/case-studies/one-zero-bank-hebrew Last Modified: 2024-03-02 04:46:39 --- ### Yahoo & Financial Company URL: https://www.taboola.com/resources/case-studies/yahoo-financial-company Last Modified: 2024-04-24 11:00:32 --- ### HPE Automotores & Media.Monks URL: https://www.taboola.com/resources/case-studies/hpe-automotores-media.monks Last Modified: 2024-02-26 11:02:54 --- ### One Zero Bank URL: https://www.taboola.com/resources/case-studies/one-zero-bank Last Modified: 2024-02-16 16:39:18 --- ### Muang Thai Life Assurance PCL URL: https://www.taboola.com/resources/case-studies/muang-thai-life-assurance-pcl Last Modified: 2024-02-05 16:44:05 --- ### HPE Automotores & Media.Monks URL: https://www.taboola.com/resources/case-studies/hpe-media.monks Last Modified: 2024-01-29 12:59:43 --- ### Vodafone Turkey URL: https://www.taboola.com/resources/case-studies/vodafone-turkey Last Modified: 2024-01-31 13:52:41 --- ### Realtime Agency URL: https://www.taboola.com/resources/case-studies/realtime-agency Last Modified: 2024-01-16 14:52:02 --- ### Dziennik.pl URL: https://www.taboola.com/resources/case-studies/dziennik.pl Last Modified: 2024-01-04 12:01:13 --- ### Renault On Demand URL: https://www.taboola.com/resources/case-studies/renault-on-demand Last Modified: 2024-01-08 12:17:35 --- ### Digital Athlete URL: https://www.taboola.com/resources/case-studies/digital-athlete Last Modified: 2024-03-11 07:48:45 --- ### BORA & Territory Media URL: https://www.taboola.com/resources/case-studies/bora-territory-media Last Modified: 2023-12-22 10:52:13 --- ### Ancestry & Performics URL: https://www.taboola.com/resources/case-studies/ancestry-performics Last Modified: 2023-12-19 15:15:23 --- ### Click Crew Media URL: https://www.taboola.com/resources/case-studies/click-crew-media Last Modified: 2023-11-29 12:14:36 --- ### ABOUT YOU URL: https://www.taboola.com/resources/case-studies/about-you Last Modified: 2024-01-09 11:04:38 --- ### BlazePod URL: https://www.taboola.com/resources/case-studies/blazepod Last Modified: 2023-10-23 13:55:58 --- ### Seven.One Entertainment Group URL: https://www.taboola.com/resources/case-studies/seven.one-entertainment-group Last Modified: 2023-11-06 12:07:27 --- ### Notix URL: https://www.taboola.com/resources/case-studies/notix Last Modified: 2023-10-11 12:26:50 --- ### Resolution URL: https://www.taboola.com/resources/case-studies/resolution Last Modified: 2023-09-29 12:47:48 --- ### AVVA URL: https://www.taboola.com/resources/case-studies/avva Last Modified: 2023-09-26 12:45:31 --- ### ICICI Prudential Life URL: https://www.taboola.com/resources/case-studies/icici-prudential-life Last Modified: 2023-09-12 11:36:25 --- ### Tadiran - HE URL: https://www.taboola.com/resources/case-studies/tadiran-he Last Modified: 2023-09-11 03:37:14 --- ### Planit & Vodafone URL: https://www.taboola.com/resources/case-studies/planit-vodafone Last Modified: 2023-12-08 11:46:42 --- ### Barceló Hotel Group URL: https://www.taboola.com/resources/case-studies/barcelo-hotel-group Last Modified: 2023-08-24 11:02:46 --- ### Ferrero Group & Publicis URL: https://www.taboola.com/resources/case-studies/ferrero-group-publicis Last Modified: 2023-08-25 14:07:27 --- ### McClatchy URL: https://www.taboola.com/resources/case-studies/mcclatchy Last Modified: 2023-12-20 14:00:28 --- ### Ariş Pırlanta & Neodigital URL: https://www.taboola.com/resources/case-studies/aris-pirlanta-neodigital Last Modified: 2023-07-27 12:05:38 --- ### Bombas URL: https://www.taboola.com/resources/case-studies/bombas Last Modified: 2023-07-19 11:59:06 --- ### Fat Tail Investment Research URL: https://www.taboola.com/resources/case-studies/fat-tail-investment-research-2 Last Modified: 2023-07-27 10:10:58 --- ### LBS and Hearts & Science Germany URL: https://www.taboola.com/resources/case-studies/lbs-and-hearts-science-germany Last Modified: 2023-07-11 14:08:04 --- ### Mayne Pharma & EVERSANA INTOUCH® URL: https://www.taboola.com/resources/case-studies/mayne-pharma-eversana-intouch Last Modified: 2023-06-29 14:43:51 --- ### Sirloin & NHN Commerce URL: https://www.taboola.com/resources/case-studies/sirloin-nhn-commerce Last Modified: 2023-06-06 11:22:23 --- ### Kapten & Son URL: https://www.taboola.com/resources/case-studies/kapten-son Last Modified: 2023-06-09 13:05:18 --- ### ynet URL: https://www.taboola.com/resources/case-studies/ynet-2 Last Modified: 2023-05-24 12:48:38 --- ### BluTV and Hype URL: https://www.taboola.com/resources/case-studies/blutv-and-hype Last Modified: 2023-05-17 16:04:43 --- ### Matterkind and The Economist URL: https://www.taboola.com/resources/case-studies/matterkind-and-the-economist Last Modified: 2023-05-05 16:24:45 --- ### The Greek National Tourism Organization and Project Agora URL: https://www.taboola.com/resources/case-studies/the-greek-national-tourism-organization-and-project-agora Last Modified: 2023-05-03 18:20:53 --- ### Beleza na Web URL: https://www.taboola.com/resources/case-studies/beleza-na-web Last Modified: 2023-05-01 15:26:54 --- ### Vodafone URL: https://www.taboola.com/resources/case-studies/vodafone-2 Last Modified: 2023-04-26 20:44:22 --- ### Spinny URL: https://www.taboola.com/resources/case-studies/spinny Last Modified: 2023-04-26 19:06:47 --- ### The Loop URL: https://www.taboola.com/resources/case-studies/the-loop Last Modified: 2023-03-29 12:55:35 --- ### SciPlay URL: https://www.taboola.com/resources/case-studies/sciplay Last Modified: 2024-09-17 13:44:03 --- ### BidBerry URL: https://www.taboola.com/resources/case-studies/bidberry Last Modified: 2023-08-22 06:55:31 --- ### Kaspersky URL: https://www.taboola.com/resources/case-studies/kaspersky Last Modified: 2023-02-06 15:05:07 --- ### Renault Australia URL: https://www.taboola.com/resources/case-studies/renault-australia Last Modified: 2023-01-27 17:43:48 --- ### Telangana Publications URL: https://www.taboola.com/resources/case-studies/telangana-publications Last Modified: 2022-12-22 20:11:02 --- ### MVF & The Eco Experts URL: https://www.taboola.com/resources/case-studies/mvf-the-eco-experts Last Modified: 2023-01-06 19:11:28 --- ### Outlook India URL: https://www.taboola.com/resources/case-studies/outlook-india Last Modified: 2023-01-06 19:12:25 --- ### Matas URL: https://www.taboola.com/resources/case-studies/matas Last Modified: 2023-01-06 19:13:05 --- ### DirectAsia URL: https://www.taboola.com/resources/case-studies/directasia Last Modified: 2022-11-23 13:58:49 --- ### Zazume URL: https://www.taboola.com/resources/case-studies/zazume Last Modified: 2022-10-26 15:19:37 --- ### AlgaeCal URL: https://www.taboola.com/resources/case-studies/algaecal Last Modified: 2022-10-26 13:06:04 --- ### Version Two URL: https://www.taboola.com/resources/case-studies/version-two Last Modified: 2022-10-18 12:52:34 --- ### CNNturk.com URL: https://www.taboola.com/resources/case-studies/cnnturk.com Last Modified: 2022-10-10 12:54:14 --- ### ALM Corp & Multinet Up URL: https://www.taboola.com/resources/case-studies/alm-corp-multinet Last Modified: 2022-10-26 13:53:17 --- ### The Independent Singapore URL: https://www.taboola.com/resources/case-studies/the-independent-singapore Last Modified: 2022-09-09 14:30:26 --- ### KIA URL: https://www.taboola.com/resources/case-studies/kia Last Modified: 2022-09-23 13:36:49 --- ### Durex & Havas URL: https://www.taboola.com/resources/case-studies/durex-havas Last Modified: 2022-08-31 19:45:01 --- ### Tadiran URL: https://www.taboola.com/resources/case-studies/tadiran Last Modified: 2022-08-14 07:14:26 --- ### Estados de Minas URL: https://www.taboola.com/resources/case-studies/estado-de-minas-aumenta Last Modified: 2026-07-30 07:32:12 --- ### Capital Media URL: https://www.taboola.com/resources/case-studies/capital-media Last Modified: 2022-09-02 17:30:05 --- ### Xaxis URL: https://www.taboola.com/resources/case-studies/xaxis Last Modified: 2022-12-22 19:28:15 --- ### Tops Online URL: https://www.taboola.com/resources/case-studies/tops-online Last Modified: 2022-08-09 17:43:27 --- ### BlueBet URL: https://www.taboola.com/resources/case-studies/bluebet Last Modified: 2022-06-10 18:09:01 --- ### Aegon Seguros URL: https://www.taboola.com/resources/case-studies/aegon-seguros-es Last Modified: 2022-08-09 17:18:01 --- ### Aegon Seguros URL: https://www.taboola.com/resources/case-studies/aegon-seguros Last Modified: 2022-08-09 17:18:09 --- ### Colvin ES URL: https://www.taboola.com/resources/case-studies/colvin-4 Last Modified: 2022-08-09 17:35:24 --- ### Colvin URL: https://www.taboola.com/resources/case-studies/colvin Last Modified: 2022-08-09 17:35:31 --- ### Grupo Atlántida URL: https://www.taboola.com/resources/case-studies/grupo-atlantida-2 Last Modified: 2022-07-26 18:55:53 --- ### Grupo Atlántida URL: https://www.taboola.com/resources/case-studies/grupo-atlantida Last Modified: 2022-08-09 17:50:07 --- ### SOMOS INVICTOS URL: https://www.taboola.com/resources/case-studies/somos-invictos Last Modified: 2022-08-09 17:50:15 --- ### StartEngine & Island Brands URL: https://www.taboola.com/resources/case-studies/startengine-island-brands Last Modified: 2022-08-09 17:45:48 --- ### Stoppa Pressarna URL: https://www.taboola.com/resources/case-studies/stoppa-pressarna Last Modified: 2026-06-11 04:48:17 --- ### Sobha Dubai URL: https://www.taboola.com/resources/case-studies/sobha-dubai Last Modified: 2022-07-26 18:48:02 --- ### AdCombo URL: https://www.taboola.com/resources/case-studies/adcombo Last Modified: 2022-08-11 13:15:47 --- ### Zupee URL: https://www.taboola.com/resources/case-studies/zupee Last Modified: 2022-08-09 17:19:08 --- ### udn.com URL: https://www.taboola.com/resources/case-studies/udn.com Last Modified: 2022-08-09 17:50:47 --- ### Menora URL: https://www.taboola.com/resources/case-studies/menora Last Modified: 2022-04-12 15:50:45 --- ### INN URL: https://www.taboola.com/resources/case-studies/inn-3 Last Modified: 2022-07-26 18:57:51 --- ### INN URL: https://www.taboola.com/resources/case-studies/inn Last Modified: 2022-08-09 17:51:03 --- ### The Post Internazionale URL: https://www.taboola.com/resources/case-studies/the-post-internazionale-it Last Modified: 2021-12-15 19:09:42 --- ### Samsung Life URL: https://www.taboola.com/resources/case-studies/samsung-life-ko Last Modified: 2021-12-15 19:11:24 --- ### Universal Music Group & Metallica URL: https://www.taboola.com/resources/case-studies/universal-music-group-metallica-de Last Modified: 2022-06-17 15:46:31 --- ### Universal Music Group & Metallica URL: https://www.taboola.com/resources/case-studies/universal-music-group-metallica Last Modified: 2022-06-17 15:46:53 --- ### Mercedes-Benz Mexique URL: https://www.taboola.com/resources/case-studies/mercedes-benz-mexique-fr Last Modified: 2021-12-15 19:37:48 --- ### Mercedes-Benz Mexico URL: https://www.taboola.com/resources/case-studies/mercedes-case-study Last Modified: 2022-05-17 14:16:06 --- ### HuffPost Korea URL: https://www.taboola.com/resources/case-studies/huffpost-korea-ko Last Modified: 2021-10-20 11:46:28 --- ### Renault URL: https://www.taboola.com/resources/case-studies/renault-2 Last Modified: 2021-10-20 18:22:37 --- ### Sony Pictures URL: https://www.taboola.com/resources/case-studies/sony-pictures-ja Last Modified: 2021-10-20 18:44:18 --- ### HuffPost Korea URL: https://www.taboola.com/resources/case-studies/huffpost-korea Last Modified: 2021-10-20 11:46:53 --- ### Food Network URL: https://www.taboola.com/resources/case-studies/food-network Last Modified: 2022-08-09 17:15:35 --- ### The Post Internazionale URL: https://www.taboola.com/resources/case-studies/post-internazionale Last Modified: 2021-10-20 11:48:08 --- ### FOX SPORTS URL: https://www.taboola.com/resources/case-studies/fox-sports Last Modified: 2021-10-20 11:57:47 --- ### CBS Interactive URL: https://www.taboola.com/resources/case-studies/cbs-interactive Last Modified: 2021-10-20 17:43:49 --- ### Samsung Life URL: https://www.taboola.com/resources/case-studies/samsung-life Last Modified: 2021-11-15 18:39:00 --- ### Renault Spain URL: https://www.taboola.com/resources/case-studies/renault-es Last Modified: 2021-11-15 18:39:27 --- ### Renault URL: https://www.taboola.com/resources/case-studies/renault Last Modified: 2021-10-20 18:22:48 --- ### Renault Spain URL: https://www.taboola.com/resources/case-studies/renault-spain Last Modified: 2021-11-15 18:33:02 --- ### Sony Pictures URL: https://www.taboola.com/resources/case-studies/sony-pictures Last Modified: 2021-10-20 18:44:26 --- ### Adidas URL: https://www.taboola.com/resources/case-studies/adidas Last Modified: 2022-01-12 16:18:22 --- ### CNN Greece URL: https://www.taboola.com/resources/case-studies/cnn-greece Last Modified: 2022-08-09 17:44:24 --- ## Ebooks ### Taboola's Sponsored Content Checklist URL: https://www.taboola.com/ebooks/taboolas-sponsored-content-checklist-2 Last Modified: 2024-08-01 09:45:24 --- ### Moments of Next in Real Life URL: https://www.taboola.com/ebooks/moments-of-next-in-real-life Last Modified: 2024-08-01 10:27:28 --- ### Checklist to Design Your Next Landing Page URL: https://www.taboola.com/ebooks/checklist-to-design-your-next-landing-page Last Modified: 2021-10-22 18:29:26 --- ### How to Start a Content Discovery Campaign from Scratch URL: https://www.taboola.com/ebooks/how-to-start-a-content-discovery-campaign-from-scratch Last Modified: 2021-10-20 19:18:43 --- ### How to Create Genuine Native Content URL: https://www.taboola.com/ebooks/how-to-create-genuine-native-content Last Modified: 2021-10-20 19:08:23 --- ## Trends And Insights ### Optimize Your Campaign with the Most Trending Keywords URL: https://www.taboola.com/trends-and-insights/optimize-your-campaign-with-the-most-trending-keywords Last Modified: 2021-10-20 19:28:18 --- ## Best Practices ### Customer Learning Series: How to Run Successfully on Taboola URL: https://www.taboola.com/best-practices/customer-learning-series-how-to-run-successfully-on-taboola Last Modified: 2024-08-01 12:45:36 --- ### Strategies for Driving Post-Click Actions URL: https://www.taboola.com/best-practices/strategies-for-driving-post-click-actions Last Modified: 2021-10-20 18:57:46 --- ### Creating Content Pages That Convert URL: https://www.taboola.com/best-practices/creating-content-pages-that-convert Last Modified: 2021-10-20 18:57:58 --- ## Engineering ### Inside RecSys 2025: How LLMs Are Rewriting the Rules of Recommendations URL: https://www.taboola.com/engineering/recsys-2025-ai-recommendation-trends/ Last Modified: 2026-02-09 09:01:10 Explore the top trends from RecSys 2025 in Prague, including LLM-powered recommendations, Relational Foundation Models, data quality advances, and real-world AI challenges. At RecSys 2025 in Prague, one trend was impossible to miss: Large language models (LLMs) and recommender systems are converging, signaling a new era for personalization. Over several days of keynotes, poster sessions, and insightful panels, researchers and practitioners explored how AI is shaping the very core of personalization, as well as what it will take to turn these breakthroughs into real-world systems. Here are some of my top takeaways. ## Relational Foundation Models: The Next Leap for Structured Data Imagine being able to train one model that could power predictions across dozens of use cases, from click-through rates to content engagement. That’s the promise behind Relational Foundation Models (RFMs), the focus of the standout keynote by Jure Leskovec, professor of computer science at Stanford University. Foundation models have transformed our understanding of unstructured data like text, images, and code. But as Leskovec put it, structured data such as transaction logs and customer journeys are still locked behind brittle pipelines and handcrafted machine learning models. RFMs aim to change that. As Leskovec’s presentation pointed out, RFMs provide a single, general-purpose model pretrained on relational structures that can perform in-context learning across a variety of downstream tasks. Leskovec compares their behavior to how LLMs handle text. For companies working with large-scale structured data, this technology promises: - Faster iteration without repetitive model design or feature engineering. - Improved predictive accuracy across diverse business use cases. - Simplified infrastructure for teams managing multiple models. For those building and maintaining predictive systems at scale, the potential is impressive. An RFM backbone could eventually support multiple use cases, from CTR prediction to user engagement modeling, all from one unified architecture. At Taboola, it’s exciting to consider how RFMs could streamline the predictive layers behind ad CTR models, personalization systems, and more. ## Tackling the Cold Start Problem in Sequential Recommendations Another standout presentation was Let It Go? Not Quite: Addressing Item Cold Start in Sequential Recommendations With Content-Based Initialization. This session highlighted one of the industry’s biggest challenges: handling new items with little or no interaction data. By limiting embedding, and retraining and seeding models with content-aware signals, the method outlined by the authors can improve sequential recommendation accuracy. This is particularly relevant for dynamic creative optimization workflows that frequently introduce new creatives or formats. It’s worth deeper exploration for anyone using GRU4Rec or other session-based models in high-velocity environments. ## Data Quality and Shapley-Based Filtering When it comes to data, the old adage, “quality in, quality out” applies. For that reason, several researchers are looking at data from new angles. A particularly interesting study applied Monte Carlo-based approximations of Data Shapley values to identify harmful data points in recommender training sets. These included issues like bot traffic and poor metadata, which can affect model performance. By filtering these outliers, researchers achieved measurable gains in performance on KNN-style recommenders, serving as a reminder that model improvements often start with better data hygiene. ## Negative Feedback and Bias Correction in Recommendations When building recommendation models, positive feedback like clicks, views, and purchases get most of the attention. But at RecSys, the keynote by Xavier Amatriain, VP of AI Products at Google, titled Recommending in the Age of AI: How We Got Here and What Comes Next highlighted the value of negative feedback. In his presentation, Amatriain highlighted the paper Balancing Fine-tuning and RAG: A Hybrid Strategy for Dynamic LLM Recommendation Updates. Its authors showed that combining periodic fine-tuning with retrieval-augmented generation helps models adapt to changing user interests. This approach avoids driving up compute costs, while keeping recommendations fresh. In large-scale YouTube tests, this hybrid method delivered measurable performance gains. If you’d like to read more on some interesting topics including, but not limited to negative feedback and bias correction, I find these papers especially helpful: - Benefiting From Negative Yet Informative Feedback by Contrasting Opposing Sequential Patterns: The authors trained two transformer models: one for positive actions and one for negative, and compared the patterns between them, slightly outperforming the state-of-the-art deep model, SASRec. - Unobserved Negative Items in Recommender Systems: Challenges and Solutions for Evaluation and Learning: Models often assume that items a user has never seen are “negative,” which can distort training results. The paper proposed a statistical correctional method called inverse probability weighing as a way to make these evaluations more reliable. - Addressing Multiple Hypothesis Bias in CTR Prediction for Ad Selection: This is one of the most relevant papers in our domain, proposing a post-processing calibration method that corrects bias in click-through rate (CTR) predictions. This approach can be paired with any model and has already improved cost per acquisition and CTR at LinkedIn. - RecViz: Intuitive Graph-based Visual Analytics for Dataset Exploration and Recommender System Evaluation: Proposes using a graphics processing unit-accelerated visualization tool that helps teams quickly explore datasets and evaluate recommender models through interactive graph views. - Revisiting the Performance of Graph Neural Networks for Session-based Recommendation: Shows that with proper tuning, older models like GRU4Rec can outperform newer graph neural networks. The results show that optimization and evaluation practices matter as much as investing in new architecture. ## The Consensus The dominant theme at RecSys 2025 was the growing integration of LLMs into recommendation systems. While these models show great promise, the field is still waiting for a true breakthrough. Key challenges include the dynamic nature of user interest and the trade-off between computational complexity and real-time performance. Those constraints continue to limit widespread adoption in production environments. A few studies, such as the one presented by Google, demonstrated ways to combine advanced LLM techniques to better meet real-world applications. That signals that the industry is making meaningful progress. Notably, it was striking how few sessions at RecSys 2025 focused on advertising or CTR prediction: a sign that some of the field’s most competitive advances remain behind closed doors, underscoring how much innovation is happening privately even as the broader research community moves forward. --- ### How We Developed Our AI-Based SQL Rewrite System URL: https://www.taboola.com/engineering/rapido-sql-rewrite-system/ Last Modified: 2025-10-31 08:59:44 Meet Rapido, Taboola’s AI-powered SQL rewrite system. Learn how LLM recipes, schema & plan context, and production benchmarking automate query tuning and speed up runtimes. For many database professionals I know, having an automatic SQL rewriting system has been something they’ve been aspiring to for many years. Building such a system has been a challenge, but with the rise of LLMs it became much more feasible. So we went on a journey and did it.. We also gave it a name that will fit what it does - Rapido.  ## The Human Approach To Query Rewriting When humans look at a query for potential rewrites, they: - Search for anti-patterns in the query based on their experience, and possibly searching for such anti-patterns in the web - Try different ways to write parts of the query (join to exists, row_number() to cross apply, etc) - Go over the query plan information trying to understand where the bottleneck is - Go over the table schemas and index/projections - Apply hints In some cases, when they have knowledge about the data in the tables and relations between the tables, they also: - Check data distribution and adjust the query where possible - Change the query based on the knowledge they have on the data of their company’s specific data model They then try a few rewrites for the query until they find a fast enough one, in a process looking about like the below: But there’s a limit to how much one person/team can do. Can we use the Generative AI revolution for doing it at scale? Let’s do it! So what would such a system at Taboola need? - Knowledge about general SQL query tuning techniques - Taboola specific optimizations, relevant to Taboola data models - Schema information (data types and indexes/projections) - Data distribution information - Query plan information - Automated testing for finding the best rewrite ## Connecting The Pieces As an AI summarized it, “at its core, Rapido is a pipeline that mimics an expert’s workflow. It starts by gathering context, then generates multiple rewrite ideas in parallel, benchmarks each one, validates the results, and finally takes steps for implementing the rewrite”. Here’s the flow: Let’s explain each step. ### Get query Rapido gets queries from a few interfaces: - UI, when a developer/analyst wants Rapido advice on a query - Automations pulling queries from production - Soon Git, for verifying queries get to production with good performance As an example, here’s Rapido UI: Developers and analysts can submit their query for checking through Rapido UI. They select their database type, whether it’s a backend or frontend query, the timeout for the query, and number of executions, for making sure we get consistent results. The user can also choose whether to use the predefined recipes or supply a custom prompt. ### Get relevant recipes A recipe is something we teach the LLM to look for and guide it which rewrite to perform. We supply an explanation for what to do, alongside an example query in which this recipe helped improve performance. Prompt engineering was key here. We send the recipe as part of the prompt with very clear instructions as to what to look for and how to apply a fix. ### General recipe examples Union to Union All “Optimize the Union operations in this query by: - Converting UNION to UNION ALL where duplicates are impossible - Optimizing individual queries within the UNION” Data Type Mismatches (Implicit Conversion) “Fix data type mismatches: Quote numeric values when filtering VARCHAR columns (e.g., WHERE varchar_col IN ('123', '456') not IN (123, 456))” Preaggregate With CTE “Rewrite the SQL query to pre-aggregate data with a Common Table Expression (CTE) to improve performance”. Merge CTEs “Sometimes people use too many CTEs. Change the query logic to use less CTEs by merging CTEs that can be merged without changing the query logic. An example for CTEs that can be merged are CTEs that query the same tables. Make sure the result is a valid Vertica SQL query and that the query logic stays the same” Not Exists To Left Join “Optimize this query by: - Converting NOT EXISTS to LEFT JOIN with IS NULL check - Maintaining correct NULL handling - Preserving exact semantic equivalence” ## Taboola Specific Recipe Examples - We guide the LLM to treat tables from certain schemas as fact tables, filter and preaggregate them using a CTE - We identify a common unneeded historic join and rewrite the query to grab the relevant column from one of the fact tables - We add a common missed column to a join which we know improves performance In addition, we have a general prompt called “magic” for general performance guidance, in which the LLM can be more flexible: “You are a Senior Database Engineer. Rewrite the Vertica query to optimize performance by: - Converting complex subqueries to efficient JOIN operations - Optimizing JOIN sequences for better performance - Simplifying complex WHERE conditions - Minimizing data movement between nodes where possible - Improving GROUP BY and aggregation patterns - Avoiding SELECT * and listing only needed columns Maintain exact semantic equivalence while focusing only on SQL structure” ### What else do we pass the LLM for more context? - Database type and version - Schema information - Data distribution information - Query plan (explain) ### Do we pass all recipes for each query? No we don’t. We parse the queries, and if, for example, the query doesn’t have “union”, we won’t pass the “union to union” all recipe. ### Why do we have a few recipes? Why don’t we use a single prompt and get a single query back? Because as you might know from your query tuning work, each query is different, and something which works for one query doesn’t necessarily work for the other. So we get a query for each recipe and let the best recipe win. ## Test Each Recipe In Production We get the rewritten query for each recipe, run it and measure performance. We run it a few times in order to verify we get a consistent runtime. We run the rewritten queries In production, in order to make sure we get production runtimes. ### How do we make sure we don’t burn the kitchen? We run the queries one at a time, in isolated servers/clusters in order not to hurt any production workload. We also limit the runtime and amount of resources each query can use. ### Logic validation: how do we verify the queries are logically equivalent? The easy part is when it’s a session option we can control on our side. In this case, we only change the session option in a configuration table. Our automation is then verifying the performance is indeed better and rolls back the change if it isn’t. If it’s not only a session option but an actual rewrite, we have a few options for verifications. In most cases, we perform a count and an unordered hash calculation over the result set. If they are the same between the original query and the rewritten one, and the performance boost is significant, we pass it to the query owner for human verification and implementing the rewrite in code. ## Results So how does it look? Let’s look at runtime results for a few queries. In this case, the original query ran for almost 60 seconds, while the “magic” recipe resulted in a 2 seconds runtime, and the “preaggregate_with_cte” recipe resulted in a 1.2 second runtime. Here, the “magic” recipe reduced the runtime from  110 seconds to 5 seconds. The other recipes did not improve the query (the “magic” was to turn an IF, which is an external function in Vertica, into a CASE). In this case, “merge_ctes” is the winner, with a runtime almost 15 times faster than the original query. Here, the winner is “not_exists_to_left_join”, with a 15 seconds performance improvement. And in this case, the “preaggregate_with_cte” recipe is the winner with runtime 56% faster than the original query, while the other recipes were actually slower than the original one. As you can see, each query is different, and having a wide range of recipes increases the possibility for a good rewrite. The UI also offers a comparison screen between the original query and the best rewrite, where the user can also see an explanation about the nature of the rewrite ## Challenges We initially worked with the OpenAI GPT-4o model and got back many queries that were “too creative”. Lowering the temperature to 0 helped but not fully. We also got queries that were cut off. Both of the issues were resolved when we moved to the o3-mini model which allows more max_tokens and follows the recipes better without the need for temperature definition Other things that we did for improving coverage: - In cases where the LLM returns a query that has a syntax error, we provide the LLM with the error message and let it try to fix it. This improves the number of compiling queries by 30% - Even though our recipes are descriptive and specific, we do allow some flexibility inside them, and have the general “magic” recipe which allows the LLM more creativity - We make sure the LLM response doesn’t end prematurely Another challenge was how to verify we supply a rewrite that is logically equivalent to the original query. As stated, we perform a count and an unordered hash calculation over the result set. However, what should we do in cases where data in the table changes frequently, or in cases where the rewrite generates a result set slightly different while the performance benefit is very big? We’re currently considering exposing such cases while mentioning the difference in percent between the data sets. ## Side Benefit While developing Rapido, we needed a way to benchmark queries for knowing if a rewrite is better than its original query. We developed such a tool which is now being used by us humans when trying different ways for writing queries and when comparing performance between different database engines. ## Impact Rapido is relatively new, and one of our current challenges is measuring the impact and understanding whether an optimization suggestion was implemented in production. What we know so far: - Up to 40% potential performance boost for specific workloads that were tested - 15% performance boost by automatic fixes where they were possible - 3200 human optimization hours were saved ## What’s Next For Rapido? Rapido keeps being developed. Here are a few things we plan for it: - More input interfaces: CI/CD pipeline, for making sure queries are deployed to production with proper performance - MCP, for allowing other Taboola application utilize Rapido - Optimizing queries over more database engines - Automated comparison and benchmarking between database engines - An optimization loop - further optimizing an already Rapido-optimized query - Optimization targets beyond runtime (e.g. memory usage, CPU time) - More recipes and enhancing the existing ones for better coverage - More automations where possible We believe the approach that led us with Rapido can be beneficial for many organizations. ## The Amazing People Behind Rapido Rapido was developed by Yakir Gibraltar and Illés Solt who are in charge of the Rapido engine, Nati Poliszuk who’s in charge of the UI, and yours truly. It’s been an amazing journey developing it together, and we’ll keep on making it better. --- ### Enhancing Developer Flow: Smart Optimizations in a Monorepo World URL: https://www.taboola.com/engineering/enhancing-developer-flow/ Last Modified: 2025-06-26 12:21:39 Discover how Taboola optimized developer workflows using remote dev environments and production-like testing to boost productivity. Every minute a developer spends waiting is a minute of lost innovation. Recently at Taboola, a critical question has emerged: how can we tackle the productivity challenges and inefficiencies that weigh down our developers? The first step was to identify the key areas needing improvement. In a monorepo setup such as the one we use at Taboola – with over 8M lines of codes and 160K unit tests – these challenges often center around long feedback loops and delayed test cycles. While monorepos provide consistency and foster cross-team collaboration, they also introduce latencies and lengthy pipelines that can slow down development. In this article, we address some of these bottlenecks and the solutions we've implemented to reduce delays and enhance developer productivity. ## Current Development and Deployment Workflow At Taboola, when developers push their code to a Git branch, an automated pipeline kicks off a build process that takes an average of 15 minutes. This includes compiling source code, generating JAR files, building Docker images, and packaging artifacts, as well as running unit tests to catch issues early. After the build completes, the newly created Docker images can be deployed to testing environments, where developers can execute integration tests to validate functionality and ensure stability before progressing to further stages. This process repeats iteratively until the changes meet functional and code quality standards. If tests fail, developers refine their code, push updates, and trigger another build. These cycles of testing, reviews, and fixes can be time-consuming, highlighting the need for optimization before the code is verified and ready for production. ## Remote Webpack: Reducing Feedback Loops for Frontend Development ### The Problem For a frontend developer, the feature development process follows a similar pattern: opening a branch, adding the code, and triggering a Jenkins build. Once artifacts are ready, the next step is the creation of an on-demand testing environment in Kubernetes with the newly built version, where Selenium tests are executed to simulate user interactions with the application, ensuring that the frontend behaves as expected across different browsers and environments. This feedback cycle introduces significant latency, as the considerable wait time between code changes and testing can disrupt the development flow. ### The Solution When faced with these productivity challenges, it’s tempting to seek new tools or technologies. However, some of the most impactful solutions can come from optimizing existing tools. In our case, IntelliJ Gateway was the starting point. This tool allows developers to access fully configured IDEs in remote environments, reducing startup and indexing time while preserving IntelliJ's powerful features. We leveraged IntelliJ Gateway’s remote development feature to connect directly to a development pod running in Kubernetes. This pod is provisioned with a full-fledged development environment and the connection is established over SSH, with IntelliJ Gateway handling the remote execution of the IDE backend, while rendering the UI locally. We extended this approach with rsync, synchronizing changes between a developer’s local machine and a remote pod in the same namespace as the Webpack service. This remote pod processes changes immediately after each recompile as it efficiently transfers only the modified files to the remote pod, creating an experience that feels just like working locally. To streamline access, the remote pod is externally monitored, ensuring that one is always available when it's needed. Developers can instantly resume their sessions without needing to build again in Jenkins or create a new pod. This setup provides a near-local development experience while benefiting from the scalability and consistency of a Kubernetes-hosted environment. ### Key Benefits - Immediate Feedback: Developers get near-instant feedback after each code change, eliminating the overhead of full rebuilds. - Improved Resource Utilization: Resolves network issues while reducing CPU and memory usage on local machines. - Streamlined Setup: While the initial setup takes 8–10 minutes, each subsequent change provides immediate results, keeping the developer workflow smooth and uninterrupted. Step Traditional Workflow  Optimized Workflow  Code Change Edit file locally Edit file locally Build Time 15 mins in Jenkins A few seconds Deploy 7 mins on average No additional deployment File Sync Full artifact upload Only changed files Test Feedback Delayed (after deployment) Immediate This feature transformed the feedback loop into a seamless process, allowing developers to focus on what truly matters: writing great code. ## Gonzales: Testing in a Production-Like Environment ### The Problem In traditional workflows, developers often rely on staging or test environments that may not fully replicate the complexity or scale of the production system. As a result, features might behave differently once deployed to production, leading to bugs or unexpected behaviors. Additionally, testing in these environments can be slow, requiring significant setup time and resources. ### The Solution Gonzales allows developers to test feature branch changes in a production-like environment within just a few minutes. The way it works is simple yet powerful: Gonzales copies the local changes from a developer's feature branch into a production machine, where the application is then run in read-only mode. This mode, a feature of the underlying system, ensures no side-effects like traffic logs or data manipulation occur while still running the application in a production-like environment. This approach allows developers to manually test their changes without affecting live user data or generating pageviews. The code under test is packaged in a JAR used exclusively when running the app with Gonzales, ensuring that it doesn't leak into the production system, even though it operates on the same machines. ### Key Benefits - Mirrored Traffic: By exposing the test machine to the same incoming traffic as the production environment, Gonzales provides an accurate reflection of real-world behavior. - Safety and Isolation: Since the environment operates in read-only mode, no side effects (such as data manipulation) can occur, ensuring safe testing. - Efficiency: This approach dramatically reduces setup time, while offering a reliable testing environment, bridging the gap between staging and production. By simulating production conditions with real traffic, Gonzales ensures that features are thoroughly vetted before deployment, reducing bugs and enhancing confidence in production releases. By optimizing our development and testing workflows, we’ve removed key bottlenecks that were slowing down feature delivery. Remote Webpack has accelerated frontend development by providing immediate feedback, while Gonzales ensures reliable testing in a production-like environment without the usual setup overhead. As organizations strive to enhance developer productivity, evaluating and refining existing workflows can be a highly effective approach. We encourage teams to audit their development processes, identify bottlenecks, and explore optimizations that maximize efficiency without adding unnecessary complexity. Credits: Alon Pilberg and Shirel Hadad --- ### How Changing a Few Lines of Query Reduced the Execution Time by 80% URL: https://www.taboola.com/engineering/vertica-query-optimization/ Last Modified: 2025-05-14 16:12:20 How Taboola cut query time by 90% on a 45B-row Vertica job using smart pre-aggregation, statistics, and projection tricks. Here’s how they did it. ## The Challenge: A 7-Hour Job on 45 Billion Rows At Taboola, we run a massive Vertica job that calculates analytics and KPIs through multiple processing stages. The core query scans 45 billion rows spanning 90 days, performing complex calculations to generate dynamic KPIs. With a runtime of 7 hours per day, this job wasn’t just consuming resources - it was delaying analysis downstream which impacts dev velocity. Optimizing such a workload required a fundamental shift in our approach, starting with a deep dive into how we process and structure the data. ## Background: Meet Genie Genie’s our A\B testing analysis framework, which allows developers to choose a list of dimensions and metrics and get daily aggregated report tables including complex statistical calculations. The requirement for Genie was to get for each dimension the aggregated value over the entire time of the A\B test, so for example if I asked How many clicks I had per country & platform per day, Genie will create a row per day, country & platform with the total Clicks over the entire period of the test (see picture for example) Sounds simple?  We get around 600m new rows per day, but if a test runs for a few months and more dimensions, this could easily force our job to scan 45B (B as 45 billion rows :)  rows and aggregate them every day from scratch. ## ## Step 1: Pre-Aggregation—The First Breakthrough Our first idea was to adopt a cumulative pre-aggregation approach Instead of every day recalculating ALL the data of the test (e.g. 90 days of data), just aggregate the last day, and accumulate it with previous “total-so-far” results (sum of last 89 days + today). This method eliminates the need to recalculate all the data every run, rather simply retrieve a single, lightweight entry. We create a cumulative job that aggregates all data per day, each day adding only the new day into cumulative table: Now, our job should scan fewer rows than before: 3B vs 45B  (>90% reduction in rows) But wait wait, you said ~600m rows, why 3B and not 600m?? Good question! Initially, we thought each cumulative day would have the same number of rows as a single day from the raw data: ~600M rows. However, because the key granularity varies between days (e.g. the same combination of country/platform isn't present every day), we have more rows per day in the cumulative data. For example if our dimensions are platform & country,  in a specific day we might see only: - Israel & Chrome, Israel & Safari, UK & Chrome And in next day contains: - Israel & Chrome, Israel & Safari, UK & Edge (yes there are still Edge users out there ) Aggregate of those 2 days = 4 rows overall (not only 3). Ok so it sounded great on paper, and initial tests showed promise. But… Reality had other plans. ## Step 2: The Mystery of Faster Queries ### Failures Before Success When we deployed the cumulative solution to production, we noticed an odd pattern: the first execution would often fail after a long period of time, only to succeed in a retry just a few minutes later (!) — exactly as we expected it to work. If the original failure was due to the complexity of the query, wouldn’t we expect the retries to also be slow? So why were the successful retries so fast?! To make sure there’s no dark magic going on, we tried running the same production query on the same environment under a test job we saw the query is still super fast, the ONLY differences we could think was: The production query always runs on a newly created database daily partition. In contrast, the retries occur hours later and after at least a few queries were already run on this partition. Same query, same resources, same environment, same load - the only factor that changed? Timing. 🤔 This raised our suspicion that additional factors were influencing the query’s performance. ### Database statistics The Vertica cost-based query optimizer relies on data statistics to produce query plans. If statistics are incomplete or out-of-date, the optimizer is liable to use a sub-optimal plan to execute a query. (Vertica doc) When we added an initial step before actually running the query in production to report the explain plan of our query, we noticed Vertica reporting missing_stats on the new day partition which means statistics were still not established on the new partition (not table! The statistics are apparently managed per partition and not per table). And this also explains why subsequent runs (the retries and our manual tests) were much faster - because the original query ran, Vertica had enough time to gather statistics on the new partition making all subsequent queries run MUCH faster. ### The Solution By running ANALYZE_STATISTICS on 1% of the data for the date partition column before executing the query, we provided Vertica with enough information to optimize its plan. The result? The query runtime dropped by a staggering 90%! ## Step 3: Rethinking Temporary Tables With the first step in our flow optimized, we turned our attention to the next step in the DAG. This part of the process relied on temporary tables and involved operations like COUNT(DISTINCT) and GROUP BY. Initially, we assumed the COUNT(DISTINCT) operation was the bottleneck. However, even after removing it, the query runtime remained largely unchanged. The realization that this query primarily relied on GROUP BY operations across multiple fields prompted us to explore how Vertica segments and sorts data across its nodes. In Vertica’s terminology, this is referred to as a "Projection." This directly influences query performance by avoiding the data shuffle between nodes. ### The Revelation: Projections and Temp Tables Digging into Vertica’s documentation, we found that temporary tables inherit their projections from the SELECT statement used to populate them. Without explicitly defining a projection, the column order in the SELECT statement could directly impact subsequent query performance. Confusing?  let’s take an example: - The original query returned a temp table with the columns: Country, Browser, Variant - The subsequent query running on the temp table grouped by: Variant, Country, Browser - Because the temp table inherited the projections from its SELECT, even though the subsequent query grouped by the same list of columns, because the order is different it had to shuffle the data across all nodes only to do the same exact group by! ### The Solution We aligned the column order in the SELECT statement with the GROUP BY operation in the subsequent query. This adjustment reduced the runtime of this step by 80%—from 134 minutes to 23 minutes! the reason is that it avoids shuffle between vertica nodes: ## The Final Result: A Transformed Process After some iteration, testing, and learning, the transformation was complete. - Trying a pre-aggregation approach to reduce the overall data scanned. - Running ANALYZE_STATISTICS eliminated inconsistencies and ensured optimal query plans. - Optimizing the projection of temporary tables further slashed runtimes. The result? A job that once took 7 hours now completes its key steps in under an hour, freeing up valuable resources and ensuring timely delivery of results. ## Lessons from the Journey Looking back, this journey offered valuable lessons for anyone optimizing large-scale Vertica queries: - Pre-Aggregate Data for Efficiency Pre-aggregated data can make even the most complex queries faster - Analyze Statistics  Ensuring Vertica starts with the right query plan. Running ANALYZE_STATISTICS at the right moment guided Vertica to generate better plans from the start, improving both reliability and performance. Especially when you query a new partition - Running ANALYZE_STATISTICS on the partition column even on 1% of the data before query execution can be a game-changer - Pay Attention to Projections and shuffled data Temporary tables inherit their projections from the SELECT statements used to populate them. Aligning the column order with downstream operations like GROUP BY can lead to substantial performance improvements. - Embrace Iteration and Discovery Optimization is a journey of trial and error. Every failure offers clues that bring you closer to the solution. --- ### A Key Component in Taboola CI/CD Pipelines for Achieving High Production Stability URL: https://www.taboola.com/engineering/real-traffic-ci/ Last Modified: 2025-03-03 17:36:42 See how Taboola integrates real traffic into CI pipelines using GoReplay to enhance testing, improve stability, and ensure high production reliability. ## Automated Verification CI pipelines In theory, testing in a controlled environment should cover all bases. In practice, nothing beats the accuracy of replaying real-world traffic in your CI/CD pipeline. At Taboola, in our continuous integration (CI) pipelines, we employ a suite of automated tests utilizing real traffic. These tests are executed on a feature branch against our production branch to measure key performance metrics and ensure correctness. This service in particular is responsible for serving Taboola recommendations at the rate of 1 million requests per second from thousands of machines across 7 data centers. This automated CI pipeline not only helps to maintain its correctness but also allows all engineers to iterate fast on this recommendation flow. In this blog post, we delve into how we incorporate real traffic within our CI pipeline and maintain high standards without traditional QA. ## Testing with Real Traffic? Testing against real traffic significantly boosts the confidence of Taboola engineers, embedding accountability into their work. To successfully implement this testing methodology, several key considerations must be addressed: - Impact on Production: Though there are great advantages of having the ability to test code against production traffic, we must also minimize the possible impact to services Taboola provides. - Traffic Quality: To test the modified code effectively, we must be comprehensive about the traffic quality it receives. It should provide good coverage of our traffic, and address most features of Taboola service. - Request Rate Stability: In order to measure the performance between the production branch and a feature branch, we must have the ability to send steady and consistent high quality traffic to both servers. Our exploration led us to evaluate tools like the NGINX mirroring module and the GoReplay framework. Both tools can replicate traffic to different destinations, but we found that the effectiveness of these tools heavily depends on the stability and volume of the incoming traffic source. Notably, GoReplay provided traffic recording capabilities that became invaluable for our process. ## The Issue with Testing against Real Traffic Ideally, we would like to have the incoming traffic at a consistent and decent rate that the system load is enough when we measure the performance metrics, and not too much that it overloads the system. Before this tool was developed, we had issues with inconsistent incoming traffic rate from time to time, demonstrated by the graph below. The incoming traffic rate was not consistently steady enough. On top of that, in this case, it created too much load on a server (while at times it created too little load on a server), and when the servers were overloaded, we were not able to effectively measure and compare the performance differences between the production and feature branches. ## Recording Live Traffic First, we regroup and define our requirements, understand the scope and what we set out to achieve. Here are our target goals: - Specific Duration Recording: It’s important we have the ability to capture traffic over a specified time frame to mirror production scenarios because the traffic patterns vary at different times in different regions of the world. - Guaranteed QPS: To accurately compare performance of two different JVM, we must ensure a consistent queries per second rate to two different machines. - Customizable Headers: Allowing the insertion of HTTP headers to guide traffic flows because the traffic requests are wired differently inside of Taboola service, in which case we can effectively control the features being tested. - Traffic Isolation: Focus on specific types of traffic as needed. - Traffic Distribution: Disperse identical traffic streams across multiple targets for parallel testing. By doing so, our comparison metrics between two different JVM instances can be meaningful. ## GoReplay: The Backbone of Our Traffic Testing Tool At Taboola, we have used both the Nginx rating limiting module and GoReplay for some time now. While both tools are quite amazing at what they do, there are still some parts of the features we wish them to have naturally. Based on our experience with both and the requirements defined early. GoReplay stands out as an essential tool for capturing and replaying traffic, offering features such as: - Rate-Controlled Recording: Capture traffic without overwhelming the systems. However, GoReplay’s recording capability is limited to the volume of incoming traffic as naturally it can only record what it receives. - Multi-Host Playback: Replaying the same traffic to multiple hosts to facilitate comprehensive testing is an essential feature of GoReplay. - QPS Management: Without consistent source traffic QPS, achieving a steady final output is challenging despite GoReplay’s rate control options. ## Customizing GoReplay for Our Needs To meet our business requirements, we decide to move forward with GoReplay and implemented custom enhancements on top of it: - Timestamp Rewriting: Modify traffic timestamps in GoReplay’s output files to simulate a fixed 100 QPS rate, regardless of actual recorded rates. - Percentage-Based QPS Control: With a stable 100 QPS source, control throughput by adjusting playback percentages with ease (e.g., 50% for 50 QPS). ## The Result After the tool was implemented, we were able to control the right amount of traffic, consistently at 75 QPS, which produced sufficient system load, allowing us to measure the meaningful impact of both branches before releasing the new code to our production environment. In addition, by utilizing the recorded traffic, we were also able to reduce the number of servers used in the CI pipelines from three to two when, in the previous setup with Nginx rate limiting module, it required one additional machine to forward requests to the other two servers. ## Exploring Additional Benefits The application of this traffic replicating approach extends beyond just our benchmarking of new features into the production environment. It can also serve in other areas of our infrastructure: - System Warm-Up: Emulate real-world traffic to effectively prepare our systems, particularly the JVM, for consistent performance. - Load Testing: With this setup, we can easily produce load to our system with real traffic. - Traffic for Local Environments: When Taboola developers write and test code locally, this approach can assist by sending real traffic to their local environments. ## Looking Forward: Future Enhancements As we refine our methodologies, future enhancements can include: - Enhanced Traffic Filtering: Fine-tune data captured to focus on specific publisher traffic. - Time-Based Variation: Use traffic samples from various times of the day in different time zones to simulate different user behaviors and loads. By adopting these advanced traffic recording and replay techniques, Taboola ensures robust testing environments that closely replicate real-world conditions, fostering more reliable CI pipelines, which contribute to our more stable production environments. --- ### From Zero Labels to 85% Accuracy: Our Journey in Conversion Intent Prediction URL: https://www.taboola.com/engineering/predicting-ad-intent/ Last Modified: 2025-02-25 14:55:13 Taboola built an AI model to predict Conversion Intent (CI) without human labels, improving ad targeting, boosting conversions, and optimizing placements. ## Introduction Understanding the commercial intent behind online content is essential for optimizing ad targeting and maximizing revenue. At Taboola, we strive to deliver more relevant advertisements by aligning ads with the intent of the content. This strategic alignment can help improve user engagement, may increase conversion rates, and aims to satisfy both publishers and advertisers. By identifying whether a web page encourages visitors to take action—a concept we refer to as Conversion Intent (CI)—we can tailor our ad placements more effectively. For example, an article titled "The Top 10 Running Shoes for 2024" signals strong commercial intent, indicating that readers might be in a buying mindset. Placing relevant ads in such content increases the likelihood of a sale, potentially benefiting the reader, advertiser, and publisher. Conversely, a headline like "They Took The Same Picture For 40 Years (You Will Surely Get Emotional)" may not exhibit explicit commercial intent. Recognizing this allows us to adjust our advertising approach to maintain a seamless and engaging user experience without disrupting the reader's journey. Conversion Intent (CI) may be key to ensuring our recommendation systems deliver relevant advertisements to publishers. Additionally, CI can serve as a useful filtering parameter within our Contextual Targeting product. For example, an advertiser promoting sports shoes could target placements where the category is “sports/basketball” and CI is marked as true. This approach ensures their ads are shown to audiences with a higher likelihood of conversion. The goal for this project was to build a model that predicts CI with 80–90% accuracy while minimizing effort and resources. Due to time constraints and the high cost of human annotations, we ruled out manually labeled data as an option. ## Development Process I approached the task during a month without utilizing human labels. The development process comprised three main steps: - Investigate Previous Work - Try Existing Models - Improve Based on Domain Knowledge Method 1 - Previous in-house model Method 2 - GPT-3.5 Method 3 - BERT model based on domain knowledge Ensemble results Proposed BERT model Sample number Publisher accuracy 0.66 0.74 0.75 0.84 0.82 200 Sponsor accuracy 0.77 0.82 0.77 0.87 0.89 199 Average accuracy 0.71 0.78 0.76 0.86 0.85 ### 1. Investigate Previous Work Leveraging existing work is crucial, especially under tight schedules. Unfortunately, there was a scarcity of open-source data or models suitable for our specific context of publisher and advertisement content. However, we had an in-house model designed to predict whether a page sells products. Although its primary goal was slightly different—detecting commercial or product links—it provided a valuable starting point. This in-house model was trained using noisy pseudo-labels obtained through web crawling and rule-based filtering. On our newly collected evaluation set, it achieved approximately 70% accuracy, serving as a baseline for further improvements. ### 2. Try Existing Models Large Language Models (LLMs) have become indispensable tools for data scientists due to their ability to handle various language tasks with minimal effort. While they may not always offer the best performance out of the box, they provide strong baselines for rapid development. At the time of this project, GPT-3.5 was the most advanced model available. After applying prompt tuning, the model achieved 78% accuracy. Although the improvement was modest, it confirmed the potential of LLMs in handling CI prediction without extensive training data. ### 3. Improve Based on Domain Knowledge Understanding the nuances of our domain provides additional avenues for improvement. Taboola operates on both the supply side (publisher pages) and the demand side (advertisement landing pages). While it's intuitive to assume that advertisement pages have higher CI than publisher pages, empirical verification was necessary. Data analysis revealed that certain types of advertisements—such as Direct Ads, Calls to Action, and Branded Content—had a higher likelihood of exhibiting CI. In contrast, most publisher items did not encourage direct actions. Leveraging this insight, we trained another model with BERT-base as the backbone and achieved 76% accuracy. ## Ensembling Models for Enhanced Performance With three baseline models—the in-house product detection model, the GPT-3.5-based model, and the BERT-based model informed by domain knowledge—we explored combining their strengths. By ensembling their binary predictions through a simple voting mechanism, we achieved a significant performance boost. The ensemble prediction reached an impressive 86% accuracy. This substantial improvement is likely due to the heterogeneity of the models, as they were trained using diverse data sources and methodologies. To ensure scalability and handle high traffic, we retrained a BERT-base model using the ensemble's pseudo-labels. This final model maintained high performance with 85% accuracy, making it suitable for deployment in a production environment. ## Conclusion The development of a Conversion Intent (CI) prediction model without human annotations presented both challenges and opportunities. Our endeavor to understand and predict commercial intent behind online content was important for improving ad targeting and potentially increasing revenue. By integrating insights from an in-house model, an LLM, and a domain-informed BERT model, we successfully leveraged heterogeneous models to collectively enhance performance, surpassing the capabilities of individual models. This process underscores that with the right methodologies and resources, it's possible to develop high-performing models on tight schedules and limited budgets. --- ### Driving Success and Innovation Across Cross-Functional Teams - Part 3: The Role of the Track Leader URL: https://www.taboola.com/engineering/driving-innovation/ Last Modified: 2025-02-03 18:16:12 Learn about the pivotal role of Track Leaders in driving innovation, managing cross-functional teams, and ensuring strategic success. ## Introduction In the previous parts of this series, Driving Success Through Cross-Functional Management in Engineering at Taboola, we explored how the Tracks methodology fosters collaboration, delivers business impact, and empowers both managers and developers. Now, let’s focus on one of the most critical roles within the Tracks framework: the Track Leader. Track Leaders are not merely project managers, they are a strategic driver responsible for ensuring that the Track achieves its business goals while fostering innovation and team collaboration. ## 1. Core Responsibilities of a Track Leader The Track Leader is responsible for leading their Track to success, based on a strategic mission statement they receive upon nomination to the role. These responsibilities go beyond project management and include strategic leadership, cross-functional coordination, risk management, and resource optimization. ### Strategic Vision and Roadmapping The Track Leader’s strategic direction starts with the mission statement they receive as part of their nomination to the role. This includes high-level objectives that guide the initiative. It is then the Track Leader's responsibility to translate this into a detailed roadmap with milestones and clear KPIs that align with the company’s broader goals. ### Cross-Functional Coordination Track Leaders must facilitate collaboration across departments and groups, ensuring that all members of the cross-functional team are aligned. This involves synchronizing the efforts of diverse teams like R&D, Product, Marketing, and other stakeholders to ensure that everyone is working toward the same goals and that communication flows smoothly across all levels. ### Risk Management and Adaptability The Track Leader must anticipate and manage risks while maintaining the flexibility to adapt when necessary. This includes making strategic adjustments to the roadmap to account for new developments or obstacles. In cases where progress becomes uncertain, Track Leaders should refer to the principle of early failure - “fail fast” (detailed below) to stop or pivot efforts early and consider alternative strategies to achieve impact. This proactive approach is crucial to minimizing resource loss and ensuring that efforts are focused on achievable goals. ## 2. Early Failure: A Key Responsibility for Maximising Impact One of the most critical responsibilities of a Track Leader is recognizing when a project or effort is no longer on track to deliver its intended impact. Early failure should be viewed as a positive step, allowing teams to course-correct or stop certain efforts to focus on more promising alternatives. - Identifying Early Failures: Track Leaders need to assess when a project has low certainty of achieving its impact. If it becomes clear that resources are being invested into a failing path, the Track Leader should act quickly to stop these efforts, reducing resource waste. - Stopping Efforts Early: By recognizing early signs of failure, Track Leaders can reassign resources to more promising alternatives. Stopping efforts early is a key action in ensuring the overall success of the Track, as it prevents further resource drain and helps refocus on strategies that can better meet the impact goals. - Positive Perception: Early failure should be seen as a strength rather than a weakness, allowing for quicker pivots to the right path, maximising the chances of delivering results within the required timeframe. ## 3. Building a Strong Track Managers Forum One of the first actions a Track Leader should take is to establish a Track internal Managers Forum. This forum is crucial for creating a leadership structure around the Track’s mission and ensuring effective collaboration. - Key Members: The forum usually includes the Product Manager (PM), team leaders, and occasionally some key tech leaders assigned to the Track. - Shared Ownership and Leadership: This forum builds a sense of shared leadership and end-to-end ownership of the Track’s goals. It encourages partnership and collaboration among team leaders and key stakeholders, fostering a collective sense of urgency to meet milestones and achieve impact. - Strong PM and Track Leader Partnership: A crucial part of the forum is the strong partnership between the Product Manager (PM) and the Track Leader. If the Track Leader is a PM rather than an R&D Manager, best practices recommend involving an R&D Manager in the internal leadership forum to closely partner with the Track Leader. ## 4. Maintaining Open Communication with Organic Managers Track Leaders must maintain open communication with the organic managers (typically group managers) who oversee the teams working within the Track. Since a group manager may have teams working in a Track led by another group manager, cooperation is key: - Support for Team Leaders: By maintaining open communication, Track Leaders can ensure that organic managers are well-equipped to support their team leaders in successfully fulfilling their roles within the Track. - Collaborative Success: Open communication creates a bridge of support between the Track and the organic structure, aligning efforts to maximise team productivity and ensure that challenges are addressed early. ## 5. Ensuring Track Visibility and Strategic Decision-Making The Track Leader must ensure ongoing visibility of the Track’s progress. This is achieved through regular updates and strategic decision-making: - Weekly Updates: Track Leaders should provide weekly updates to maintain visibility of the Track’s progress against the roadmap. This helps teams stay focused and ensures that any roadblocks are quickly identified. - Key Decisions and Senior Management Review: Track Leaders are responsible for identifying key strategic decisions that require input from the Track Managers Forum. When building or revising a roadmap, it should be reviewed by senior management to ensure alignment with the company’s strategic direction. ## 6. Driving Commitment and Urgency Track Leaders are responsible for building and maintaining a strong commitment to the Track’s mission, goals, and deadlines: - Sense of Urgency: The Track Leader must foster a sense of urgency within the Track to ensure that milestones are met and KPIs are achieved on time. This involves continuously motivating the team, aligning efforts, and addressing challenges as they arise. - Tracking Progress Against KPIs: It’s the Track Leader’s role to constantly track progress against the defined KPIs, ensuring the team is on track and adjusting as needed to hit key business objectives. ## 7. Encouraging Parallel Workstreams and Alternatives In order to optimize outcomes, Track Leaders should consider working on multiple alternative solutions simultaneously. This method can shorten the time needed to achieve impact: - Working on Alternatives: Track Leaders can invest in several alternatives at once, assessing their potential to meet the required milestones and improve KPIs. This may require allocating more resources initially but can shorten the time to impact. - Early Failure in Parallel Workstreams: As with the overall project, early failure is key here. If an alternative is found to be unpromising, the Track Leader must stop that path quickly and reallocate resources to the most promising option. ## Conclusion The role of the Track Leader is multifaceted and requires a unique blend of strategic vision, adaptability, team leadership, and communication. By managing cross-functional coordination, encouraging early failure, and driving innovation, Track Leaders are instrumental in ensuring that their Tracks not only achieve their goals but also contribute to the company’s broader success. --- ### SDK testing with hot swapper URL: https://www.taboola.com/engineering/sdk-testing-with-hot-swapper/ Last Modified: 2025-01-14 14:37:11 In the following article, I describe how we came up with a way to improve the chances that our SDK library gets smoothly integrated in our customers' Applications and reduce issues when going to production. The main idea is to take a number of significant clients' applications and replace your existing SDK code with a new code, allowing you to see how the apps perform before you release a new SDK version. ### Why releasing a reliable SDK is so important Developing an SDK for mobile apps is very different from developing a standalone app. You can think of an SDK as a guest in someone else's house. You need to behave, you can’t put your legs on the table or wipe your hands on the sofa (well, in most countries you can’t). So what I mean is that you can’t interfere with the app’s normal behavior, break some flows or worse than that, cause memory leaks or god forbid crashes. ### The common scenario and why you can do better Now let’s say you developed a new feature or fixed a bug in your SDK. You’ve checked it on your device, passed the QA process (manual, automation, several devices and OS versions) and everything works fine. Great, you can release a new version… can you? Well, No. At least not yet. Building an SDK means there are all kinds of apps that use your product with different implementations and numerous customizations. Of course you can’t send your new release to all your customers and tell them to test the new released version. What you might do is choose several of your biggest customers and tell them to do it, but still the tests and fix cycles will take time, not to mention you don't want them to see the bugs you have. You must be stable from day one, making sure your SDK does not interfere with the app running it. The worst thing that can happen is your customer losing users because your SDK is messing up with the app functionality #### Another Such Scenario It’s getting even more tricky when your customer is updating their own app, regardless of your SDK. For example, a mobile app can change the layout of the app, resulting in a new use case that you didn’t check.  So how can we make sure that releasing an SDK does not result in any regression? How can we keep track on the app changes regardless of the SDK? ### Enter SDK Hot Swapper To solve this issue we had to come up with a new idea - SDK hot swapper ! Imagine you hold in your hands your customer’s application and you can replace the current version of your SDK with a new SDK version you want to test. That’s exactly what our system does. ## So, how does this magic happen? In a gist, to replace an SDK in an APK you uncompress the APK, using ApkTool, replace the folder with your SDK files and re-compress & sign. The main tool we are using is Apktool. This tool can disassemble an application. This action is called baksmali and what it practically does is disassembling the classes.dex file (all the application’s classes transformed into bytecode) to smali files, which, unlike .dex files are readable. The opposite operation of assemble the apk is called “smali”. * Fun fact - “smali and baksmali” is assemble and disassemble in Icelandic. ### Sounds good! Now tell me how to do it... Here’s a step by step guide: - Disassemble the APK you want to test (So that you can find your SDK’s smali files in the APK and replace them).  You can do it by running this command: apktool d - The files are arranged by folders. Each package has its own folder so you can easily find your code's package name. You can also find your SDK package within the source code files. - Create a new apk with your new sdk. (It’s important to build this apk in release mode,assuming your customer’s app, which is in production, is also built in release mode). Once you have the new apk you need to run the apktool on this apk too, as described above (Now you have two disassembled apks in your hands). Replace the package folder of your new SDK smali files with the folder containing the files of your old SDK. - Now it’s time to assemble the apk back. In order to that you need to run this command: apktool b -o This will rebuild the application with the new version. -  Last but not least, you need re-sign the application. There are several ways to do it, the one I use is jarsigner. This is the command you need to run: jarsigner -verbose -sigalg SHA1withRSA -digestalg SHA1 -keystore my-release-key.keystore alias_name * Tip: As of JDK 7, the default signing algorithm has changed, requiring you to specify the signature and digest algorithms (-sigalg and -digestalg) when you sign an APK ### Final Result: You have the customer's app APK running with your new SDK version. Now you can run tests on this application when it is not yet in production, and without any work required from your customer. ** Tip: you can use the apktool not only to replace your sdk, but also to edit the AndroidManifest.xml - I will leave it to you to decide what and how to change there. --- ### Stop reading, start talking. A new way to share knowledge. URL: https://www.taboola.com/engineering/stop-reading-start-talking-a-new-way-to-share-knowledge/ Last Modified: 2025-01-14 14:37:11 Knowledge sharing is critical for every company that wants to grow and improve. The bigger the company - the harder it gets. Inefficiency, a lack of alignment within your peers, difficulty training new workers - you name it. In this post we will take a look at the existing methods for knowledge sharing. How they can’t keep up with growth and fast paced changes, and why people are your best resource for knowledge.   ### What is Knowledge? In general, there are three main types of knowledge that need to be shared in a software company - Technical Knowledge, Product Knowledge and Business Knowledge. When a new employee begins their job, most companies will help them to learn, using some of the more traditional methods to share knowledge: - Learn from others - via frontal training or assigning a mentor - Allow Self learning - online course, or from the company’s knowledge center (Atlassian, Convo, Jive) that holds company’s videos, or an internal Wiki, etc. ### Which types of knowledge are hard to deal with? Technical knowledge is being shared all the time on the web. Online courses, YouTube videos and professional forums teach and share the existing knowledge between professionals and novices alike. The other two types of knowledge (Product Knowledge and Business Knowledge) represent the company’s knowledge and they are almost impossible to share and manage, as the knowledge and the people who have it - changes constantly. The more developers and services you have - the more effort you will invest in sharing your knowledge and keeping it up to date. When there’s a gap between the existing knowledge being shared and the knowledge that isn’t, your best resource is your own people. They are the ones making the changes, they are the ones who can share knowledge with you when you need it. They hold the company's knowledge, which is what is needed. ### Sometimes all you need is a cup of coffee At Taboola, we deliver fast. We write thousands of lines of code every day and use continuous deployment. That means there’s a lot of knowledge to share, a lot of features to develop and a lot of changes are being made. That’s when knowledge sharing becomes a real issue and people become your best source. They are up to date. But if you’re a new employee, you soon find out that people are less available than a Wiki page and it takes you time to learn the names and responsibilities of your peers. Sometimes all you need is a cup of coffee with that specific person who can help you the most. 10 minutes talk and you can get the job done. ### Instead of knowledge, search for people “It is nothing for one to know something unless another knows you know it.” ― Persian Proverb In our latest Taboola Hackathon, my team and I tried to tackle that gap. How can you tell who is that specific person who has the most relevant and up to date knowledge? We thought it would be useful to have a search engine for the people who are the most relevant for what we need. That search engine has to be dynamic, automatic and not reliant on manual management. To do that, we would need to find an information source that gets updated anyway and not as part of an effort to maintain knowledge management. This description suits many systems we all use to manage our work as developers. For example Git, Slack, or any other task management systems. But how can we use them to find people? Let’s take Jira for example. In Jira, every task is represented by a ticket. Each ticket holds information about the task: who should develop the task, what the task is, who does the Code Review, etc. Now let’s say we want to find someone in our company that knows something about a product called ‘Items Manager’. A naive approach would be to find the person in our company who has solved the most tickets containing the phrase ‘Items Manager’. It could be within the tickets’ title, description, comments, and so on. ### Our solution - creating Wiser During our latest Hackathon we took that naive approach and made it more sophisticated. We thought having one source of information was not enough. Our results would be more accurate if we had multiple information sources. It would be even better if every source had different inputs, with different associated scoring. That way we would have an aggregated result that takes many parameters into consideration, making the result more relevant. Wiser was created in the Hackathon, and it’s our attempt to tackle the difficulty of reaching the knowledge that is not being shared. We decided to start with Jira as our source, since it holds a lot of useful information. The result was a list of people who were very relevant for each field of knowledge we were searching for. We are planning to keep on working on Wiser and make it available for Taboola’s employees to use. Hopefully making it a service available for other people as well in the future. ###### An example for a search result with Wiser ### Not every solution suits everyone Every company has different needs. Some companies has developers all over the world, others are geographically close. Every company is at different stages and needs different methods for knowledge sharing and management. Not every solution suits every company. Try to think about your company’s characteristics and find the solution that suits you best. Knowledge Sharing is a complex issue every company has, and it gets harder as the company grows. When the traditional knowledge sharing methods can’t keep up with the changes, people become your best resource of knowledge. So the next time you search for knowledge in your company, remember that sometimes all you need is a cup of coffee with that person who can help you. --- ### Collaborative Trial: On Optimizing Recommendation Testing URL: https://www.taboola.com/engineering/collaborative-trial-on-optimizing-recommendation-testing/ Last Modified: 2025-01-14 14:37:11 Taboola is responsible for billions of daily recommendations, and we are doing everything we can to make those recommendations fit each viewer's personal taste and interests. We do so by updating our Deep-Learning based models, increasing our computational resources, improving our exploration techniques and many more. All those things though, have one thing in common - we need to understand if a change is for the better or not, and we need to do so while allowing many tests to run in parallel. We can think of many KPI’s for new algorithmic modifications - system latency, diversity of recommendations or user-interaction to name a few - but at the end of the day, the one metric that matters most for us in Taboola is RPM (revenue per mill, or revenue per 1,000 recommendations), which indicates how much money and value we create for our customers on both sides - the publishers and advertisers. RPM can get a little noisy - it is influenced by external factors such as current events (how interesting the news is right now) or time of year (aren’t we all waiting for the holidays?) but also by internal factors such as system upgrades and our own experiments - which can affect one another and add extra noise to the mix. All those mean that RPM can change by as much as 5-10% from day to day. But an algorithmic improvement, on the other hand, might be even lower than 1%. So how can we tell if a change is for the better? All it takes is proper data handling and some pretty basic statistics. ## The Simpsons Let’s start with data handling. With the enormous amount of data we handle at Taboola, you can rest assured that we’ve seen many interesting phenomena that caused us to scratch our heads. For instance, take a look at the data in the table below, which shows how well two variants are doing on a four day period. If you look carefully you’ll see that in each of the 4 days, the test variant has higher RPM values than the default variant. However, when aggregating results from all 4 days the default variant all of the sudden has a higher RPM. This brings up two questions: - How is that possible? - Is the test variant winning or losing? default variant test variant date RPM count revenue RPM count revenue 2020-05-01 1.358 123,651,344 167,896 1.399 41,385,732 57,888 2020-05-02 1.190 157,182,560 186,992 1.200 77,811,946 93,370 2020-05-03 1.068 151,712,429 161,976 1.075 74,929,584 80,538 2020-05-04 0.988 141,997,417 140,288 0.995 70,248,390 69,864 Grand Total 1.144 574,543,750 657,152 1.141 264,375,652 301,661 Table 1: Daily stats for two variants over the course of 4 days First off, this is not only possible, but can actually happen quite often. This phenomenon is called “Simpson’s paradox”, where combining several groups together cancels out a trend that appears individually in each group (you can read more on Simpson's Paradox here). The answer to the second question is hidden in the first day - take a close look at the second table below, which presents the traffic counts for each variant and the ratio between them: date default count test count count ratio 2020-05-01 123,651,344 41,385,732 0.33 2020-05-02 157,182,560 77,811,946 0.50 2020-05-03 151,712,429 74,929,584 0.49 2020-05-04 141,997,417 70,248,390 0.49 Grand Total 574,543,750 264,375,652 0.46 Table 2: Comparing the number of recommendations each variant had It can be seen that in the first day of the experiment, the ratio is way off (0.33 compared to 0.49-0.5 on other days). Looking closely into the matter we found out that we included the day of the variant setup in the experiment, so it was a full day for the default variant and a partial day for the test variant. Because the first day happened to be the most profitable day by far (check out the RPM columns in the first table), and because the test variant got a smaller chunk of it, its overall measure was lower even though it had better measures on all 4 days. What we learn from this example is that taking a look at the traffic split between variants, and making sure it is consistent, is important if we wish to avoid strange skews in the data. Our testing procedure in Taboola tests for such cases and others, and gives the researcher a green light if the experiment is winning with confidence, a red light if it is losing with confidence and an alert if some issues were found that need a closer look.   ## It’s T-Time And how exactly do we gain this confidence? This brings us to the statistical part of this blogpost, or more specifically, to a paired sample t-test. Paired sample t-test is used when we can match each observation from one sample to an observation from the other. For our case we use the daily average revenue as a single data point, and the pairing is done over days. This means that for each variant we have a sequence of numbers and just like a normal t-test, we try to reject the null hypothesis: : The means of the test variant RPM and default variant RPM are equal But while a normal t-test goes like this: (  and  are the unbiased estimators for the sample mean and sample variance respectively) A paired sample t-test goes like this: The great advantage of the paired sample t-test over the normal t-test is that it allows us to cancel the time related noise and focus on the difference between the two variants. For example, take a look at the results given in the following figure: Figure 1: Left: average revenue (per recommendation) of two variants over the course of 21 days Right: the difference in the average revenue along the same 21 days Using the normal t-test to compare the two variants, we get a p-value of ~0.7, which means we have about 30% confidence in rejecting the null hypothesis that the two variants have equally distributed means. This happens because the difference between days is much greater than the average difference between test and default. However, using the paired sample t-test yields a p-value of ~0.001, or 99.9% confidence in rejecting our null hypothesis (I will not dive into the subtleties of the confidence term, let’s keep it simple). If you need more intuition as to why the paired sample t-test improves the confidence to such an extent, compare the figure above to the one below - which shows a permutation of the days for test and default data such that we can no longer pair by days. In a sense, the un-paired t-test uses the data representation below while the paired t-test uses the data representation above. Figure 2: Same test and default variants, but this time days are permuted randomly so that pairing is not known     Whenever using a statistical test, it is important to ask ourselves if it’s the right one. In our case, we made sure our data meets these 3 requirements for using a t-test: - The sample means for both populations are normally distributed. For large samples this assumption is met due to the central limit theorem, but for smaller samples this needs to be tested. We use daily aggregated data so our sample is as big as the amount of days in the experiment, which is typically not very large. To test this requirement, we used the Shapiro-Wilk normality test and got good results. - The two samples have the same variance. This is only required when the samples are not of equal size. We have equal size samples, and t-test is very robust to cases with unequal variance, so no worries there. - The sampling should either be entirely independent or entirely paired. Our daily data is linked through the general behavior on each day, which means we need to perform a paired test, and that is exactly what we did. ## Final Words When we are testing new variants of our system we face contradicting interests. On the one hand, we want to make decisions fast - removing losing variants quickly and diverting more traffic to the winning ones yields greater value. On the other hand, the more time we give our variants, the more certainty we have in their performance, reducing the chance of making wrong decisions. This post gives a peek at how we use some pretty basic statistical principles and combine them with general “know-hows” of data handling in order to balance those interests, and make sure we gain as much information as possible from the time we give our variants. Have any thoughts? Let us know!   --- ### How Taboola Handles Deep Learning Training-Serving Features Discrepancies URL: https://www.taboola.com/engineering/discrepancy-solution/ Last Modified: 2025-01-14 14:37:10 ## Introduction In machine learning, one critical challenge is ensuring that the features used during model training (offline) match those used during inference time (online serving). Discrepancies between training and serving features can lead to significant performance degradation, making it crucial to identify and address these inconsistencies as fast as possible. At Taboola, we specialize in content discovery and native advertising, enabling users to find and engage with personalized content across the web. Our advanced machine learning powered recommendation systems serve billions of recommendations every day, helping publishers, advertisers, and brands reach their target audiences effectively. In this blog post, I’m going to discuss the challenges of training-serving feature discrepancies in machine learning models and how they can affect model performance. I’ll explain how we tackle these discrepancies at Taboola, including the design and implementation of our solution, Sherlock. Finally, I’ll discuss some of the key discoveries made and the significant impact Sherlock has had on our recommendation systems. ## Discrepancies?? No Way… Training-serving feature discrepancies can occur for several reasons. The most common reason is the way we handle large features. Due to their significant storage requirements, these features are often not reported back as-is after serving in the online environment. Instead, before training, they are re-calculated from sampled data and various data sources. The re-calculated features may differ from the original calculations performed during serving. This may be due to data differences, changes in external data between serving and training times, or calculation logic differences. Another cause of discrepancies can be cache misses and database queries timeouts in the online serving environment. For Taboola to be able to return recommendations to the client in a matter of hundreds of milliseconds we must wrap database queries in a very strict limit. These misses and timeouts can result in certain features having no values in serving while in the report back pipeline they will have values due to less strict timeouts. Additionally, integration bugs can introduce further discrepancies. Poorly integrated components or erroneous data pipelines can cause mismatches between the features used during training and those available during serving. ## Meet Sherlock In order to tackle this issue, we developed Sherlock. Sherlock is a robust system designed to detect and alert on training-serving feature discrepancies in our models. Developed as part of our continuous efforts to improve the reliability and performance of our recommendation systems at Taboola, Sherlock allows us to promptly detect discrepancies and quickly address them, ensuring consistent and accurate model performance. ## How Sherlock Works Sherlock pipeline is composed of three main parts: - Sampling: Saving a sample of feature values that were used for the serving. - Preprocess: Building the training feature values. - Compare: Comparing the serving and training feature values and raising warning/alerts based on the results. Figure 1: Sherlock’s main parts ## Stage 1 - Sampling Saving all feature values from the serving environment is impractical due to the enormous volume of data it would generate. To tackle this challenge, we implemented a sampling strategy where feature values are reported for 1 in every 10,000 requests handled at Taboola. This approach significantly reduces the data volume while still providing a representative sample of the feature values used in serving. The sampled feature values are included in Pageview objects in HDFS. A Pageview is a record that encapsulates various details about a user’s visit to a webpage, including recommendations served, user activities such as clicks, and more. To further streamline the process, we run a Spark job that aggregates these Pageviews containing the sampled features values once an hour. This job collects all relevant Pageviews from the previous hour that have the sampled data and copies them to a dedicated location in HDFS, making them easily accessible for the next stage of Sherlock’s workflow. Figure 2: sampling   ## Stage 2 - Preprocess Preprocess refers to a group of Spark jobs designed to prepare data for the model training jobs. These jobs read Pageviews data from HDFS based on a range of dates and perform a series of operations, including filtering, enrichment with external data sources, and calculations. The result is a comprehensive dataset containing all the features of the model, along with train and test folders filled with TFRecords used as inputs to the training job. All these outputs are then stored in HDFS. For Sherlock, our primary focus is on the features dataset. To accommodate Sherlock’s requirements, we developed a special mode within our existing Preprocess jobs, characterized by the following key points: - Reading Input Data from Sampled Pageviews: The job reads input data exclusively from the path specified in stage 1, which contains only the Pageviews containing the sampled serving features values. - No Filtering or Sampling: Unlike our standard preprocess before training job, this special mode processes all the data without any filtering or sampling, ensuring that all the data is processed. - Features Calculation: The job calculates all the features for training as it would in a standard preprocessing run, ensuring that the training feature values are calculated just as they are calculated in the standard production process. - Including Serving Sampled Feature Values: In addition to the calculated features, the job includes the features from the serving sampling exactly as they were recorded. This results in a dataset where each feature is represented twice: once with the serving values and once with the Preprocess calculation (the one that should have been used while training the model). Figure 3: preprocess ## Stage 3 - Compare The Compare stage is the core of Sherlock, where the serving and training values of the features are analyzed and discrepancies are detected. This job is implemented in Python and leverages the Pandas package, enabling the efficient and fast comparison of large amounts of data. Since there are various types of features, such as integers, strings, lists and more, some comparisons are straightforward while others are more complex. Basic types are automatically detected from the data, whereas complex types are defined per feature in configuration files. Each feature is configured with warning and alert thresholds to identify significant discrepancies. Additional metrics, such as the percentage of Out of Vocabulary (OOV) values, are also calculated for each feature. Once the comparison is complete, the results are uploaded to Google BigQuery and are also made visible in our easily accessible logs. To automate the monitoring of these results, we have defined auto-scheduled queries over BigQuery that check for features whose comparison results exceed their warning or alert thresholds. For features that exceed the warning threshold, a Slack message is sent to a dedicated channel. For features that exceed the alert threshold, a Slack message is sent along with a PagerDuty notification to the on-call person, ensuring prompt attention to critical discrepancies. By implementing these automated checks and notifications, Sherlock ensures that any significant discrepancies between training and serving features are quickly identified and addressed, maintaining the integrity and performance of our machine learning models. Figure 4: compare  ## Discoveries and Impact of Sherlock Since its release, Sherlock has identified multiple broken features, allowing us to make critical fixes that have significantly improved our models. Here are some examples of the types of bugs Sherlock has uncovered: - Non-Equal Logic in Feature Building: One common issue was the use of different logic for building features during serving and training. The fix involved ensuring that the same function was used on both sides, eliminating duplicate logic that could be mistakenly altered on one side only. - Inconsistent Feature Calculations: In some cases, features were calculated differently for serving and then recalculated with different logic for inclusion in the Pageview resulting in a broken report back value. The solution was to calculate the feature only once and use the same value twice. When this wasn’t feasible, we made sure the calculation logic remained consistent. - Configuration differences: We have had multiple cases where the building logic of the feature is equal for serving and training but the configuration wasn’t the same. For example, the OOV values weren’t equal on both sides resulting in inconsistent models. One of the most notable fixes involved correcting a broken feature that led to a significant increase in Taboola’s global Revenue Per Mille (RPM) by 1.31%. This improvement underscores the substantial impact Sherlock has had on our system’s performance. A significant impact of Sherlock is to allow us to make the right calls regarding our models. For example, when Sherlock was first activated, it flagged several features as broken—features that were about to be removed from our models due to their perceived lack of value (which, of course, was because they were broken). Instead of removing them, we fixed the features and saw the positive impact they had when functioning correctly. ## Conclusion Training-serving feature discrepancies can be a real pain, reducing the accuracy of machine learning models and potentially leading to financial losses. When we noticed that features which previously boosted our models’ RPM stopped doing so, it became clear that we needed a solution like Sherlock. The impact of Sherlock has been substantial for Taboola. By ensuring that our features are free from discrepancies, it has enhanced the reliability and accuracy of our machine learning models. Sherlock enables us to identify and address bugs promptly, even during the development and testing phases of new features. Moreover, it provides a safety net, alerting us if a feature is accidentally broken, and helps maintain the integrity of our recommendation systems. --- ### LLM-Gateway: Streamlining Large Language Model Integration at Taboola - Part 1 URL: https://www.taboola.com/engineering/llm-gateway/ Last Modified: 2025-01-14 21:23:42 ## Introduction & Overview As the landscape of LLMs rapidly evolves, leading providers such as OpenAI, Azure GPT, Gemini, and others are emerging as pivotal players, each pushing the boundaries of what’s possible. Businesses across various sectors are increasingly harnessing the power of LLMs to drive intelligent, data-driven solutions that enhance operational efficiency and user engagement. This blog delves into how Taboola has developed the LLM-Gateway, a unified solution that addresses operational challenges, streamlines interactions with various LLM providers, and enhances the scalability and reliability of our internal services. ## Leveraging the Power of LLMs at Taboola At Taboola, we recognize the transformative potential of LLMs. From automating customer service interactions to powering advanced data analytics and natural language understanding, LLMs are reshaping our operational strategies and elevating the experiences we deliver. Their versatility and adaptability allow us to innovate continuously, maintaining our competitive edge in a fast-paced industry. ### Key Applications of LLMs at Taboola To enhance user experience and optimize advertisement effectiveness—ensuring ads reach the right audiences on publishers’ sites—we have integrated LLMs into several core components. These integrations include intelligent systems such as the Intelligent Abby AI Agent, GenAI Ad Maker, and Automated Advertisements Filtering for Publishers. By leveraging LLMs, we have significantly improved process efficiency and overall performance across these applications. - Intelligent AI Agent - Abby: To maximize advertisers’ success, we developed Abby, an intelligent AI agent designed to streamline the creation and management of ad campaigns. Abby leverages LLMs to understand and respond to advertisers’ requirements, enabling effortless campaign setup and enhancing the onboarding process. Additionally, Abby assists in managing ongoing campaigns by suggesting target audiences and geographic locations based on campaign budgets and key performance indicators (KPIs). Through interactive conversations with advertisers, Abby improves overall campaign performance by providing data-driven recommendations and optimizations. - GenAI Ad Maker: Taboola’s GenAI Ad Maker empowers advertisers with the capability to generate high-quality, customized ads that align with their brand’s unique identity and target audience. Utilizing LLMs, GenAI Ad Maker can create compelling advertisement titles and descriptions from limited input information. For existing ad campaigns, it offers functionalities to rephrase content and rectify any issues, thereby enhancing campaign editing efficiency. This tool streamlines the ad creation process, enabling advertisers to produce effective ads quickly and consistently. - Automated Ad Compliance for Publishers: We use LLMs to automatically classify and tag ads, flagging any potential inappropriate content before it reaches publisher sites. By matching these tags and classifications with our content policy, we make sure that only suitable ads are shown to our publisher partners' audiences. This approach not only boosts ad performance but also helps publishers avoid unwanted content, leading to a better experience for users. Large Language Models (LLMs) are crucial for building Taboola’s key features, enabling functionalities that would be extremely challenging without them. LLMs power the Intelligent AI Agent Abby, facilitating seamless campaign setup and optimization through advanced language understanding. They drive the GenAI Ad Maker, allowing for the efficient creation and customization of high-quality ads, and automate advertisement filtering to ensure appropriate content on publisher sites. Without LLMs, developing these sophisticated and scalable features would be significantly more difficult. ## The Challenges After Adopting LLMs Integrating Large Language Models (LLMs) into our services at Taboola has unlocked unprecedented levels of efficiency and capability. However, this extensive adoption also surfaced a range of significant challenges: - Limited trackability: One of the major issues we faced after heavily adopting LLMs is the limited trackability. With multiple services utilizing LLMs simultaneously, it became increasingly difficult to gain clear visibility into resource consumption, token usage, and model performance. - Escalating costs: Each LLM API request consumes tokens, and as the volume of requests across our services increased, monthly expenses began to soar.What began as a manageable expense quickly snowballed into a significant financial strain. - Difficulties with Seamless Failover for Continuous Availability: Implementing seamless failover to ensure continuous availability has become a significant pain point. Each of our services needs to build its own retry and fallback mechanism, resulting in repetitive boilerplate code across multiple services. This not only increases development complexity and leads to inconsistency in how failover is handled. - Complex Integration and Lifecycle Management: One of the key obstacles the Taboola internal services face is the complexity of integrating with different LLM models and managing their life cycles. Each model has its own set of requirements, versioning, and updates, making it difficult to ensure compatibility and smooth operation across various services. - Security Risks in Managing Shared API Keys: In our organization, several services access LLM models using the same API key, which makes it difficult to manage granular access control. This setup creates a vulnerability: if we need to restrict or prevent one service from using the LLM model, it’s almost impossible to do so without impacting other services that rely on the same key. ## LLM-Gateway: Our Unified Solution for LLM Challenges To address the cross-cutting challenges mentioned above, we developed the LLM-Gateway—a centralized intermediary layer between our internal services and various LLM providers and models. The LLM-Gateway offers a unified interface that simplifies interactions with multiple LLM models, while providing enterprise-grade features such as enhanced observability, easily integrated interfaces, intelligent traffic load balancing, customizable failover strategies, a configurable caching mechanism, and robust security protocols. ### Design Decisions Behind the LLM-Gateway Before diving into the architecture of the LLM-Gateway, it is crucial to understand the two core concepts behind the scenes: Virtual Models and a Unified Request & Response Data Schema. #### Deep Dive into Virtual Models At the heart of the LLM-Gateway lies the concept of Virtual Models. A Virtual Model abstracts the complexities of underlying LLM Vendor Models, presenting a unified interface to client applications. This abstraction allows clients to interact with LLMs without needing to manage details such as regional deployments or specific model parameters. ##### Virtual Model Request Routing To better understand the integration between LLM Vendor Models, the following diagram illustrates how each client service uses a Virtual Model to connect with the actual LLM Vendor Models. Between the Virtual Models and LLM Vendor Models, there is an additional layer called the Service Policy. This layer allows the Virtual Model to configure how it utilizes the associated LLM Vendor Model when serving a request. The Request Routing Flow: - LLM-Gateway clients submit request with Virtual Model: The client service sends a request to the LLM-Gateway, specifying the desired Virtual Model identifier. - LLM-Gateway server validates Service Policy: The LLM-Gateway retrieves the associated Service Policy, which includes rules such as TPM limits and model preferences for an LLM Vendor Model, and performs pre-processing operations before forwarding the request to the LLM Vendor Model. - LLM-Gateway server forwards requests to LLM Vendor Model: Once the pre-processing operations are finished, LLM-Gateway selects the associated LLM Vendor Model and forwards the request accordingly. #### The Request and Response Data Schema To maximize the effectiveness of Virtual Models, we standardized the request and response formats into a unified data model. This approach simplifies client interactions with multiple LLM vendor models by abstracting the complexities of model management and routing. A pivotal design decision was to align our API structure closely with the widely-adopted OpenAI ChatCompletions API. This alignment leverages the extensive ecosystem and familiarity that many developers have with OpenAI’s interface, simplifying integration across our services. While adhering to the OpenAI framework, we have also introduced support for extra parameters to integrate with our in-house Llama3 model. This slight extension maintains compatibility with internal models while preserving the simplicity and familiarity of the OpenAI API. In the demonstration above, clients familiar with the OpenAI ChatCompletions API can easily switch to the LLM-Gateway. Additionally, the standardized data formats enable efficient caching mechanisms, improving performance by reusing responses across services. This approach not only streamlines integration but also boosts development productivity by reducing the overhead of managing diverse model interactions. ### Exploring the Internal Mechanism of the LLM-Gateway Building on top of the Virtual Model and the Unified Request and Response Schema, we developed several features to address the practical challenges mentioned earlier. This section delves into the inner workings of the LLM-Gateway, highlighting the mechanisms and processes that power its core functionalities. From intelligent load balancing and dynamic routing to robust fallback strategies, we’ll explore how these internal systems work together to deliver high performance, reliability, and efficiency. #### LLM-Gateway - Core Features The diagram above illustrates the layered architecture of the LLM-Gateway. It depicts the key components across different layers, including LLM-Gateway clients, the LLM-Gateway internal infrastructure, and various LLM Vendors. Next, we will explore the core features that address client challenges and make the LLM-Gateway a powerful solution: ##### Smart Load Balancing The Smart Load Balancing feature was developed to mitigate Difficulties on Seamless Failover for Continuous Availability issue, it intelligently routes requests based on predefined Service Policies associated with each Virtual Model.   In the diagram, a Virtual Model like taboola-service-A-gpt-4o can distribute traffic between multiple LLM Vendor Models based on TPM configurations. Here’s how it works: For the virtual model taboola-service-A-gpt-4o, with a total TPM limit of 40K (30K + 10K), the traffic is distributed as follows: - us-west-gpt-4o: Handles 75% of traffic, based on 30K TPM limit settings. (75% = 30K/40K) - us-east-gpt-4o: Handles 25% of traffic , based on 30K TPM limit settings. (25% = 10K/40K) ##### Customizable Fallback Routing Another feature for dealing with Difficulties in Seamless Failover for Continuous Availability issue is Customizable Fallback Routing. This feature allows clients to specify fallback Virtual Models in the request parameters. If the primary model fails or exceeds its limits, the gateway automatically reroutes requests to the designated fallback model, ensuring uninterrupted service. ## In the above graph, ServiceA sends the ChatCompletions API request with the fallback models, and LLM-Gateway does the error fallback internally when the primary model `taboola-llama3` process fails. ##### Configurable Caching System To address the Substantial Costs problem, we provide an Configurable Caching System that allows clients to use an HTTP request header to manage whether to retrieve cached results or request fresh data. This flexibility ensures that up-to-date responses are available when needed, while caching reduces unnecessary costs. By offering this configurable caching option, the LLM-Gateway enhances efficiency and provides clients with a flexible way to manage and optimize their LLM-related expenses. ##### Unified Model Management To overcome Model Integration and Lifecycle Management pain points, the Unified Model Management leverages Virtual Models to simplify the management of model lifecycles. As models evolve or new versions are introduced, LLM-Gateway clients can easily switch to their desired LLM vendor models by updating the configuration of their Virtual Models—no changes to client code are necessary. This unified approach significantly reduces the complexity and overhead associated with managing model lifecycles and integrations, benefiting both our organization and our clients. ##### Client Application Authorization With the introduction of Client Application Authorization, the LLM-Gateway now effectively addresses both Trackability Limitations and Security Risks with Shared API Keys. This feature centralizes access control, ensuring that each client application is assigned distinct authorization credentials. It strengthens security by preventing unauthorized access and reducing the risk of data exposure. Additionally, it offers comprehensive visibility into how each client application interacts with LLM resources during the entire request lifecycle. With real-time insights into resource usage, token consumption, and the performance of various LLM models, this feature empowers us to monitor and optimize operations effectively. ### Summary The LLM-Gateway serves as a centralized intermediary between Taboola’s internal services and various LLM providers. It simplifies model management through Virtual Models, allowing seamless updates and lifecycle management without altering client code. Additionally, the gateway enhances security and visibility by centralizing access control and monitoring resource usage, providing clients with real-time insights and control over LLM operations. These solutions collectively create a streamlined, scalable, and secure environment for leveraging LLM technology. ## Real-World Impact of the LLM-Gateway Since its launch, the LLM-Gateway has revolutionized how our client applications interact with LLM models, addressing critical difficulties such as cost efficiency, seamless failover, and intelligent load balancing. In this section, we delve into the real-world impact of the LLM-Gateway, supported by key monitoring metrics and insights from our applications. ### System Metrics and Monitoring One of the most significant advantages post-launch has been the ability to collect comprehensive monitoring metrics—insights that were previously unattainable when client services interacted directly with LLM vendors. These metrics provide invaluable visibility into system performance, resource usage, and operational efficiency. #### Requests Metrics Client request count:  This metric is essential for assessing usage patterns, identifying peak usage times, and ensuring that the gateway can handle the load effectively. Client request duration Monitoring request durations helps in identifying performance bottlenecks, optimizing processing times, and ensuring a seamless user experience across all services. Client requests traffic across different LLM Model providers This information is crucial for load balancing, optimizing resource allocation, and negotiating with LLM vendors based on usage patterns. #### Caching Metrics Caching hit rate per client application  This metric is vital for evaluating the efficiency of the caching system, identifying opportunities for cache optimization, and minimizing redundant LLM API calls to control costs. #### Token Metrics Token usage per client application By analyzing this metric, teams can identify high-usage applications, implement token-efficient practices, and ensure that usage remains within budgetary constraints. ## ### Lesson Learned #### Caching vs Cost Since the launch of the LLM-Gateway, one of the most impactful strategies has been the implementation of a robust caching mechanism. Given that each LLM API request incurs token-based costs, our caching strategy has been instrumental in reducing redundant calls and optimizing overall performance. - Cost Savings: One of our client applications experienced a dramatic reduction in monthly costs, saving approximately 75% of total expenses. This significant saving was achieved by maintaining a consistent cache hit rate between 60-80% across various applications. - Performance Improvement: By serving responses from the cache instead of making repeated LLM API calls, we have not only reduced costs but also enhanced response times and overall system efficiency. These observations highlight the critical role caching plays in mitigating the high costs associated with LLM usage, making it a cornerstone of our approach to optimizing performance and maintaining cost-effective operations. #### Request fallback The Request Fallback mechanism has been another standout feature of the LLM-Gateway. Integrated seamlessly into the standard API request body, this feature has been widely adopted by our clients due to its simplicity and effectiveness. - Increased Availability: Clients experience higher request availability without implementing complex retry logic. - Reduced Complexity: The ease of integrating fallback directly into the request body has simplified client implementations, reducing development overhead and improving reliability. ## Conclusion The launch of the LLM-Gateway has been a pivotal advancement in optimizing how our clients and organization interact with Large Language Models (LLMs). By addressing difficulties such as cost management, high availability, security, and comprehensive monitoring, we have developed a solution that significantly enhances performance, reliability, and flexibility in LLM deployments. On an organizational level, the LLM-Gateway has provided critical visibility into LLM request usage, enabling better resource management, performance analysis, and data-driven decision-making. By centralizing monitoring and tracking, the gateway allows us to optimize usage and control costs more effectively. In our next article, we will explore advanced technologies supported by the LLM-Gateway, such as BulkChatCompletions—a bulk version of the ChatCompletions API—and other enhanced caching features. These technologies further maximize the LLM-Gateway’s performance, empowering businesses to harness the full potential of LLMs while maintaining efficiency and cost-effectiveness at scale. --- ### Zooming Past the Competition URL: https://www.taboola.com/engineering/zooming-past-the-competition/ Last Modified: 2025-01-14 14:37:12 Imagine you’re walking down the street and you see a nice car you’re thinking of buying. Just by pointing your phone camera, you can see relevant content about that car. How cool is that?! That was our team’s idea that awarded us first place in the recent Taboola R&D hackathon aptly named - Taboola Zoom!   Every year Taboola holds a global R&D hackathon for its 350+ engineers aimed at creating ideas for cool potential products or just some fun experiments in general. This year, 33 teams worked for 36 hours to come up with ideas that are both awesome and helpful to Taboola. Some of the highlights included a tool that can accurately predict the users’ gender based on their browsing activity and an integration to social networks for Taboola Feed. Our team decided to create an AR (Augmented Reality) application that allows a user to get content recommendations, much like Taboola’s recommendations widget, based on whatever they’re pointing their phone camera at. ## What is Zoom? The app itself is an AR experience similar to that of Google Glass, which allows you to interact with the world using your phone camera. Using the app, one just needs to point their camera onto an object to immediately get a list of stories from the web related to that object. To make this idea a reality we used technologies from several domains - AI, Web Applications and Microservices. ## Under the Hood The flow is pretty simple: - The user opens a web application on his phone - an HTML5 page that behaves like a native app. - The app sends a screenshot of the captured video every second to a remote server. - The server processes the image using computer vision technologies. - The server searches for web articles with thumbnails that are the most similar to the processed image. - The retrieved images are filtered using a similarity threshold, and are sent back to the user to be shown in a slick widget atop of the user’s display. - When the user clicks the widget the relevant article is opened in the browser.   ## Zooming In Each component of the system is implemented using different technologies:   ## Web Application We decided to implement the user interface using HTML5, which allowed us access to native capabilities of the phone such as the camera. Additionally, we used WebRTC API and Canvas API.   ## Computer Vision Service For every image, the service processes the image and returns an embedding - a numerical vector representing information extracted from the content of the image. Understanding the content of an image is a well known problem with plenty of solutions. We chose to use Google’s Inception model, which is a DNN (Deep Neural Network) trained to classify the object found in an image. A DNN is a construct of layers of neurons - similar to nodes in a graph, where each layer learns certain patterns in the image. The first layers learn to output simple patterns such as edges and corners, while the last layer outputs the type of the object, e.g. dog, cat etc. We chose to use the output of the layer before last - as that produced the best results.   ## Database The only component that was already available to us was Taboola’s internal database of articles, containing, among other things, the thumbnail and the article URL. If such a database was not readily available to us, we could have just built our own by scraping images using a library such as BeautifulSoup.   ## Server We used Flask as our web server. On startup, the server queries Taboola’s internal database of articles from the web. It then sends each image to the computer vision service, which returns an embedding. The embeddings are then stored into a designated data structure called FAISS (Facebook AI Similarity Search). It allows us to perform a nearest neighbors search. Each image sent from the user is similarly transformed into an embedding. It is then used as a query to the above data structure to retrieve its nearest embeddings, meaning, images with similar patterns and content. Only images which are considered similar above a predefined threshold are then returned to the user. So to recap, the entire architecture relies on three main components: - web application - computer vision service - articles database ## Do it Yourself The entire project was developed in under 36 hours by a team of 5 people. This app touches several interesting and exciting domains that are “hot” in the industry - AR and AI. It was a breeze to implement thanks to the commonly available tools and libraries. If there’s only one thing we want you to take from this is that it’s not that difficult. We invite you to be aware of the potential that lies in these domains and to be on the lookout for interesting and exciting ideas. Once you find one, go and have your own private hackathon! ## Finally If you want to play around with the app, open https://zoom.taboola.com on your phone- use Chrome on Android and Safari on iOS, and try hovering over different objects to see the various results - keep in mind this is a pre-alpha hackathon project. We want to thank our wonderful teammates who worked tirelessly to create this amazing app - Amir Keren, Yoel Zeldes, Elina Revzin, Aviv Rotman and Ofri Mann. --- ### Driving Success Through Cross-Functional Management in Engineering at Taboola – Part 2: Strategic Leadership and Innovation in Tracks Methodology URL: https://www.taboola.com/engineering/tracks-process/ Last Modified: 2025-02-19 17:24:12 ## Introduction In the first part of this series, we explored how the Tracks methodology empowers teams through cross-functional collaboration, delivers measurable business impact, and fosters innovation. In this second part, we’ll dive deeper into the high-level management of Tracks, including how mission statements and resources are reviewed, and how leaders like group managers and track managers balance their responsibilities. We’ll also take a closer look at how innovation is driven through the Discovery mode and how the Tracks methodology supports leadership and strategic decision-making. ## Tracks Management Forum and Portfolio Process The success of the Tracks methodology at Taboola is driven by the careful oversight and resource allocation facilitated by the Tracks Management Forum. This small forum includes VPs and a Sr. Product Director and is responsible for orchestrating the leadership of Tracks, from defining their mission statements to assigning resources. - Coordinating the Tracks: The Tracks Management Forum ensures that each Track is aligned with Taboola’s broader strategy by defining the mission of each Track, selecting a capable Track Leader, and allocating resources (teams and other assets), which are reviewed and approved by the SVP R&D, VP Product  and members of the Senior Executive Team (SET) as part of the portfolio process. - Portfolio Process for Resource Approval: Resources for Tracks are approved based on a repeated 6-month portfolio process, led by the SVP R&D along with the VP of Product and other members of the SET. This process takes place before the beginning of each year and again mid-year to ensure that Tracks are properly resourced to meet their goals. ## The Three “Hats” of Leadership In Taboola’s tech organisations, leaders wear three distinct "hats" that define their responsibilities: - People Management: Leaders at all levels—whether team leaders, group managers, or VPs—are responsible for managing the growth and performance of their teams, supporting their professional development, and fostering a positive and productive working environment. - Project Management: Depending on their role, leaders oversee either smaller tactical projects or large-scale initiatives, ensuring timely delivery and resource alignment. - Craft (Technical Expertise): Leaders are responsible for maintaining and enhancing the technical expertise within their teams, driving innovation and ensuring high-quality output. At the team leader level, all three hats—people management, project management, and technical expertise—are worn simultaneously within their own team. This means team leaders are not only responsible for managing their team’s day-to-day performance and professional development but also for driving successful project execution and ensuring a high level of technical proficiency within the team. They play an essential role in fostering a positive, productive working environment. While team leaders focus on managing the full scope of their teams' work, group managers in the Tracks structure have a more flexible role, overseeing their teams while allowing certain projects or tasks to be led by other Track Leaders. In the cross-functional model, group managers might lead Tracks while some of their teams are assigned to Tracks led by others. Unlike team leaders, who wear all three hats—people management, project management, and craft—directly within their own teams, group managers in the Tracks structure often do not directly manage the day-to-day tasks or projects of all their teams. Instead, their role is to ensure that their team members are successful within the Tracks they are assigned to, even when these projects are led by another Track Leader. This includes offering guidance, supporting team leaders, and ensuring their team’s alignment with the broader goals of the Track. This flexibility ensures that the best people are always working on the most impactful projects, regardless of departmental boundaries, while group managers remain responsible for their team’s overall development and performance. ## Opportunities for Emerging Leaders While most of our Tracks are led by managers who already manage a group and are additionally nominated as Track Leaders, we also recognize potential in emerging leaders. In some cases, we nominate managers to lead Tracks even before they become group managers. These are typically leaders who have experience managing teams and demonstrate the capability to drive larger, cross-functional initiatives. For these emerging leaders, taking on the Track Leader role allows them to focus more on Project Management and Craft—ensuring successful project execution and maintaining high technical standards across their cross-functional teams. Since they are not yet managing organic teams, People Management becomes less central to their role, allowing them to concentrate on driving the Track’s goals and technical outcomes. This flexibility offers a valuable path for managers to grow into broader leadership roles. By focusing on leading high-impact projects and honing their technical expertise, they gain the experience necessary to eventually balance all three hats—People Management, Project Management, and Craft—as they grow into full group managers in the future. Note: Similarly, senior Product Managers (PMs) may also take on the Track Leader role as part of their career growth, allowing them to experience the benefits of cross functional management of non-PMs, as described above. ## Benefits of the Tracks Methodology for Managers The Tracks methodology at Taboola offers significant advantages for managers, enhancing their leadership capabilities, improving strategic decision-making, and driving career growth through cross-functional collaboration. ### Greater Visibility and Direct Impact on Business KPIs In the Tracks model, managers hold end-to-end accountability for projects, giving them a direct influence on business outcomes and key performance indicators (KPIs). Successfully leading high-impact Tracks provides increased visibility within the company, opening pathways for career advancement through measurable achievements. ### Efficient Resource Management Tracks provide managers with the flexibility to dynamically allocate resources based on evolving project needs. This approach ensures that the best talent is aligned with the most mission-critical projects, improving overall efficiency and accelerating progress toward business objectives. ### Expanding Leadership Across Groups and Departments Leading a Track means managing cross-functional teams across multiple groups and departments. This experience helps managers develop broader leadership skills, enabling them to handle complex, multi-team initiatives and work effectively beyond the boundaries of their own department. ### Driving Innovation and Creativity The Tracks methodology empowers managers to make strategic decisions and champion a culture of innovation. It encourages calculated risk-taking and experimentation, which is further supported by Discovery Mode—a structured process for exploring uncertain opportunities and testing new concepts (see more on Discovery Mode in the next section). By leading innovative initiatives, managers can foster creative problem-solving across teams. ### Strategic Career Growth Leading high-impact Tracks allows managers to build a robust track record of success, positioning them for further leadership opportunities. In addition to sharpening their project management and technical skills, managers gain valuable insights into broader business objectives, enhancing their ability to align technical execution with business strategy. ### Balancing Structure with Leadership Flexibility The Tracks methodology offers a structured framework for managing leadership responsibilities while remaining adaptable to changing business priorities. This balance of structure and flexibility allows managers to focus on delivering results while adjusting team direction as business needs evolve. ## Discovery Mode: Exploring Uncertain Opportunities Within Tracks While Tracks are typically focused and impact-driven, Taboola also utilises Discovery mode for concepts, projects or sub-projects that need further exploration or research. Discovery allows us to explore opportunities that may not have immediate clarity but show potential worth investigating. - Exploration with Lower Certainty: Discovery projects allow us to inspect ideas and run Proofs of Concept (POCs) to determine whether they should evolve into a full Track or sub-project. - Defining Future Tracks: Sometimes, a Discovery phase leads to the establishment of a new Track or the addition of a sub-project to an existing Track. If the potential is significant, Discovery ensures that it receives the attention it deserves. - Dedicated Resources: Discovery Mode is not a "nice-to-have" or an afterthought; it’s a crucial part of the portfolio process, with dedicated resources allocated to it. This ensures that exploration and validation of new opportunities receive proper attention and are aligned with company goals. - Flexible Outcomes: Not all Discovery projects move forward, but the process provides a space for safe exploration, minimising risk and ensuring that only the most promising ideas are pursued. By embracing a fail fast approach—described in more detail in Part 3—teams can quickly identify which ideas to pursue and which to discard, minimising resource investment in less viable paths. This approach, rooted in early failure, encourages teams to stop or pivot efforts early when progress becomes uncertain, focusing on alternative strategies that can better achieve impact. This agility allows us to refine concepts early on, ensuring that only the most promising initiatives progress, while mitigating risks and fostering innovation. ## Conclusion: Empowering Strategic Leadership and Innovation Through the Tracks Methodology The Tracks methodology has proven to be a powerful framework at Taboola, empowering managers with the flexibility to lead cross-functional teams, allocate resources efficiently, and drive impactful innovation. By offering a structured yet adaptable approach, Tracks not only help achieve key business outcomes but also provide valuable opportunities for leadership growth across R&D and other departments. The balance of accountability, ownership, and adaptability equips both managers and teams with the tools needed to navigate complex challenges and deliver results. Through mechanisms like Discovery Mode and the principle of early failure, the methodology ensures that innovation remains a core component of how we operate, enabling us to pivot quickly and focus on high-potential initiatives. In Part 3, we will further explore the role of Track Leaders, focusing on how they manage cross-functional teams, recognize and act upon early signs of failure, and ensure that their Tracks remain aligned with Taboola’s strategic objectives. --- ### Driving Success in Engineering at Taboola: Cross-Functional Management Blueprint URL: https://www.taboola.com/engineering/taboola-tracks-maximize-conversions/ Last Modified: 2025-02-19 17:24:23 In today’s fast-paced business environment, flexibility and collaboration are key to driving innovation and delivering impactful results. At Taboola, we’ve embraced cross-functional management as a way to bring together the best talent from across the organisation to work toward common goals. Our approach, embodied in the Tracks methodology, empowers teams to take end-to-end ownership of projects and deliver measurable business impact. As we celebrate the first anniversary of Maximize Conversions—one of the successful initiatives born from this methodology—it’s a great time to reflect on how Tracks have transformed our approach to project management, innovation, and growth. This journey has been further validated by being named a finalist in the 2024 Digiday Technology Awards for Best Native Advertising Platform, largely due to successful collaborations like the Hyundai case study. Beyond Maximize Conversions, other impactful innovations, such as Generative AI from the Creative & Formats Track, have continued to drive engagement and deliver superior business outcomes. ## Part 1: Introducing the Tracks Methodology: Building Cross-Functional Teams for End-to-End Success In this first part, we’ll explore the fundamentals of the Tracks methodology. We’ll look at how it fosters collaboration across departments, drives business impact, and empowers teams to take ownership of their work, creating a culture of accountability and innovation. ### Tracks: Cross-Functional teams that Drive Impact At Taboola, Tracks are medium-term cross-functional teams that typically last several months to a few quarters, designed to drive impactful outcomes within a defined time frame. Tracks focus on developing demanding technical abilities through structured roadmaps, enabling teams to achieve clear business goals. - Purposeful Collaboration: Tracks bring together cross-functional teams from various departments—R&D, Product, Marketing, and more—under a shared mission. This collaboration enables us to break down silos and foster creative problem-solving. - Mission-Oriented with Clear KPIs: Each Track is guided by a roadmap with defined milestones and KPIs, ensuring that every project stays aligned with the company’s strategic objectives. - Leadership and Accountability: A Track is typically led by an R&D Manager or Product Manager, ensuring that leadership is aligned with both technical and business needs. Teams are given clear direction, while leaders are accountable for driving progress. ### Track Leaders’ Responsibility Each Track leader—usually an R&D Manager or a Product Manager—is tasked with guiding their team toward achieving the Track’s goals. The Track leader has several critical responsibilities: - Defining the Roadmap: The Track leader must define a clear roadmap, outlining the milestones that need to be reached to meet the project’s KPIs and deliver meaningful business impact. - Driving Execution: Track leaders are responsible for the day-to-day execution of the project. They coordinate cross-functional teams, ensure smooth collaboration, and manage any challenges that arise during the project’s lifecycle. - Accountability for Results: Ultimately, the Track leader is accountable for its success. Their role is to ensure that the project stays on course, meets its goals, and delivers measurable results in line with Taboola’s business objectives. ### Enhanced Flexibility in the Tracks Model One of the greatest strengths of the Tracks methodology at Taboola is its flexibility in team assignments and leadership, which allows us to adapt to business priorities more effectively compared to the classic organic structure. - Dynamic Team Assignments: In the Tracks model, teams aren’t confined to their original departments. Instead, resources (people) can be reassigned based on the specific needs of the project. This flexibility ensures that the best talent is placed on the most impactful projects, regardless of their department. In contrast, the classic organic structure typically limits team members to their designated functions, which can restrict their ability to contribute to high-priority initiatives outside their core teams. - Leadership Across Tracks: In the organic structure, leaders manage their own teams and projects within a specific department. However, the Tracks methodology allows for greater leadership flexibility. A group manager may lead a Track with teams from other departments, while some of their own teams may be working under a different Track leader. This cross-functional leadership enables us to assign the best-suited leaders for each project, promoting both technical excellence and strategic alignment. - Flexibility in Response to Changing Priorities: As business priorities evolve, the Tracks model allows for quick reallocation of resources. If a new business opportunity arises or a project’s goals shift, the Tracks structure enables us to pivot more easily by adjusting team assignments and resources. In contrast, the organic structure can be slower to adapt, as formal restructuring processes may be required to shift resources between departments or projects. - Cross-Functional Collaboration as a Standard: The Tracks methodology inherently encourages cross-functional collaboration by breaking down departmental silos. In the organic structure, collaboration across departments often requires more effort and coordination. With Tracks, cross-functional collaboration is built into the process, allowing teams from different areas of the organization to work together more seamlessly. Diagram: Illustrating the Tracks structure, including the Tracks Management Forum, Track Leaders, and cross-functional collaboration between R&D, Algorithms, Product, and Marketing teams. Note: While the diagram separates Track Leaders and R&D Managers, in practice, certain R&D Managers may also act as Track Leaders for specific Tracks. As was described above in the "Leadership Across Tracks" section, this flexibility enables us to assign the best-suited leaders for each project, promoting both technical excellence and strategic alignment. ### Tracks as a Catalyst for Innovation The Tracks methodology fosters a culture of innovation at Taboola by enabling: - Cross-functional collaboration that brings diverse perspectives together to solve complex problems. - Focused, mission-driven innovation where creative problem-solving is aligned with strategic business goals. - Flexibility for iterative improvement, allowing teams to experiment, take risks, and refine their ideas. One of the most powerful tools for fostering this creativity is Discovery Mode, which provides a structured way to explore uncertain opportunities and test new concepts (see more about innovation in Part 2 under the section on Discovery Mode). ### Benefits of the Tracks Methodology for R&D Developers and Talents The Tracks methodology isn’t just about delivering business results—it also empowers R&D developers and talents to thrive. By providing opportunities for cross-functional collaboration, high-impact projects, and personal development, Tracks create an environment where innovation and career growth flourish. - Exposure to New Technologies: Working on Tracks gives developers the chance to work with cutting-edge technologies and tackle complex challenges, keeping their skills sharp and relevant. - Ownership and Autonomy: Tracks give developers the end-to-end responsibility, allowing them to take the initiative, make decisions, and see their work through to completion. This ownership boosts creativity and engagement. - Career Growth and Visibility: Developers who contribute to successful Tracks gain visibility within the organization, accelerating their career growth and opening up opportunities for leadership roles. ### Not Every R&D/Product Activity Should Be Part of a Track Tracks are designed to drive large-scale initiatives aligned with leading company goals, with achievements expected within a mid-term timeframe. They focus on high-impact outcomes with clear roadmaps, milestones, and well-defined KPIs. Additionally, Tracks emphasize cross-functional collaboration, bringing together teams from different departments, such as R&D, Product, and Marketing. However, not all R&D activities fall within this scope. Some developments are critical but not directly tied to such large initiatives, including ongoing improvements and maintenance to existing products. After a Track achieves its purpose, responsibility may revert to the regular organic teams for continued maintenance and evolution. ### Maximize Conversions: A Track Success Story One of the early success stories that emerged from Taboola’s Tracks methodology is Maximize Conversions, which was launched just one year ago. This AI-powered bidding strategy was designed to transform how advertisers manage their campaigns on Taboola, simplifying the process by allowing advertisers to focus on their business goals while the system optimised performance behind the scenes. The Challenge: Before Maximize Conversions, advertisers had to manually adjust their campaign bidding based on their goals, which was time-consuming and often inefficient. The Taboola team recognized the need to develop an automated system that could improve advertiser results by leveraging AI to optimise bids in real-time. The Approach: A cross-functional Track was created to address this challenge, pulling together talent from R&D (Software engineers, Algorithms engineers), Product, Analysts and Product Marketing. The team worked collaboratively in high focus to design, test, and refine the AI-based bidding strategy. This involved multiple iterations, each informed by data and feedback from advertisers during early trials. The goal was to not only increase the amount of conversions and improve cost, but also simplify campaign management for advertisers, enabling them to achieve better results with less manual effort. The Result: Since its launch, Maximize Conversions has delivered significant results: - Over 1,000 advertisers have joined the platform, including major brands like Babbel and Hyundai. - Campaigns using Maximize Conversions have seen an average 110% increase in conversions. - Advertisers have launched double the number of campaigns each quarter, showcasing the effectiveness of the tool. In the Hyundai case study, Maximize Conversions played a crucial role in reducing costs: - A 30% reduction in cost per session. - A 26% reduction in cost per car configuration. Additionally, innovations like Motion Ads, developed in the Creatives & Formats Track, also played an important role in driving engagement for Hyundai and other brands. Together, these innovations contributed to Taboola being named a finalist for Best Native Advertising Platform at the 2024 Digiday Technology Awards. Images: - Celebrating the first anniversary of Maximize Conversions with 1,000+ advertisers and a 110% conversion increase. - Hyundai's partnership with Taboola achieved a 30% lower cost per session using Maximize Conversions. Shaping Taboola’s Future Approach: The success of Maximize Conversions, along with other early Tracks, was pivotal in shaping Taboola’s broader approach to cross-functional management and innovation. The collaboration between teams from various departments and the rapid iteration cycles exemplified how the Tracks methodology can drive business impact efficiently. As we celebrate the first anniversary of Maximize Conversions, it’s clear that this Track demonstrated the value of empowering teams to innovate and deliver results aligned with Taboola’s strategic goals. ## Conclusion: Empowering Innovation and Impact through the Tracks Methodology The Tracks methodology has had a transformative impact on how we operate at Taboola. By breaking down silos, fostering collaboration, and providing the flexibility to focus on high-impact projects, Tracks have empowered both our organisation and our people to innovate and grow. In the next part, we’ll explore the high-level management of Tracks, the roles of group and track managers, and how innovation is driven through Discovery mode. --- ### Who Eats Memcached Network URL: https://www.taboola.com/engineering/who-eats-memcached-network/ Last Modified: 2025-01-14 14:37:12 Taboola is the leading recommendation engine in the open web market, serving close to 1 million requests per second during peak while keeping p99 response time subsecond. During recommendation flow, caches are heavily utilized to boost performance, save complicated computation, and reduce load to external data stores. In Taboola, we build cache services based on memcached and have a memcached cluster per data center dedicated to recommendation flow. These memcached clusters serve over 10 million requests per second. In this article, we discuss one particular case where our memcached cluster network got saturated and became unstable. ## The problem As Taboola continues to grow (Yahoo partnership, Apple News and Stocks, and many more), so are our recommendation requests. In turn, our memcached cluster powering recommendation requests starts to face a skewed access pattern as illustrated by the graph below. Each line represents a memcached node - positive for transmit traffic and negative for receive traffic. In particular, the orange line is a skewed node with much higher network traffic than others. And network traffic of the skewed node is saturated to a point that clients connecting to the node start to encounter occasional, unpredictable packet loss. ## Observability into cache distribution Before we dive into the issue, we need to understand how cache in Taboola works. As a simplified flow, clients see read-through cache while there are two layers of cache underlying - in-memory cache and memcached. - Access to caches reaches in-memory cache first. - If in-memory cache misses, then reach memcached. - If both in-memory cache and memcached miss, fetch from data stores and/or perform whatever computation necessary. In a memcached cluster, each memcached node works independently and is unaware of each other. Cache entry sharding is performed by our clients using consistent hashing. New nodes are added occasionally and in general the number of nodes is relatively stable. The first theory to the issue is the celebrity problem, i.e., certain entries are hot and accessed much more frequently than others, and thus leads to the skewness. But this is quickly eliminated, since hot entries likely reside in-memory cache all the time and are served in-memory without needing memcached fetch. We also double confirm that no suspicious high evictions among in-memory caches. Our theory is then turned to skewed cache entry distribution among memcached nodes. Initial check on memory indicates that there are some discrepancies among memory usage of memcached nodes, though much less extreme than network traffic. However, given our memcached cluster serves tens of different caches, skewed entry distribution of one cache or few caches are likely hidden from coarse-grained memory view. To find the potential skewed entry distribution, we need to be able to analyze per cache entry distribution. While memcached has the tool to dump cache statistics, it’s an aggregated view and has no knowledge of determining which entry belongs to which cache. In order to better understand the entry distribution, first we add a new convention to our cache framework and gradually migrate all of our caches to prefix cache keys with cache names. Second, we utilize memcached lru_crawler to dump cache keys and together with cache name prefix to collect entry distribution per cache. We make this process automatic and feed the statistics into our continuous monitoring pipeline backed by Prometheus. After the statistics are available, unfortunately none of the tens of caches show skewed entry distribution. On the good side, we ruled out another theory and built another nice observability into our memcached cluster. ## Locate the villain Then something catches our attention - the skewed memcached node shifts from one node to another at a certain time point, as illustrated by the graph below, while it has been consistent on a particular node for a long while already. Looking deeper we realize that the shift timing correlates to the migration of adding cache name prefix to cache keys. While the root cause is still unknown, we now have an easy way to a) validate whether it’s one or few caches causing the issue, and b) find exactly which caches are troublemakers. The investigation is then straightforward with binary search technique. We divide tens of caches into two groups and perform another cache key tweaking migration on those two groups separately and observe migration of which group leads to skewed node shifting. And from the group having shifting effect, we repeat the process of dividing it further into two subgroups to test shifting effect. After a few interactions, we are able to locate exactly one cache who is causing the skewness. So now we find the villain, but the puzzle is not yet solved. The problematic cache’s in-memory cache hit rate is ~83%, while not great but also not that bad as well, and its eviction rate is reasonable. Why is it causing the issue then? ## The root cause There are several attempts to address the issue, including reducing the size of largest entries being up to hundreds of kilobytes and some experiments to reduce cache access, but none helps and the skewness persists. Then another anomaly is noticed - while the problematic cache has a reasonable in-memory hit rate, the stale access ratio is >90%. Stale access is an optional feature in Taboola to help latency. The basic idea behind stale access is simple - in some scenarios where there is contention between latency and data freshness, we can optionally favor latency by postponing data fetch and using slightly stale data. When stale access is enabled, which is the case with the problematic cache, the flow becomes: - In-memory cache now has two TTL values, as illustrated in the graph below. Freshness TTL to determine whether an entry is considered fresh or stale. - Entry TTL to determine when an entry needs to be evicted from the cache. - The invariant is freshness TTL < entry TTL. - When access to an in-memory cache returns an entry whose freshness exceeds freshness TTL but still within entry TTL, it’s marked a stale access and then the entry is served as usual. - Meanwhile, a background task is launched to fetch data from memcached and data sources if memcached misses, and then update the fresh copy back to in-memory cache. In general we expect stale access ratio to be low, since subsequent access to the same entry of a stale access likely gets fresh data resulting from background update. And high stale access ratio hints that something unexpected is happening. After some digging, it turns out that indeed this is where the bug is! By convention we assign the same value to both freshness TTL and memcached TTL. But for the problematic cache, there is a bug slipping through leading to memcached TTL being much greater than freshness TTL. This results in the background task of updating stale data from memcached succeeds but it’s updating a stale entry with another stale entry, when entries in memcached exceed freshness TTL. Practically it means stale entries are not refreshed at all until passing memcached TTL. And it also explains the symptom that in-memory cache hit rate is high as stale access is still a hit. In short, stale entries are not refreshed, resulting in massive stale access, in turn resulting in massive memcached access. This essentially turns the issue into the celebrity problem which we eliminate in the very first beginning, i.e., hot entries are not really buffered in-memory and majority of access results in memcached access. With the root cause found, the fix is quite trivial. After the fix is applied, the network traffic of the skewed node is reduced to one fifth and the skewness is resolved, as illustrated in the graph below. ## Conclusion This exploration shows a simple issue becomes challenging when it’s live on production with millions of access per seconds and tens of different clients involved. We hope you enjoy reading this problem-solving journal with us and find values in techniques we used to solve this particular problem. --- ### Optimizing Ad Space Through Mixing Organic Items In The Open Web URL: https://www.taboola.com/engineering/optimizing-ad-space-through-mixing-organic-items-in-the-open-web/ Last Modified: 2025-03-11 13:13:10 ## Introduction Taboola is the world's leading native advertising platform, serving thousands of websites with a diverse array of products, including the Taboola Feed, mid-article ads, and homepage personalization, generating hundreds of millions of impressions daily. Despite our relentless efforts to optimize performance through sophisticated features, advanced models, and robust infrastructure, navigating such intricate ecosystems inevitably unveils areas of sub-optimization. While some widgets exclusively feature ads, most exhibit a blend of organic items and advertisements (some also contain eCommerce and paywall content), as mixing organic content helps greatly in avoiding ad blindness and generating page views. Taboola currently uses static slot allocation for organic content and ads, which fails to adapt to user preferences and inventory opportunities. In this blog post, I'll delve into our solution designed to address this challenge, outlining the hurdles encountered and the adjustments made to overcome them. ## Current Setup At the core of our strategy lie a nuanced understanding of our publishers, where we deliver two main values: - Direct monetization through ad clicks. - Increased page views (PVs) fueled by organic clicks. Direct revenue through ad clicks is straightforward to evaluate for both Taboola and the publishers, while organic clicks generate substantial revenue for the publisher through alternative monetization methods. However, this does not translate similarly for Taboola. The balance between these two values (Revenue/PVs) is a unique attribute tailored for each publisher. Ideally, it should correlate with the ratio between the revenue accrued from Taboola and other revenue streams. For instance, if a publisher derives much greater value per PV from other monetization forms, such as subscriptions, than from a Taboola ad click, the inclination would be to showcase more organic content and minimize ad slots to get more subscribers. Thus, organic content would be prioritized over ads in Taboola’s widgets. Figure 1 - Current Schema For sponsored items, we have two types of models: one for predicting the eCTR (estimated click-through rate) and another for predicting the eCVR (estimated conversion rate) and setting the ad’s CPC (cost per click). We calculate eCTR * CPC = eCPM for each item and rank them accordingly. For organic items, we use only one model to predict the eCTR and use its output to rank them. ## Unified List Solution for Ad Space Mixing Two main principles guide the solution: - PVs derived from clicks on Taboola’s organic recommendations generate value for the publisher and thus for Taboola too, so we should be able to put a price on that value. - In order to find the optimal allocation for our ads and organic items, we must rank them together. Putting these two principles together, we came up with the idea of using an eCPC for organic items (organic eCPC) and using this organic eCPC for calculating the organic eCPM (estimated cost-per-mille), allowing us to rank both organic items and ads at once. But how can we determine this organic eCPC? ## Implementation Our solution includes three steps: - Determining the organic eCPC for each publisher. - Estimating eCPM for both organic items and ads. - Ranking the items accordingly. ### Organic eCPC The main challenge revolves around determining the right organic eCPC to meet our specified goals, whether it's achieving targeted revenue per day or maintaining a specific ratio between revenue and engagement. As a preliminary step, we decided to simulate the ranking we would have achieved using a single organic eCPC for all the organic items on one of our international websites. Upon analyzing the results, we found that even though the share of organic and sponsored items seemed reasonable, cohorts of users were receiving only organic items. This wasn't initially alarming, but we thought it wise to check if these users had something in common. To our surprise, they did—the vast majority of these users were from third-world countries. This, of course, makes sense, as most campaigns set up in Taboola had location-based targeting. Users from different countries had different inventories of ads to choose from, and the items available for the third-world users had mostly low CPCs, so our organic items were beating them almost every time. To solve this problem, we simply calculated the average eCPM for each country and normalized our eCPC accordingly, allowing us to set the eCPC once for each country. ### Determining the Initial Organic eCPC As mentioned above, to gauge the potential impact of our solution, I conducted an analysis using actual requests from different placements in several major publishers. We checked what the DCG (discounted cumulative gain) would be for both revenue and organic clicks for different shadow bids, ending up with a Pareto frontier depicting the ratio of revenue to page views generated. This surpassed our current static state by hundreds of percent for both organic clicks and revenue. But how can we find the optimal organic eCPC for this website? Unfortunately, there is no definitive answer (unless you know what the average RPM is from all income sources for the website after a click on an organic item). In most cases, websites choose to withhold this information from Taboola. We had to use arbitrary proxy goals, whether it was aiming for the current revenue/PVs ratio, maintaining the same number of page views generated as in the static setting while maximizing revenue, etc. Regardless of the goal, we used the eCPC that got us this value in the Pareto analysis as an estimate to build upon. Figure 2 - Offline Analysis Using NDCG ### Controller The open web is an ever-changing environment, and one aspect of it is that the average CPM isn’t static either. This means we must continuously update our organic eCPC for each publisher. To do so, we implemented a simple “controller,” updating the eCPC once a day based on a single week’s data, adjusting it according to the optimization goal. For example, if we aim for a specific number of clicks per user and are exceeding it by 5%, our calculation would be (1 - 0.05*learning_rate) * current organic eCPC. In most cases, using the initial organic eCPC from the offline analysis, it takes a week or two to match our optimization goal. ### Calibration Challenges Another significant challenge we encountered was calibration. Ensuring the calibration of both our organic and ad models is paramount. Each placement we support must have accurately calibrated predictions. For instance, if our models predict a 10% chance of an item being clicked, this prediction should align closely with real-world outcomes for both organic and sponsored items. If it doesn’t, we will end up with suboptimal rankings. To address this challenge, we employed two key strategies. Firstly, we revamped the way we stratify the data for our model, transitioning from per-publisher stratification to per-widget stratification, ensuring our model can make more accurate corrections. Secondly, we implemented isotonic regression techniques to ensure alignment between predicted probabilities and observed outcomes for all our models. Figure 3 - Suggested Schema ## Online Results ### Measuring Impact Online Traditional metrics like RPM and CTR are inadequate in our case, as they pit engagement against revenue. Instead, we focus on clicks and revenue per user. Moreover, filtering out super users and bots eliminates noise from the data. We evaluate the effect on both affected widgets and entire page views to gain comprehensive insights. Initial findings indicate promising lifts of 8.7% in RC clicks per user and 10.5% in revenue per user. Additionally, there's a discernible decrease in ad density, reducing the number of sponsored items by 18%, signaling a more balanced and optimized ad space. ## Conclusion In conclusion, our unified approach not only enhances revenue potential but also fosters a more tailored and engaging user experience, reinforcing Taboola's position as a leader in native advertising optimization. ### Key Takeaways - The unified ranking approach performs very well; it optimizes ad space allocation successfully, yielding significant lifts in both engagement and revenue. - Calibration is crucial; we need to account for the numerous different widgets on different sites with different layouts, as well as users from different countries with different average CPCs and ad inventories. - Choosing the right metrics is essential: Focusing on metrics like RC clicks per user and revenue per user allows for accurate evaluations. ### Work to be Done - Address weekly seasonality in organic eCPC: Users behave differently on weekends, which changes the average CPM. - Implement the LinkedIn Gap Effect . - Analyze long-term effects. - Apply this approach to different types of content, such as eCommerce and paywall pieces. ## Supplementary Information Offline Analysis Simply put, in the offline analysis, I used our data records to simulate what the ranking would have been given the deeper and smarter bid predictions, and summed up the estimated amounts of revenue and organic clicks (eCTR * oCPC for ranking). There are two caveats that might differentiate the analysis from reality: - I used the DCG method of dividing the estimate by log(1+slot index); this is not accurate, and it would have been best to use an accurate calibration factor for each slot. - Some items that should have been blocked might have been included. I removed the items that were already marked as blocked, but this might not be enough. I'll add that before running the analysis, I checked the calibration for both organic and sponsored CTR predictions. They both seemed very accurate (post-multiplication), with about a 3% error on average, allowing us to trust the cumulative metrics. ## References Guo, C., Pleiss, G., Sun, Y., & Weinberger, K. Q. (2017, July). On calibration of modern neural networks. In International Conference on Machine Learning (pp. 1321-1330). PMLR. Yan, J., Xu, Z., Tiwana, B., & Chatterjee, S. (2020, August). Ads allocation in feed via constrained optimization. In Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (pp. 3386-3394). --- ### Optimizing TensorFlow Serving Performance on Skylake CPUs URL: https://www.taboola.com/engineering/optimizing-tensorflow-serving-performance-on-skylake-cpus/ Last Modified: 2025-01-14 14:37:13 ## Introduction Taboola is the leading recommendation engine in the $80+ billion open web market. The company’s platform, powered by deep learning and artificial intelligence, reaches almost 600 million daily active users. The company’s ML inference infrastructure consisted of 7 large-scale Kubernetes clusters across ten on-premise data centers and tens of thousands of cores. Over recent months, we've witnessed an increase in the Kubernetes worker nodes load averages that initially seemed aligned with rising traffic levels. However, further investigation revealed significant discrepancies in servers load-to-request rate ratio across data centers. Closer examination pointed to a notable correlation with the proportion of Skylake CPUs in each cluster, particularly in Chicago and Amsterdam, where Skylake usage was significantly higher (46% and 75%, respectively) compared to other locations (20-30%). ## The Skylake CPU Frequency Challenge It's been observed and documented across the web that Skylake CPUs experience frequency downclocking when utilizing the AVX512 instruction set, with the impact intensifying as more cores employ it. While we previously adopted custom-compiled binaries with optimized mtune/march flags to disable or enable AVX512 on specific CPU models in TensorFlow 1.x, migrating to TensorFlow 2.x initially seemed to rectify the issue with performance improvements out-of-the-box. ## Unveiling the Root Cause: oneDNN's JIT and AVX512 Finding the root cause in an environment that includes heterogeneous CPU families, constant models, and traffic level changes can be challenging so our troubleshooting journey began with the basics like verifying the servers’ governor is set to performance, there was no change in servers tuned profile and examining environmental factors like servers room temperature to rule out external influences. Taking a closer look into performance metrics showed that all the servers in the datacenter had load average increase. (We use a “least requests” load balancing algorithm to compensate for the heterogenetic CPU types.) It was interesting to observe that there was more of an impact on Skylake based servers. This impact included frequency reduction, and degradation in inference requests per CPU core metric. Next we shifted our focus to investigating AVX512 usage on the Skylake servers. Initial probes using the arch_status under the /proc filesystem suggested minimal AVX512 involvement: grep -v "\-1" /proc/*/arch_status   But as the documentation noted, there is an option for false negatives. Running more comprehensive analysis using perf uncovered a contrasting scenario: perf stat -e cpu/event=0x28,umask=0x18,name=core_power_lvl1_turbo_license/,cpu/event=0x28,umask=0x20,name=core_power_lvl2_turbo_license/,cpu/event=0x28,umask=0x40,name=core_power_throttle/   The perf output illuminated substantial level 2 core throttling attributed to AVX512, examining the ratio between level2 to level1 throttling showed 4 of 5 ratio. In parallel, our Tensorflow workload where in the process of upgrading from version 2.7 to 2.13 analyzing a few Tensorflow 2.7 workloads using perf showed much lower AVX512 usage with a ratio of 1 of 3. Note that in our environment, the Tensorflow process is not pinned to specific cores, so the results have a lot of "noise" from cross-process CPU time on the same physical core. However, TensorFlow serving code didn't explicitly enable AVX512 compilation, with only AVX and SSE4.2 flags present in the .bazelrc file. The key discovery came from TensorFlow 2.9 release notes: oneDNN was now enabled by default. This library dynamically detects supported CPU instruction sets like AVX512 and employs Just-in-Time (JIT) compilation to generate optimized code. ## Crafting a Workaround: Limiting oneDNN's Instruction Set Following oneDNN documentation, we implemented a workaround to constrain the instruction set based on the CPU family. Added to the image entrypoint bash script a simple test to detect CPU type and set the following environment variable to restrict oneDNN to AVX2: ONEDNN_MAX_CPU_ISA=AVX2 ## ## Outcome: Performance Uplift and Reduced Load Deploying this fix resulted in a notable improvement: - ~11% increase in average CPU frequency. - ~11% decrease in peak 15-minute load average. - ~10-15% boost in inference requests per used CPU core. Key takeaways and considerations: - Prioritize CPU compatibility: When deploying deep learning workloads across diverse CPU architectures, carefully evaluate their instruction set support and potential performance implications. - Stay informed about framework updates: Regularly review release notes and changelogs of frameworks like TensorFlow to understand introduced features and their potential impact, especially on specific hardware configurations. - Leverage performance analysis tools: Employ tools like perf to gain fine-grained insights into CPU behavior and pinpoint performance bottlenecks. ## Let's Be Greedy: Compiling Tensorflow with -march Tuning TODO - Add the tensorflow compile with icelake flag testing…. After the successful instruction set tuning for Skylake, we wanted to check if we can squeeze a few more drops of performance. We compiled a few Tensorflow serving binaries with different march options by adding the following to .bazelrc and pass the relevant build parameter at build time:   # Tune for Icelake and newer CPU Famely build:icelake --copt=-march=icelake-server build:icelake --host_copt=-march=native build:icelake --copt=-O3 # Tune for SkyLake (Not include AVX512) build:skylake --copt=-march=skylake-server build:skylake --host_copt=-march=native build:skylake --copt=-O3 ... # More CPU famalies ...   After pushing the images to our local repository we created a custom image and COPY the different binaries. We set the ENTRYPOINT to launch a bash script that will execute the binary based on the CPU Family. After deploying the new image to production and monitoring its performance we didn't observe any performance boost, seems the heavy lifting is done by the OneDNN library ## Moving Forward: Continuous Monitoring and Tuning Our experience highlights the importance of proactive performance monitoring and optimization, especially in heterogeneous infrastructure. We'll continue to closely monitor cluster performance, explore deeper optimizations where applicable, and stay up-to-date with advancements in TensorFlow and related libraries. We are already excited to test the impact of AMX mixed precision on our servers running Intel Sapphire Rapids CPU family. --- ### Android Working On A Multiple Library Project URL: https://www.taboola.com/engineering/android-working-on-a-multiple-library-project/ Last Modified: 2025-01-14 14:37:13 We are going to talk about a multi-library project in Android. Not something ordinary, but not something out of the ordinary either. You may have come across it in your line of work or you may be looking into converting your library into sub-modules for better structure and organization. No matter the case, you should be well aware of what lies in front of you before diving in. Writing your library in Android is neat. You get a chance to write some code that can help other developers (or even yourself). Since libraries can’t be a standalone project by themselves, they are usually always paired in a project with an application. This allows developing the library to be a simple process where you add a feature/fix a bug and then you can test it directly with the application you have in the project. Thus, simulating (in a local way) how a developer will integrate your library. But, what if your library relies on another library you are developing? If you are not aware of it, you should know that a library (read AAR) cannot contain another local library within it. It can rely on libraries remotely (via dependencies), but not on something local. This is not supported in Android and while some solutions popped up over the years (FatAar), these didn’t always solve the problem and are not up to date. There is even a Google Issue Tracker requesting this feature that has been open for quite some time and is receiving plenty of attention from the community. But let’s identify which walls we can break and which we cannot. Imagine your project hierarchy looks like this: So, since InnerLib can’t be part of your original project, where can it reside? And also how would you be able to work locally while developing features inside InnerLib? We are going to answer these questions in this article. ## Git Submodule For most technical problems, there isn’t always just one solution. Usually, there are more, but each solution has its drawbacks. It is all a question of which drawbacks you are more comfortable living with at the end of the day. To answer our first question, where can InnerLib reside, we have several options: - Make InnerLib a submodule of our original project - Make InnerLib a remote dependency of its own If you are not aware of submodules in Git, Git’s documentation is a good place to familiarize yourself with them. Quoting from it (the first paragraph): “It often happens that while working on one project, you need to use another project from within it. 👉 Perhaps it’s a library that a third party developed or that you’re developing separately and using in multiple parent projects. 👈 A common issue arises in these scenarios: you want to be able to treat the two projects as separate yet still be able to use one from within the other.” shows us that this is exactly our use case. Using a submodule has its benefits. All your code is in one place, easy to manage and easy to develop locally. But submodules have some weaknesses. One, is the fact that you must always be aware of which branch your submodule is pointing to. Imagine a scenario where you are on a release branch in your main repository and your sub-module is on a feature branch. If you don’t notice, you release a version of your code with something that is not ready for production. Whoops. Now think about this within a team of developers. One careless mistake can be costly. If the first option seems problematic to you, then hosting our library in another repository is your second choice. Setting up the repository is pretty simple, but how do you work locally now? ## Working Locally Now that we got our project set up properly, we will probably have a line similar to this in our OuterLib build.gradle file: How can we make the development cycle efficient and easy to work with? If we develop some features in InnerLib, how do we test things out in OuterLib? Or in our application? One solution that might come up is to import our InnerLib locally to our OuterLib project, while having InnerLib .gitignored in our OuterLib project. You can do so easily by right clicking on the name of the project in the left hand side menu in Android Studio and going to New → Module. Then in the window that opens up, you can choose the Import option at the bottom left That sounds easy and simple so far, but what’s the catch? Each time you modify a file that belongs to InnerLib, the changes won’t be reflected inside InnerLib since it is ignored. So, each change you want to make has to happen inside of InnerLib and then you have to import it again inside OuterLib to see the changes. This doesn’t seem right. There must be a better way of doing this. With just a few lines in our settings.gradle file, we can make sure our files stay in sync when we make changes in InnerLib. When we imported InnerLib into our project, Android Studio made a copy of InnerLib and cached it. That is why we needed to re-import the library for every change we made inside of it. We can tell Android Studio where to reference the files from using the projectDir attribute. Our settings.gradle might look something like this: To reference our InnerLib locally, we would have to change settings.gradle into this: Using this approach, our InnerLib files will be linked to our working directory and thus every change we make will be reflected immediately. But, we would like flexibility when working locally on OuterLib with a remote version of InnerLib. What we wrote above inside the settings.gradle file will only allow us to work locally and surely we don’t want to commit that as it is. ## Maven Local If the approach above doesn’t sit quite right with you, there is a different one you can take. Just like you would publish your library publicly with Maven, you can do the same thing locally with Maven local. Maven local is a set of repositories that sit locally on your machine. Below are the paths for mavenLocal depending on the operating system of your machine: - Mac → /Users/YOUR_USERNAME/.m2 - Linux → /home/YOUR_USERNAME/.m2 - Windows → C:\Users\YOUR_USERNAME\.m2 In essence, you can publish your library locally and then link to it in your project. Doing it this way, we can link our project to InnerLib. In order to allow this configuration in our project, we need to do the following things: - Add mavenLocal() as a repository inside our repositories clause. This is to allow our project the ability to search for repositories locally 2. Change our implementation line inside our dependencies clause to reference our InnerLib as if it we are referencing it remotely 3. To publish InnerLib locally, we will create a file called publishingLocally.gradle that will contain the following: 4. Inside your application level build.gradle file, add the line: apply from: '/.publishingLocally.gradle If this option seems a bit too good to be true, it is. While on one hand, we can develop things locally seamlessly just as if we were working with a remote library. On the other hand, if we make any change inside InnerLib while working locally, it is required to publish it locally again. While this isn’t a costly task, it does create a need to perform tedious tasks over and over. ## A Solution For Working Locally & Remotely We want to avoid the constant need to re-publish our InnerLib package whenever we make a change locally. We need to figure out a way to make our project be aware of those changes. In the Working Locally section, we found out how to do that, but we had an issue with committing the settings.gradle file. To solve this problem so we can work both locally and remotely with our InnerLib, we will use a parameter we will define in our gradle.properties file. The gradle.properties file is a place where you can store project level settings that configure your development environment. This helps make sure that all the developers on a team have a consistent development environment. Some settings you might be familiar with that are found inside this file are AndroidX support (android.useAndroidX=true) or the JVM arguments (org.gradle.jvmargs=-Xmx1536m). To help us solve our situation, we can add a parameter here to indicate whether we want to work locally or not. Something along the lines of: workingLocally = false This parameter will grant us the ability to distinguish between which settings we are working with, either locally or with production code. First, let’s alter what we have in our settings.gradle file by wrapping it in a condition that checks if our parameter is true: This way, we indicate to the project to get the files for our InnerLib locally from our machine. Another place where we need to change our logic is in our build.gradle file. Here, instead of getting the code to our library remotely in our dependencies block, we can indicate whether we are depending on it locally or not. ⚠️ Word of warning: You should never commit the gradle.properties file when working locally. The journey was long and to some, very exhausting. But now we have a full-proof setup for working locally and remotely on a multiple library project. If you encounter any issues or would like to give your take on this, feel free to leave a comment. --- ### Optimizing CTR Prediction through Clustering-Based Feature Engineering URL: https://www.taboola.com/engineering/optimizing-ctr-prediction-through-clustering-based-feature-engineering/ Last Modified: 2025-01-14 14:37:13 With more than 500,000 recommendations per second, Taboola is the world’s leading native advertising platform in the world. When it comes to recommending content, choosing the right features can make a huge impact. Taboola is working with many publishers, e.g., news, shopping, sports websites, etc. The most basic feature to represent a publisher could be its identifier. Using basic features like publisher-id doesn’t tell us much about what are the user‘s preferences. Furthermore, the sparsity of a feature such as publisher-id requires a lot of data to achieve good results. Lastly, identifier based features are prone to the cold-start phenomena, in which newly added publishers do not appear in the dataset, and are not familiar to the model yet, and more importantly, they don’t have enough traffic yet for the model to extract other meaningful features. A better approach is to use features that characterize publishers by their articles (i.e., news items) and the users that visit them. Helping the model to better understand the publisher and select a more personalized content effectively, and enhancing user engagement and satisfaction, while improving our models’ performance and generalization. In this blog I will describe the process of data analysis for extraction of meaningful publisher-related features for better CTR prediction. ### Getting Cozy with Your Data This journey begins with a main dataset containing information about items recommended by Taboola with information about user and publisher, over the first week of 2024. Table 1: Example of our main dataset from the 1st day of 2024. The current dataset contains information such as: - Placement: The name of the section where the recommended item appeared within the page. - Session Referrer: The domain of the website that has referred the user to the publisher in the current session. - Organic Clicks: Amount of user historical organic clicks (i.e., articles clicks). - Sponsored Clicks: Amount of user historical sponsored clicks (i.e., ads clicks). - Pvs: Amount of page views of the current itemId. - Session Depth- number of pages the user passed through until he/she got to the current page view. - Clicks: 0/1 for user click/no-click on the item. - Source/ Target Upper/Alchemy Taxonomy - The category the article belongs to (i.e., Sports). Source is the current article, Target is the recommended article by taboola. In addition, I used another complementary dataset for "publisher_common_features". This dataset provides information about each publisher's top five common user geolocations, such as country, city, region, and DMA code (for US publishers), and the five most common categories / taxonomies read by the users in that publisher (i.e., sports, entertainment, etc.). Table 2: Example of our complementary data set. ### Sampling Snacks from the Dataset Buffet First, I filtered the main dataset to retain only samples from users that appeared in more than one of the days in the week, and had rc/sp clicks values. On top of this, I sampled only 250K samples from every week day. Once we have a data frame ready we can start handling missing values and remove outliers. ### Borrowing Data from the Neighbors' Yard As you can see in the snippet above, there are some missing values in the Country and Region columns, and a significant number of missing values in the Session Referrer column, we will handle these two differently. For the Country and Region, which are both user features, we can infer some of the values by finding the userId in another sample of the dataset and set the Country and Region from that sample to the missing values one. After applying this technique, we had a very small number of missing values, so we could drop those samples. With the Session Referrer feature, the case is different. The missing value in this column, points out that there is no information about the referral of this session, and we can simply set the value in those samples to “OTHER”. ### Good Bye Long Tail! When working with a dataset that is collected based on user sessions online, it is important to remove outliers, for these outliers are mostly bots with no intention of clicking suggested content. Sponsored Clicks Organic Clicks Before Removal After Removal # Table 3: Distribution of rcClicks and spClicks before and after the removal of outliers. In table 3, you can see examples of anomaly removals. In the distribution of rcClicks and spClicks there are long tails that can be removed. By using z score of more than 20, we got this nice and gradual log-normal distributed curve. ## Unearthing Hidden Gems in the Data Mine As mentioned before, we want to extract some meaningful features describing our publishers. We start by analyzing which properties of an item recommendation are likely to have influence on the behavior of the user (i.e., clicks). One hypothesis was that some impact can come from the following features: - is_weekend*: Binary feature for weekend sessions. - upper_taxonomy_match*: 0/1 for match between the source and target upper taxonomy**. - alchemy_taxonomy_match*: 0/1 for the match between the source and target alchemy taxonomy. - in_market: Calculated as 0/1 for a match between the user’s location such as: country, region, city and DMA code - is_sp_lover/is_rc_lover: if the user had more than 40% rcClicks and spClicks he was considered as sp/rc lovers (i.e., sponsored or organic clicker). - is_sp/rc_lovers_publisher: 1 for publishers with more than 50% of the users being sp/rc lovers, 0 otherwise. - is_TAXONOMY_publisher: For each TAXONOMY (i.e., article category) we added a binary feature if the TAXONOMY is one of the top 5 common upper taxonomies of the source page - common_target_taxonomy: If the traffic that was explained by the users for this taxonomy (i.e., sports) of this publisher was more that 70%, the publisher was considered as this “taxonomy” publisher. otherwise- “None”. *For our purpose these features will be used as calibration/ context features **Taxonomy is a way of classifying items into groups with common characteristics. In our case, it will be the item/ publisher’s main topics (e.g. news, sports, etc…). Upper taxonomy refers to high level categories, and alchemy refers to more detailed and specific categories. ## How Publishers Found Their Soulmates In order to get some knowledge about how meaningful our features are, we will first use visualizations of our data to see which features show separation in the publishers’ data. After performing one-hot encoding to the categorical features, and scaling the numerical features with Min-Max scaler, we used 2D and 3D PCA for visualizing the distribution of publishers. Figure 1: Distribution of publishers representation in 2 and 3 dimensions using PCA. Here we can already see some separation with accordance to the publisher taxonomy and whether it is a sponsored lovers publisher. Since the existing publishers' features do not necessarily define which publishers are similar in the context of CTR prediction, I decided to try representing the publishers using different methods that may better fit our target. ### Sharing Users Like Secret Handshakes As part of our efforts to partition our publishers effectively, we decided to represent our data as a network graph. In this graph, nodes represent publishers, and weighted edges between two nodes represent the number of shared users between the two respective publishers (see example in Figure 2). Through this approach, we aim to find patterns and insights of why certain publishers are grouped together within the network. There are some very good techniques to get a partition of nodes in a network. We used the Louvain-method for community detection. In order to avoid isolated or disconnected nodes and ensure a cohesive network structure, we removed publishers with low connectivity to the rest of the network. After analyzing the partition, it turned out that the feature that defined this clustering is the publisher language, which is an existing feature in our dataset. ### Aiming at the relevant task - CTR Prediction I used one of the known ctr prediction models, DeepFM (Guo, H., Tang, R., et al., 2017) from DeepCTR package. This model utilizes factorization machines, as well as deep neural networks to predict item click/non-click based on sparse and dense features. During this process, every categorical feature value is mapped to an embedding space of predefined size. After training the model with our data, I extracted the embeddings of our items and examined their clustering via t-SNE. With this embedded space representation of our publishers, I expected that the new publisher features will: - Cluster the items in a different manner. - Improve CTR prediction. Here is an example of how the publishers are distributed in the newly trained embedding space: Figure 4: Illustrating the distribution of publishers in the embedding space. Each point represents a publisher, with its position determined by t-SNE. The plot is color-coded to highlight different features of publishers, such as taxonomy like news or sports and whether large portion their users are sponsored content clickers. As demonstrated in the visualization provided in Figure 4, we observe a clear separation among the data points, in some features (e.g., Sports, News, SP lovers) more than others (e.g., Entertainment), indicating the effectiveness of the meaningful features incorporated into our representation. This separation is a testament to the significance of the features we have included, which capture essential aspects of the data and contribute to its distinct clustering. ### The Moment of Truth To validate my hypothesis, I’ve conducted a comparative analysis between the outcomes of the original model, trained on the original dataset, and those of a modified model trained exclusively on data associated with a specific cluster. Ensuring fairness, all models were trained and tested on an equal number of samples. Table 4: Comparison between different models that were trained on specific clusters of publishers. As we expected, the models that were trained and tested with only samples from a well defined cluster, performed significantly better than the one that was trained on the unclustered data. ### Separation with Style One valid suggestion that can be derived from the results in table 4 is to separate the training dataset into dedicated groups of publishers per common category. This can benefit the model, and will eventually lead to better tailored recommendations. ## Wrapping It Up: The Part Where We Say: "Bye and Thanks for All the Data!" In my journey to improve content recommendations, I inspected a data-driven clustering approach, in order to find meaningful publisher features in Taboola's network. I analyzed user interactions, locations, and content types to find subtle clues that could boost recommendations and keep our users engaged. We discovered interesting patterns among publishers. This opens up exciting possibilities for generating more personalized recommendations. To this end, we obtained an actual improvement in our models' performance, with notable enhancements observed in key metrics such as AUC and log-loss. These positive outcomes validate the efficacy of our approach and reinforce our confidence in the potential of data-driven clustering to improve content recommendations. ## Acknowledgments This blogpost was created by Ofri Tirosh, and mentored by Taboola mentors Gali Katz and Dorian Yitzhach as part of the Starship Internship Program together with the Software and Information System Engineering Department at Ben-Gurion University. ## References Guo, H., Tang, R., Ye, Y., Li, Z., & He, X. (2017). DeepFM: a factorization-machine based neural network for CTR prediction. arXiv preprint arXiv:1703.04247. --- ### Utilizing LLMs for Localized Recommendations URL: https://www.taboola.com/engineering/utilizing-llms-for-localized-recommendations/ Last Modified: 2025-01-14 14:37:14 Do you ever go to a local boutique brewery and buy a Carlsberg? Drive to your local farmer to buy Walmart packaged tomatoes? Of course not. You want to get a local experience, beer that is brewed from your neighboring produce. News is no different. When browsing your hometown's news site, you hope to find uniquely related articles. ## Local content por favor! Taboola collaborates with a substantial number of local news publishers through serving organic content across their site. These publishers are dedicated to serving local news. However, identifying whether a news article is about local news or not, can be challenging. Nowadays, straightforward solutions might be: a) explicit tagging by the editors of articles being local or national and b) extracting the location that the article is referring to from the article’s text with Named Entity Recognition models (NERs). The first solution requires a non-negligible effort from publishers and the second works well only when the name of the location is explicitly mentioned. Now consider the following examples: “A Dolphins football player felt sick today” or “Governor DeSantis has consistently improved his city’s attractiveness”. From an initial examination, no explicit location appears in those titles, but when looking deeper, the Dolphins are actually a football team in Miami, and Ron DeSantis is currently the Governor of Florida. Implicit locations understanding could increase the location coverage significantly. ## ChatGPT? Don’t mind if I do Detecting the implicit locations seems pretty complex. For the examples above, we could have used a Knowledge Graph that will match between Dolphins to Miami Dolphins which is a sports team in Miami. Can Large Language Models (LLMs), such as ChatGPT and its kind, which seem to have a built-in Knowledge graph, do that matching easily? In order to answer this question, we took 2K english titles of news articles, and sent them to ChatGPT, asking to extract the location entities from the title, such as city, state and country. The task was not straightforward, and we had to implement prompt engineering principles such as In-context Learning , for giving the engine a context about the task, for example: “Imagine you are an editor of a newspaper, try to extract the city, state and country from the given titles”. In addition, we brokedown the task into steps using Chain-of-thought technique, and added guidelines for specific categories that the articles are related to, such as sports, business and politics, for easy identification of the location entities (e.g., “Let’s work step-by-step. If this is a sports item, first search for the sports team name, then check its hometown, and then look for the state and country”). Finally, we asked to return the result in simple json format (see Table 1 below). a) Json result of sports item b) Json result of business/ political item {“title”: “A football player in the Dolphins felt sick today”, “title_city”:”Miami”, “title_state”:”Florida”, “title_country”:”United States”, “sports_team”:””Miami Dolphins”, “sports_hometown”:”Miami”, “sports_state”:”Florida”, “sports_country”:”United States”} {“title”: “Republican crusade to convert UNC-Chapel Hill will lead to its demise”, “title_city”:”Chapel Hill”, “title_state”:”North Carolina”, “title_country”:”United States”, “political_party”:”Republican”, “business”:”UNC-Chapel Hill”, “business_city”:”Chapel Hill”, “business_state”:”North Carolina”, “business_country”:”United States”} Table 1: Few shot example of sports item (a) and business item (b) and the LLM result as a json. The json contains dedicated attributes per category in order to make the task easier of the model to extract the location entities. In (a) there are specific attributes regarding the name of the sport team and its hometown. In (b) we ask specifically for the political party and the business name. We used Few Shot Prompting to send specific examples, positive and negative ones, in order to improve the consistency of ChatGPT. Once done, we compared the results of both GPT 3.5 and GPT 4 against our current NER engine. The specific questions on sports such as “What is the sports team name and where is it playing?” contributed 3% more to the coverage of local items in GPT 3.5 and 6% more in GPT 4 (see Fig. 1a). While specific questions on business or politics did not contribute that much (e.g., “Where is the business headquarters?”, “What is the political party related to that person and where is it based?”). In sum, GPT 4 was 6 times better than NER, GPT 3.5 was 3.5 times (haha!) better than NER, and in total NER had an answer that was different from GPT 3.5 in 1% of the cases and from GPT 4 in 0.5% of the cases. Figure 1a: Classifying news articles based on titles using NER vs. ChatGPT. (a) One day of 2K unique titles of one day sent to NER, gpt-3.5-turbo-0301 and openai-gpt4-32k for extracting the location of the titles (city/ state/ country). The blue bars are the location entities extracted only from the title (implicit and explicit). The red bars are the added value to the coverage contributed from the specific questions on sports. Yellow and green signifies the added value to the coverage contributed by specific questions on business or politics. We tested this further with ~8K unique titles over a week only with GPT 3.5 and only on sports sections and saw that there is a potential of using LLMs to increase the coverage of location entities (see Fig. 1b). Figure 1b. Classifying location in news in sports placements based on titles using NER vs. ChatGPT. Sports section coverage on 577 unique titles. The blue bars are the location entities (city/ state/ country) extracted only from the title (explicit or implicit). The red bar is the added value to the coverage contributed from the specific questions on sports. Yellow and green marks the specific added value to the coverage contributed specifically by questions about business or politics. ## Performance and cost This POC compared two versions of GPT vs. a standard NER. The major differences were the cost and duration (see Table 2). GPT 4 was far more expensive than GPT 3.5 turbo ($180 vs. $10), and was also 1.5 hours slower. On the other hand, it was more reliable and returned more valid results on the first try than GPT 3.5 (98% vs. 83%). Model GPT3 (gpt-3.5-turbo-0301) GPT4 (openai-gpt4-32k) Response time per prompt 5~20 sec 10~30 sec Price $0.002/ 1k tokens prompt $0.03/ 1k tokens response $0.06/ 1k tokens Tokens input: 3,776,796 output: 1,054,222 input: 3,776,796 output: 1,097,167 Titles returned 83% 98% Cost $9.66 $179.13 Duration 4:34:13 hours 6:07:02 hours Table 2: Performance and cost of the POC, sending 2K titles to GPTs and to a standard Named Entity Recognition engine. Prices were taken from https://openai.com/pricing. ## Are you telling the truth my friend? ChatGPT and similar LLMs are pruned to hallucinations. Thus, we first added a basic validation for all the engines in the POC, including NER, to verify that the city, state and country do exist in the world, and that the city is indeed inside the state and the state inside the country. Is it enough? - unfortunately no.  What if the location entities exist in the world, but are not related to the article at all? To tackle this, we fetched a dataset of 9K unique titles from our local publishers' editors and their tagging of local/ national as the ground truth. We added an extra logic on the ChatGPT output to decide whether the items are local/ national according to the relevant publishers. Then, we calculated the precision and recall. This was done for 2 flavors of inputs: (I ) Unique titles of news items. (II) Unique titles of news items and their 1st paragraphs. The results of the first iteration (unique titles only) were 92.1% precision and 27.3% recall. It means that we succeed in understanding what is national, but we missed many local items. When thinking of it, the editors have more knowledge on every news article than the LLM. The editors read the full text of the article, and have a broader context, while our LLMs got only the title. As a side note, we could use location classification instead. Meaning that we will send the publisher location alongside the prompt and ask the model whether the item is related to that particular city. We did not do it, mainly because we planned to use the item location information in other places in the system and sometimes users from neighboring regions are also interested in cities outside of their locality. ## Trying again - this time with more text We added the 1st paragraph of each article to the title (input version II), and tried again. The results in Table 3 present the LLMs precision and recall while also sending the 1st paragraph. As can be seen, the precision stayed pretty high (86.8%), and the recall improved (35%). We still consider many local items as national, but our intuition regarding a broader context was validated and it seems that the next step is to add the full text of the article. This, of course, will be done with considerations of scale and performance, since articles can vary in length, and some can be very long. It is important to note that, in our case, precision is prioritized over recall. This means that accurately classifying the locality of items is more important to us than capturing all local items. This is mainly because we aim to promote local items in specific slots provided by local publishers. These slots require a small number of items (1-3). Therefore, as long as we can accurately identify enough items, we can ensure that at least one local item will appear in each section. Table 3: A comparison of the LLM outputs to a ground truth. 9K titles and 1st paragraphs of news items from our local publishers were sent to LLM for extraction of location entities. An extra logic was added to the output to understand whether the item is local or national according to the relevant publisher. Then, precision and recall were measured. The precision is high (86.8%), meaning only few national items mistakenly considered local (false positives), while the recall is low (35%), meaning we miss many local items and consider them as national. ## POC looks promising - let’s productize! Taboola constantly crawls new articles. In order to classify these articles using LLM, we needed to integrate the classification flow into our existing crawler pipeline (see Fig. 2). The classification process was performed asynchronously and offline. We sent batches of article titles and their first paragraphs to ChatGPT for classification. The results were saved in Cassandra and made available during serving time. Taboola's models are continuously trained for real-time click-through rate (CTR) prediction. To incorporate new location information into the training process, we included it as a categorical and historical feature in the recommendation model. This allows us to determine whether the location is local or national based on the publisher's policy. The model then personalizes the ranking of relevant candidate items for the user and relevant context, and returns a list of recommendations as the result. Figure 2: Recommendation system integration with LLMs. Offline flow (Orange) - sending asynchronous batches to LLM by keeping the rate limit. Sending unique prompts and saving the result in C*. Recommendation model is being trained on the LLM entities retrieved from C*. Online flow (Green) - recommendation request is being sent for a given item, user and context, retrieving the LLM result from a cache and building features, performing inference to the pretrained model and getting a list of ranked recommendations. ## Did it work? Yes! We succeeded in serving 48% more local content to our in-market users in local publishers. We have chosen a few selected local publishers, and measured the system impact by measuring the amount of local items served to the user after adding the item location information to the model. Before the treatment we served 1 local item every 15 items, the treatment on the other hand served 1 out 6 local items (see Fig 3.), more than doubling the amount of local items. In terms of user engagement, we have noticed the most significant user engagement improvement on users within the same city, with additional 16% clicks-per-user. Figure 3: A/B test impact analysis. The average number of local items served measured in placements with 3 slots for in-market users. With Treatment, we served 1 local item out of 6 items (2 pageviews * 3 slots). In Control we served 1 local item out of 15 items (5 pageviews * 3 slots). ## To summarize The successful use of LLMs has increased the coverage of news items, enabling us to provide more local content on our local publishers' websites. Having a broader context, similar to that of the editors, will result in better coverage outcomes. However, it is important to take into account the potential impact on latency and cost. ## Considering a similar POC? You might want to: - Supply context and break your instructions to step-by-step phases. - Specify the exact format of the expected output. - Give a few negative and positive examples. - Validate your responses and don’t be afraid to do a retry. - Evaluate the results and make sure the answers are good enough for your use-case. Don’t trust ChatGPT. ## Good luck! ### Acknowledgements Many thanks to the great team of algo experts @Taboola, Hai Sitton, Dorian Yitschaki, Yevgeny Tkach, Elad Gov Ari and Maoz Cohen for reviewing this blogpost. ### References - Brown, T., Mann, B., Ryder, N., Subbiah, M., Kaplan, J. D., Dhariwal, P., ... & Amodei, D. (2020). Language models are few-shot learners. Advances in neural information processing systems, 33, 1877-1901. - Wei, J., Wang, X., Schuurmans, D., Bosma, M., Xia, F., Chi, E., Le QV. & Zhou, D. (2022). Chain-of-thought prompting elicits reasoning in large language models. Advances in Neural Information Processing Systems, 35, 24824-24837.‏3. --- ### Real-time Language Sync: Web App to Browser Extension URL: https://www.taboola.com/engineering/real-time-language-sync/ Last Modified: 2025-01-14 14:37:14 In previous articles, I went into an efficient implementation of internationalization using react-i18next for browser extensions and discussed the usage of event-driven models using messages, so I assumed you were already familiar with them. If you still need to read those, I encourage you to do so, as they set the groundwork for the content in this article.   ## Introduction This article explores the implementation of real-time language synchronization between a web app and a browser extension. The primary objective is to seamlessly share language preferences, with the potential to extend these principles to other information like credentials and user details. To illustrate these concepts, we'll examine a web application built with React using client-side rendering, accompanied by a browser extension. The extension implementation relies on an event-driven model built using React and Manifest v3, which will consider aspects like service workers and their lifecycle to support our real-time synchronization.     The diagram above highlights our focus on a one-way synchronization process from the web application to the extension. Whenever users update their language settings in the web application, it triggers real-time synchronization within the extension, ensuring both tools remain in sync. I’ve chosen a client-side cookie approach to facilitate this since the background script lacks direct access to the web application’s local storage. This approach offers a significant advantage — the possibility of push notifications within the extension, minimizing browser resource consumption, and establishing a single source of truth for user language within the browser. We are going towards a cookie-less world, but cookies remain a powerful tool to keep a state within the browser. If the requirements and complexity grow in maintaining the language as a cookie, we will explore alternative approaches in the future. ## Cookie Implementation in Web App Now that the scope of the article is clear, let’s dive into implementing the cookie logic within the web application. To facilitate real-time language synchronization, we start the process by creating a client-side cookie for language updates: import Cookies from "js-cookie" const languageCookieKey = "{your-webApp-name}-language" const cookieOptions = { // ... // Only set secure if we are running on an https environment. // secure: true => force cookie to be only readable by https requests. secure: window?.location?.protocol === "https:", } export function setLanguageCookie(language) { // Save language in cookie on the client side - only single browser support required Cookies.set(languageCookieKey, language, cookieOptions) } Once the logic to persist language in the cookie is established, the next step is to set up language updates whenever a user changes their language. This change could occur during the login process, where users may have distinct language preferences within the same computer or when they update their language preferences within the web application's settings panel. Here's how we handle these scenarios: function onAuthSuccess(user: User) { // Initialise your client environment after authentication //... setLanguageCookie(user.language) i18n.changeLanguage(user.language) } // ... function onUserLanguageChange(language: LanguagesSupported) { // ... setLanguageCookie(language) i18n.changeLanguage(language) } This structured approach ensures that language updates are seamlessly captured and stored in the client-side cookie, setting the foundation for real-time synchronization between the web application and the browser extension. ## Extension service worker initialization Let's begin by examining the initialization process. As you may know, the background script in Manifest V3 utilizes service workers for its implementation and is expected to run as long as tasks are performed. It's crucial to note that information stored in memory is temporary and will be cleared once the worker shuts down. Therefore, during background initialization, ensure the language is retrieved and set based on the stored cookie. try { const languageCookieFromWebApp: chrome.cookies.Cookie | null = await chrome.cookies.get({ url: `${your - webApp - url}`, // Remember taht URL change for localhost and live name: "{your-webApp-name}-language", }); if (languageCookieFromWebApp) { const language = languageCookieFromWebApp.value if (language) { chrome.storage.local.set({ language }); } } } catch (error: unknown) { console.error("Failed to initialise isSkimlinksEmployee cookie in cache", error); } Once the cookie is set, the subsequent steps align with what we've discussed in the react-i18next setup article. If there's any ambiguity, feel free to seek clarification through comments. ## Real-time Sync with Content Scripts Now, let's proceed to handle cookie notifications and event propagation for real-time language synchronization within the content scripts. To achieve this, let's set up an event listener to monitor any cookie changes. // Cookie listener for onChange events in the background.ts // cookiesHandler can be found in https://gist.github.com/EduardoAC/35ba733a64854993483ab543de066aa4 chrome.cookies.onChanged.addListener(cookiesHandler) export async function cookiesHandler({ cause, cookie, removed }: chrome.cookies.CookieChangeInfo) { // We are interested in cookies which are set/updated in the browser const hasUpdateCookie = !removed && cause === "explicit"; // We are interested when the cookie is removed from the browser const hasClearCookie = removed && (cause === "explicit" || cause === "expired_overwrite"); const isRelevantCookieUpdate = hasUpdateCookie || hasClearCookie; // We monitor only cookies within the our webApp domain const yourWebAppDomain = extractDomain(`${your-webApp-url}`); // Check for the language cookie belonging to your-webApp-url domain const isLanguageCookieFromHub = cookie && cookie.domain && cookie.domain === yourWebAppDomain && cookie.name === "{your-webApp-name}-language"; if (isRelevantCookieUpdate && isLanguageCookieFromHub) { const language = cookie.value; // Assuming value is the language code if (language) { updateLanguage(language); } } } export async function updateLanguage(language: LanguagesSupported) { chrome.storage.local.set({ language }); broadcastMessageAllTabs({ type: "languageUpdated", data: { language: language, translations: i18n.getDataByLanguage(language), }, }); } ## Message Handling in Content Scripts We've successfully gone through updating the language and propagating it from our web application to our extension background script, sending an event updating the content script active tabs. Now, what's left is to listen to these events and update the UI accordingly. function initMessageHandler() { // ... useEffect(() => { // Listening for updates by the service worker chrome.runtime.onMessage.addListener(handleMessageListener); return () => { chrome.runtime.onMessage.removeListener(handleMessageListener); }; }, []); } const handleMessageListener = (message: Message) => { // ... if (message.type === "languageUpdated") { const { language, translations } = message.data; // namespace (ns) group set of translations (separated files) i18n .addResources(language, "${ns}", translations) .changeLanguage(language); } }; In the implementation, I've decided to initialize translation when the content script loads within the tab page, creating a new instance with the translation from the language at the time (Using react-i18next within Chrome extension). However, we need to monitor language updates to allow real-time language synchronization. As you can see above, we integrated the listener as part of the React lifecycle to gain more granular control for the onMessage listener. And that's it. I hope you find this article useful, and drop me a line if you have any questions. ## Conclusion Our exploration into real-time language synchronization between web applications and browser extensions centers on delivering a fluid and efficient user experience. Utilizing client-side cookies in the web application enables seamless language preference sharing and lays the foundation for extending these principles to encompass broader user details. Whether introducing a client-side cookie for language updates or initializing the extension's background with service workers, our approach ensures a persistent language environment. The one-way sync model effortlessly transmits language updates, providing a harmonious user experience and recognizing the enduring potency of cookies in maintaining the browser state. We've successfully bridged the gap between the web application and extension in content scripts, achieving synchronized language updates. A refined message handling system ensures swift UI responses to language changes, enhancing the user experience with real-time multilingual capabilities. Happy coding! ## Annexe: Two ways real-time sync between browser extension and web app While this article primarily focuses on real-time language synchronization between a web application and a browser extension using client-side cookies, it's worth contemplating scenarios where the extension updates the client cookie, prompting language synchronization within the web application. One potential approach involves the implementation of a polling mechanism to monitor changes in cookies. The provided code snippet serves as an illustrative example: import Cookies from "js-cookie"; // Function to check for changes in cookies function checkLanguageCookieChanges() { // Get the current value of the cookie const currentLanguage = Cookie.get("{your-webApp-name}-language}"); // Compare with the previous value if (currentLanguage !== checkLanguageCookieChanges.previousLanguage) { // Cookie has changed i18n.changeLanguage(currentLanguage); // Update the previous value for the next check checkCookieChanges.previousLanguage = currentLanguage; } } // Initialize the previous value checkLanguageCookieChanges.previousLanguage = Cookie.get( "{your-webApp-name}-language}" ); // Check for changes every 1 second (adjust as needed) setInterval(checkCookieChanges, 1000); This example provides a basic insight into how polling can work for the web application. For those interested in going deeper into more effective polling mechanisms, check out “Forever Functional: Three Ways Of Polling by Federico Kereki” Work on efficient polling methods offers valuable insights. Feel free to explore these avenues or reach out if you’re keen on a comprehensive example in this subject. --- ### Behind the Streams: How We Took Our CDC Infrastructure to the Next Level URL: https://www.taboola.com/engineering/cdc-infrastructure-next-level/ Last Modified: 2025-01-14 14:37:14 In today’s fast-paced world of data-driven applications, real-time access to database changes is no longer a luxury, but a necessity. Whether it’s for real-time cache updates, continuous data replication, or fraud detection, the ability to capture and process these change events in real-time unlocks a world of possibilities for modern data architectures. This blog post describes the Change Data Capture (CDC) solution we built for streaming tables’ change events (both DML and DDL operations) from our backend MySQL cluster to our main Kafka cluster, utilized by a diverse range of services. We’ll explore the chosen architecture, focusing on Debezium, Kafka Connect, and Avro serialization. In addition, we'll explain in detail how we configured Debezium in a non-standard way to meet our unique requirements and ensure a seamless migration from our previous CDC service. We’ll also discuss the challenges that emerged during the process and how we successfully overcame them. ## The Challenge At Taboola, we were previously using an in-house developed service (built on top of the widely-used library mysql-binlog-connector-java) to capture and stream DML events from MySQL tables to Apache Kafka, enabling real-time cache updates across our various services. While this service performed well for several years, its maintenance was becoming increasingly challenging: - The library stopped receiving updates and bug fixes. As a result, we occasionally encountered newly discovered bugs that required manual intervention. - We experienced intermittent instability issues due to its custom implementation for managing Kafka offsets. Therefore, we decided to look for a well-maintained open-source alternative that would meet our needs. We also wanted to take advantage of this opportunity to find a more versatile CDC software that could easily integrate into our wide range of data pipelines. ## Introducing Debezium After a thorough evaluation of various open-source CDC platforms, Debezium stood out as the leading choice in terms of functionality, maintenance and robustness. Additionally, it supports many other popular DBMSs apart from MySQL, including: PostgreSQL, MongoDB, SQL Server, Oracle and Cassandra. Another significant advantage is Debezium’s large and active community, which provides exceptional support and resources. With these factors in mind, we decided to use Debezium to facilitate the MySQL part of the overall solution. However, to fully meet our specific needs, we found it necessary to go beyond the default settings and customize it in several ways. We’ll dive deeper into these customizations and how they unlocked the full potential of Debezium in the following sections. ## Key Requirements and Overall Goals First, we needed to address all the requirements for building an optimal solution while ensuring backward compatibility with our previous service. To achieve this, we formulated the following set of requirements: ### Consumer requirements: - Single target topic All events, from all captured tables, should be streamed into a single Kafka topic. This was essential to ensure backward compatibility with our previous service. - Event record structure The record key should be a string that represents the full table name in the format schema_name.table_name. The record value should be a nested structure, including the changed row values both before and after the change. ### Producer requirements: - Fault-tolerance The producer must be resilient and highly available. - MySQL GTID support It must support the MySQL GTID mechanism to ensure that no events are missed, even in the event of a MySQL source server failover. - Minimal latency Events’ latency from MySQL should be as low as possible, ideally within the sub-second range. - Minimize requests rate and efficient batching To prevent overloading the Kafka cluster and reduce network round trips, the producer should maintain a reasonable request rate, ideally between 5 to 10 requests per second. Additionally, aggregating as many events as possible within each request will enhance batching efficiency. - Storage space efficiency Aim for compact-sized streamed records to reduce Kafka topic size and optimize network bandwidth usage. ## Solution Overview After conducting thorough research and extensive testing, we reached the following solution: Figure 1. The solution design (in high-level) The solution is based on the following two main components: 1. Kafka Connect with a customized Debezium connector We deployed a Debezium connector for MySQL, customized to our specific requirements, on top of Kafka Connect. This is the recommended deployment type by Debezium developers. Kafka Connect is an open-source framework designed to seamlessly integrate Apache Kafka with external systems. It provides connectors for a wide variety of external data sources and sinks, for ingesting data into Kafka or exporting data from Kafka. It is fault-tolerant and scalable, making it a powerful tool for data integration. 2. Avro serialization By default, the Debezium connector produces records in JSON format. While this is generally acceptable for databases with a low to medium transaction rate, it can present challenges in write-intensive workloads, as in our case.For us, Avro serves as an excellent alternative to the JSON converter. It serializes the record keys and values into a compact and efficient binary format. This, in turn, reduces the overall load on the Kafka cluster and results in smaller Kafka topics. In our particular case, we only needed to serialize the record value in Avro format since we designed the record key as a simple schemaless string value. To enable Avro serialization, we needed to set up a Schema Registry service. This service acts as a centralized repository for managing schemas used for data serialization within Apache Kafka. It plays a pivotal role in data governance, offering features such as data validation, schema compatibility checking, versioning, and evolution. Currently, we are only using its core functionality, but we look forward to exploring the rest of its valuable features and integrating them into our data streaming infrastructure, taking it to the next level. ## Debezium Custom Configuration ### What we already had out-of-the-box We were already able to accomplish some of the requirements by utilizing Debezium’s out-of-the-box capabilities: - Event record structure Each record value produced by Debezium includes the changed row state both before and after the change, plus additional useful metadata such as the change timestamp, transaction ID and more. On the other hand, the record key defaults to a structure composed of the changed table’s primary key fields - which was not compatible with our requirement. We resolved this issue by applying multiple transformations on the produced records (see Transformations Config below). - MySQL GTID support, Minimal latency Both are supported by Debezium out of the box. - Fault-tolerance Debezium is designed to handle a range of failures (such as a MySQL database crash) without ever missing or losing a change event. Additionally, the high availability capabilities of Kafka Connect significantly enhance the resiliency of the Debezium connector. In order to fully meet the rest of our requirements, we made additional custom configuration changes to the connector, which include: - Transforming the produced change events records by using Kafka Connect Single Message Transform (SMT) feature. - Customizing the serialization formats of the record key and value by overriding the default properties of Kafka Connect converters. - Overriding some of the Kafka producer properties used by this connector. - Fine-tuning several internal configuration properties of the Debezium MySQL connector. Let’s take a closer look at the corresponding sections in the connector’s configuration. ### Transformations config Single Message Transform (SMT) is a simple interface for manipulating records within the Kafka Connect framework. As the name suggests, each transformation operates on every single message (record) in the data pipeline as it passes through the Kafka Connect connector. For example, you can add, rename, or drop a field, or even change a record’s topic. In our case, the Debezium source connector passes records through these transformations before they are written to the Kafka topic. We configured four different transformations to be applied on each event record produced by Debezium: #### 1. RemoveOptionalDefaultValue "transforms.removeOptionalDefaultValue.type": "com.taboola.kafka.connect.transforms.RemoveOptionalDefaultValue", This is a custom transformation developed at Taboola as a workaround for an issue encountered with the Avro and JSON converters. For reasons unknown to us, these converters replaced all explicitly specified NULL values in optional fields with their corresponding default values, as defined in the record’s schema. While this behavior may be suitable for many other use cases, it did not align with our requirement to preserve the exact row values from the source database. The transformation operates on the record schema by removing the default value associated with each optional (nullable) column. As this behavior has recently become configurable in both Avro and JSON converters, we’ll likely no longer need to use this transformation in the near future. For more detailed information, please refer to: - Debezium Jira: DBZ-4263 - Apache Kafka Jira (JSON Converter): KAFKA-8713 - Confluent GitHub (Avro Converter): Issue #2314 2. ByLogicalTableRouter "transforms.addSourceTableToKey.type": "io.debezium.transforms.ByLogicalTableRouter", "transforms.addSourceTableToKey.key.field.name": "__source_table_name", "transforms.addSourceTableToKey.key.field.regex": "^mysql_backend_updates\\.(.+)$", "transforms.addSourceTableToKey.key.field.replacement": "$1", "transforms.addSourceTableToKey.topic.regex": "^mysql_backend_updates\\.(.+)$", "transforms.addSourceTableToKey.topic.replacement": "$1",   This transformation is provided by the Debezium platform (see Topic Routing section in Debezium docs). We configured it to perform two actions: - Add a new field __source_table_name to the record key. This field contains the full table name from the MySQL source (e.g: "__source_table_name": "tbla.config_tbl_a"). - Modify the target topic name in the record metadata from Debezium’s default format connector_logical_name.schema_name.table_name to schema_name.table_name. This affects the Connect record schema name that Debezium generates, which is derived from the target topic name (see Figure 2 below). It ultimately affects the Avro schema name registered in the schema registry by the value converter (more details in the following Value Converter section). Please note: This transformation will work as expected as long as the captured table has either a defined primary key or unique key. Figure 2. How ByLogicalTableRouter transformation operates based on our configuration 3. ExtractField "transforms.extractKeyField.type": "org.apache.kafka.connect.transforms.ExtractField$Key", "transforms.extractKeyField.field": "__source_table_name",   This is a built-in transformation provided by Kafka Connect (see ExtractField in the docs). It is applied to the record key - which, as previously mentioned, comprises the primary key fields of the changed table along with the __source_table_name field (that was added by the previous transformation). This transformation extracts the value of the __source_table_name field and sets it as the new record key. In other words, it transforms the key from a structure with multiple fields and values into a single string value (see Figure 3 below). For example, the structured key { "id": 123456, "__source_table_name": "tbla.config_tbl_a" } would be transformed into the string "tbla.config_tbl_a". Figure 3. How ExtractField transformation operates on the record key 4. RegexRouter This is a built-in transformation provided by Kafka Connect (see RegexRouter in the docs). By default, Debezium assigns the target topic for each event record based on the table that event is associated with (resulting in a distinct Kafka topic for each table). This transformation simply changes the target topic for each record to "table_updates". Figure 4. How RegexRouter transformation modifies the record’s metadata "transforms.changeTopic.type": "org.apache.kafka.connect.transforms.RegexRouter", "transforms.changeTopic.regex": ".*", "transforms.changeTopic.replacement": "table_updates",   Applying these transformations allowed us to successfully address the remaining consumer requirements: - Single target topic - Event record structure ### Converters config Following our (unique) data serialization choices, we needed to configure the suitable Kafka Connect converters for both the record key and value. 1. Key converter: "key.converter": "org.apache.kafka.connect.storage.StringConverter",   Since the record key was transformed from structure into a simple string value (using the ExtractField transformation), it should now be serialized as string. Therefore, we set the StringConverter as the key converter. 2. Value converter: "value.converter": "io.confluent.connect.avro.AvroConverter", "value.converter.schema.registry.url": "http://schema-registry.example.com:8081", "value.converter.value.subject.name.strategy": "io.confluent.kafka.serializers.subject.TopicRecordNameStrategy",   Since we decided to serialize the record value in Avro format, we set the AvroConverter as the value converter. However, because we are producing events with multiple schemas (one schema per table) under the same topic, it didn't make sense to use the schema registry’s default subject name strategy: TopicNameStrategy. This would have resulted in all of the different schemas being registered under one subject named table_updates-value in the schema registry. To address this, we changed the subject name strategy to TopicRecordNameStrategy. This strategy registers each schema under a separate subject in the form of topic_name.schema_name. This allows us to keep the different schemas separate, and it also allows us to manage multiple versions of each schema. For instance: - The subject table_updates-tbla.first_tbl.Envelope stores the schema of tbla.first_tbl table events (and all of its versions). - The subject table_updates-tbla.second_tbl.Envelope - stores the schema of tbla.second_tbl table events (and all of its versions), and so forth. ### Kafka Connect producer config By default, connectors inherit their Kafka client configuration properties from the Kafka Connect worker configuration. To achieve our desired request rate and efficient batching, we found it necessary to override the following producer-specific properties within the connector configuration: "producer.override.compression.type": "lz4", "producer.override.linger.ms": "250", "producer.override.batch.size": "10485760", "producer.override.max.request.size": "104857600", "producer.override.buffer.memory": "2147500000"   - The producer.override.compression.type property specifies the compression algorithm applied to the produced record batches. We changed its value from the default ‘none’ (no compression) to ‘lz4’ because, based on our experience, LZ4 offered the fastest performance. We also increased the value of producer.override.batch.size from 16KB to 10MB to allow for larger batches. This led to better compression ratios and ultimately increased the overall throughput. - To reduce the overall number of requests made to the Kafka cluster, we increased the value of producer.override.linger.ms to 250. This means that the connector will wait up to 250 milliseconds for additional event records to accumulate before sending them together as a batch, rather than dispatching each record as it arrives. ### Debezium internal config: To prevent Debezium itself from becoming a bottleneck within our streaming pipeline and to fully optimize its performance, we fine-tuned the following internal configuration properties: "poll.interval.ms": "200", "max.batch.size": "300000", "max.queue.size": "1000000",   - We reduced the poll.interval.ms property from its default of 500 milliseconds to 200 milliseconds. This allows the connector to poll MySQL for change events more frequently. - Furthermore, we set significantly high values for the max.batch.size (default: 2048) and max.queue.size (default: 8192) properties. This ensures that the connector can keep up with the MySQL database, even during periods of exceptionally high workloads. ### Putting it all together Here’s a complete example of such customized Debezium MySQL connector: { "name": "mysql-backend-updates-source-connector-01", "connector.class": "io.debezium.connector.mysql.MySqlConnector", "tasks.max": "1", "poll.interval.ms": "200", "max.batch.size": "300000", "max.queue.size": "1000000", "database.hostname": "mysql-server.example.com", "database.port": "3306", "database.user": "dbz_user", "database.password": "dbz_password", "database.server.id": "123456", "database.server.name": "mysql_backend_updates", "table.include.list": "tbla.config_tbl_a, tbla.config_tbl_b", "snapshot.mode": "schema_only", "snapshot.locking.mode": "none", "time.precision.mode": "connect", "tombstones.on.delete": "false", "database.history.kafka.bootstrap.servers": "kafka-broker.example.com:9092", "database.history.kafka.topic": "dbz-mysql_backend_updates-history", "transforms": "removeOptionalDefaultValue, addSourceTableToKey, extractKeyField, changeTopic", "transforms.removeOptionalDefaultValue.type": "com.taboola.kafka.connect.transforms.RemoveOptionalDefaultValue", "transforms.addSourceTableToKey.type": "io.debezium.transforms.ByLogicalTableRouter", "transforms.addSourceTableToKey.key.field.name": "__source_table_name", "transforms.addSourceTableToKey.key.field.regex": "^mysql_backend_updates\\.(.+)$", "transforms.addSourceTableToKey.key.field.replacement": "$1", "transforms.addSourceTableToKey.topic.regex": "^mysql_backend_updates\\.(.+)$", "transforms.addSourceTableToKey.topic.replacement": "$1", "transforms.extractKeyField.type": "org.apache.kafka.connect.transforms.ExtractField$Key", "transforms.extractKeyField.field": "__source_table_name", "transforms.changeTopic.type": "org.apache.kafka.connect.transforms.RegexRouter", "transforms.changeTopic.regex": ".*", "transforms.changeTopic.replacement": "table_updates", "key.converter": "org.apache.kafka.connect.storage.StringConverter", "value.converter": "io.confluent.connect.avro.AvroConverter", "value.converter.schema.registry.url": "http://schema-registry.example.com:8081", "value.converter.value.subject.name.strategy": "io.confluent.kafka.serializers.subject.TopicRecordNameStrategy", "producer.override.compression.type": "lz4", "producer.override.linger.ms": "250", "producer.override.batch.size": "10485760", "producer.override.max.request.size": "104857600", "producer.override.buffer.memory": "2147500000" } ## Conclusion Setting up this new infrastructure, along with the customizations we implemented, enabled us to fulfill all the requirements, resulting in a smooth migration from our in-house solution. Moreover, it established the groundwork for implementing CDC-based data pipelines in various other use cases, extending beyond real-time cache updates. Migrating to an open-source-based solution posed quite a few challenges along the way, involving a learning curve regarding Debezium’s internals and the Kafka Connect framework. Nevertheless, the overall successful outcome made the journey completely worthwhile. ## Final Thoughts We've been using this solution for over a year now, seamlessly integrating it into multiple streaming data pipelines developed to address various use cases. It has consistently proven to be high-performing, reliable, well-maintained, and versatile. Furthermore, all of the mentioned open-source components (Debezium, Kafka Connect, Confluent Schema Registry) have vibrant and supportive communities. Personally, I’ve been particularly impressed by the support from the Debezium community. I encourage you all to join their Zulip chat and check out the insightful discussions there. On a separate note, we’ve actively contributed to Debezium by proposing new features and providing a few bug fixes (which may even be a topic for another blog post). We believe that Debezium is a valuable tool for data engineers and architects, and we are excited to see how it continues to evolve in the future. Notable References - Five Advantages of Log-Based Change Data Capture - Using Debezium, CDC for Apache Kafka, with PostgreSQL and MongoDB - Putting Several Event Types in the Same Topic – Revisited --- ### Optimising Chrome Extensions: Part 1 - Beyond Redux, Post-Manifest v3 URL: https://www.taboola.com/engineering/optimising-chrome-extensions-part-1/ Last Modified: 2025-01-14 14:37:14 The landscape of Chrome extension development underwent a significant shift with the introduction of Manifest v3. In this article, we'll explore the journey of why we decided to part ways with the Redux ecosystem in our Chrome extension after adopting Manifest v3. We aim to provide insights and guidance for fellow extension developers facing similar decisions in selecting their tech stack. ## The Redux Era in Chrome Extension Development Our initial foray into Chrome extension development was rooted in manifest v2, where we employed a robust React-Redux stack. This setup, bolstered by the webext-redux package by Tyler Shaddix, facilitated seamless state management across background scripts and content scripts. The Chrome runtime connect mechanism allowed real-time synchronisation using the subscription pattern, enabling easy usage of Redux across all our scripts. ### Challenges Encountered However, no architecture is without its challenges. One notable drawback was the duplication of the store in memory across every tab where our extension ran. As our application grew in complexity, this led to potential inefficiencies, particularly considering today's systems that might run many tabs simultaneously. Additionally, the changes introduced in manifest v3, where the background script no longer persisted in memory, presented a hurdle. Connections reset unless explicitly persisted in the cache, aligning with the new architecture's goal of handling incoming messages efficiently. ### Adapting to Manifest v3: A Redux Dilemma While Redux had served us well, its drawbacks in the context of Chrome extension development became apparent: - Duplication of Store: Storing the same data in memory across multiple tabs posed scalability challenges. - Service Worker Synchronization: Maintaining synchronisation between reloads and shutdowns of the service worker became a requirement, introducing complexities like caching and restoring the store and making changes to enable broadcasting state as runtime connect is lost between SW executions. - Increased Message Overhead: Numerous messages between content and background scripts, especially during data loading from the network, added overhead that could be mitigated. - Redux as a Cache: In some scenarios, using Redux as a glorified cache became counterproductive as there are better approaches, particularly on extension, to overcome these difficulties. ### Alternative Paths for Redux Enthusiasts While our journey led us to part ways with Redux in our Chrome extension, we understand that some developers may still be keen on persisting with this familiar tool. Fear not, for alternatives exist in the post-manifest v3 era, offering compatibility and viable solutions: - Npm package @eduardoac-skimlinks/webext-redux For developers determined to stick with Redux using webext-redux, I've introduced a series of changes to support manifest v3. This new package implements the suggested alterations for manifest v3 compatibility, addressing specific challenges highlighted in PR-282 within Webext-redux. By embracing this alternative, you can seamlessly continue leveraging Redux while ensuring alignment with the requirements of manifest v3. - Npm package reduxed-chrome-storage This standalone provider offers a dedicated solution for storing the Redux store in the Chrome storage cache. While it may require a shift in your approach, this alternative ensures persistence and efficient data handling within the Chrome ecosystem. Choosing the right path post-manifest v3 depends on your specific needs, preferences, and the intricacies of your extension. As you embark on this alternative journey, consider the nuances of each solution to find the best fit for your Redux-powered Chrome extension. Remember, adaptability is key in the ever-evolving landscape of Chrome extension development. ## Moving Beyond Redux: Embracing Efficiency In our journey towards adapting to the changes brought by manifest v3, it became evident that the challenges posed by Redux in the Chrome extension context were becoming more pronounced. We recognised the need for a more efficient and streamlined approach to handle state management without the inherent complexities of Redux. ### Minimising User Impact The imperative to minimise the impact on the end-user experience was at the forefront of our considerations. The duplication of the store in memory across tabs, an inherent characteristic of Redux, emerged as a significant concern. Our quest for alternatives took shape, intending to reduce unnecessary redundancies, ensuring a smoother and more resource-efficient user journey. ### Exploring efficiency mechanism for caching data on the extension #### Caching Headers In our pursuit of efficiency, we considered implementing caching headers in responses. However, a significant challenge emerged regarding URL paths. Using URL matching, the cache risked retrieving items from another user's session, notably when two users shared the same computer. This issue was exacerbated by using an authentication header instead of a cookie for these requests (More details in Section 4.2 of ). Unfortunately, adapting for caching headers would require substantial changes due to our existing API structure and credential handling. Furthermore, it's essential to note that once caching headers are stored, they cannot be purged automatically. Manual user action is required, introducing an additional layer of complexity. As a result, our proposed solution had to be temporarily shelved. This setback underscores the intricate considerations and limitations in optimising our extension's efficiency, particularly when faced with challenges that, for now, remain insurmountable. #### Chrome Storage Following challenges encountered with the browser cache, we determined that the optimal solution for persisting data between service worker executions was to leverage the Chrome Storage API. This approach enabled asynchronous persistence of the data required for the extension to function. However, our exploration revealed complexities associated with handling concurrency. For an in-depth understanding of these complexities and how we manage concurrency in Chrome Extensions, please check out our detailed explanation in Managing Concurrency in Chrome Extensions. This additional resource delves into the intricacies of ensuring seamless data persistence within the Chrome Storage API. ### Optimising pub-sub messaging efficiency Messaging in Redux necessitates a constant back-and-forth to keep all tabs synchronised, given that the store is shared across them. However, upon a thorough analysis of our system, we identified two main flaws contributing to the high message volume. Firstly, recognising that each tab requires information based on its context, we realised that tabs don't need to retain all data in the Redux store. For instance, if Tab 1 needs data A to render and Tab 2 needs data B, Tab 1 doesn't need to keep both A and B in Redux. We pivoted towards a content script-centric approach, where each script requests and retains only the required data, minimising the message volume to essential updates. Secondly, we revisited how data was requested in Redux, opting for a more holistic approach. Rather than asking for specific sections of the state in the store, we crafted messages with enough abstraction. These messages are designed to be comprehensive, eliminating the need for multiple requests. This approach strikes a middle ground, simplifying the process without the complexity of fetching all data in a single message, akin to GraphQL. This strategic shift aimed to free us from the constraints of duplicated stores across multiple tabs, seamlessly aligning with our overarching goal of optimising the extension's performance. ## Benefits of Moving Beyond Redux #### Simplified Architecture The decision to distance ourselves from Redux brought about a significant simplification of our extension's architecture. Given the evolving landscape driven by manifest v3, this streamlining process proved especially crucial, which nudged us towards adopting more efficient and streamlined development practices. #### Empowering Content Scripts with Greater Autonomy Rather than constructing our extension around a centralised state for the entire application, we've embraced a paradigm where each content script manages its internal state independently. In this approach, we leverage the service worker as both a cache and a sophisticated data processing system, facilitating efficient persistence of network data. Despite the advancements in our architecture, synchronisation of shared data remains essential. We achieve this by utilising the service worker to broadcast data across all scripts, a mechanism akin to how webext-redux listened for state changes. ## Future-Proofing Extension Development #### Adapting to Manifest v3 Moving beyond Redux was not merely a reaction to challenges but a proactive response to the transformative changes introduced by Manifest v3. Adapting to this evolving architecture was paramount for ensuring our extension's continued success and relevance in the dynamic Chrome ecosystem. #### Continuous Quest for Optimization The ever-evolving nature of Chrome extension development demands a perpetual quest for optimisation. Our exploration of tailored solutions aligned with this principle, ensuring our tech stack stays agile, responsive, and well-prepared to tackle future changes and challenges. In the dynamic landscape of Chrome extension development, our strategic shift beyond Redux is a testament to our commitment to efficiency, user satisfaction, and the perpetual pursuit of optimisation. ## Conclusion: Crafting a Tailored Tech Stack As Chrome extension developers, adapting to evolving architectures is crucial. While Redux has its merits, manifest v3 prompted a reassessment. By addressing the unique needs of our extension and opting for efficient data and state handling, we've developed a more tailored and effective approach beyond Redux's constraints. Flexibility, adaptability, and an unceasing quest for optimisation are essential in the dynamic realm of Chrome extension development. "Moving beyond Redux in our Chrome extension post-manifest v3 wasn't about abandoning a tool but embracing our extension's evolving needs. By minimising user impact and exploring alternatives aligned with the new architecture, we've crafted a tech stack suiting the dynamics of post-manifest v3 Chrome extension development." Stay tuned for our upcoming deep dive! Our follow-up post will reveal the practical implementation of our tailored tech stack, maximising extension performance. Expect step-by-step guidance, insightful code snippets, and real-world examples as we guide you through the seamless transition from Redux to our optimised approach. Elevate your Chrome extension development with our hands-on guide, pushing the boundaries of performance. Watch this space for cutting-edge insights and actionable tips! --- ### Optimising Chrome Extensions: Part 2- Managing your state and communication in React URL: https://www.taboola.com/engineering/optimising-chrome-extensions-part-2/ Last Modified: 2025-01-14 14:37:14 In the previous article, we discussed the limitations of using Redux and the need to adapt our approach to Chrome extension development due to the changes introduced by Manifest v3. However, it's essential to understand the role that Redux played in our extensions and how to replace it with the new architectural strategy using only React. In this article, we'll focus on managing the state in your Chrome extension using React in content scripts, briefly caching data in the background script, and facilitating effective communication between these components to ensure data synchronisation. ### Understanding State in Content Scripts When we talk about "state," it can be a somewhat ambiguous term. The state could refer to a state machine that controls the user interface, or it could be about how we persist data in our React application. In this article, we'll focus on the latter, referring to how you persist data within your React application, as the nature of your project and the user experience you aim to provide will determine the specific state requirements for data persistence. If you're interested in discussing user experience, feel free to leave a comment (spoiler: I firmly believe a browser extension should NOT use react-router).Now, let's dive into the concept of state, which remains a crucial part of our Chrome extension, even as we move away from Redux. We can split the state into two categories: - Shared/Global State: This category encompasses data shared across all content scripts, such as user data, settings, and feature flags. Any changes made to this data should prompt an update in both the background script and all content scripts simultaneously. - Local/Pseudo-Global State: This category covers specific data sets required for a particular content script at a given time. While it may be shared indirectly with other content scripts, it doesn't necessitate active notification to other scripts. Let's take as an example a multilanguage browser extension that displays a toolbar with the reviews for a site based on Google or Trustpilot reviews, illustrated by the image below.   In this scenario, the selected language is considered a shared/global state. The review itself and the language options for the dropdown can be classified as local/pseudo-global state, even if they are present in multiple tabs since their visualisation doesn't directly affect other content scripts. However, if we decide to replace the review display with a user rating input, allowing the user to provide their rating on the page, then the rating changes become a shared state. This is because other tabs may be displaying the same page, and they need to reflect the user's review. ### Why Does This Distinction Matter? Distinguishing between shared and local states is essential because elements belonging to the shared state require different treatment than local ones. When shared state values change, the background script must broadcast these changes to all relevant tabs that need the information to be updated. ## Managing Communication Between Content Scripts and Service Workers We've discussed the various natures of data in content scripts and touched on how shared/global state information flows. Now, let's explore the most common data flows in browser extensions based on the nature of the data, assuming you're dealing with client-side rendering content scripts. ### Fetching Information with Messages The most straightforward method of communication is through messages, allowing the content script to request data from the background script. For example, when the content script needs information to display a review, it sends a message to the background script, as illustrated in the image above. Then, the review handler is responsible for handling the message and making an attempt to retrieve the information from the cache or memory in order to minimise constant network traffic. However, if it doesn't find the data, it triggers an API request to obtain the necessary information and then stores it in the cache. Simultaneously, the response is sent back to the content script that initiated the request, as shown above. ### Broadcasting by the Background Script Sometimes, there is a need to inform all content scripts about changes initiated externally, such as a user's language change through a shared cookie with your main website. When such a scenario arises, it becomes essential for all content scripts to update their internal state accordingly. To achieve this, the background script broadcasts the change to all content scripts. For example, as illustrated in the image above, we actively listen for language cookie changes. When the language cookie is updated, this triggers an update of the background script's state, as well as all tabs that are running our content script. These tabs will require updating to align their language preferences in their internal state. ### Combined Flow: Update/Fetching and Broadcast In specific scenarios, you may encounter a situation where a change in the extension's state requires sending a message. This message, in turn, initiates a broadcast to all content scripts. Let's consider a practical example to illustrate this mixed flow. For instance, going back to our example about the language selection change, if one of the toolbar elements in a tab experiences a language change, it will be required to notify the background script, triggering internal updates within the background script and external systems like the language API and cookies, as well as ensuring alignment across all other content scripts, as shown in the image above. Notably, you can exclude the message sender from the broadcast, thanks to the tab ID available in "senderTab. ## Illustrating State and Flows with Code Fragments Now that we have a clear understanding of the concepts and the defined flow. Let's dive into some code fragments to illustrate the implementation of each concept. Keep in mind that these are not fully optimised implementations but rather fragments designed to solidify the knowledge you've gained with practical examples. A complete implementation of the review example will be provided in the next article, so stay tuned. ### Sending Messages from Content Scripts Let's start by defining a promise-friendly implementation for sending messages from a content script to the background script. Here's a sample code snippet: export interface Message { type: string subtype: string data: T } export interface Response { status: number data: V } export function sendMessage<T, V>(message: Message) { return new Promise((resolve, reject) => { try { // You may require to adjust Response based on your project implementation chrome.runtime.sendMessage(message, (response: Response) => { if(response.status < 400) { resolve(response.data) } else { reject(response.data) } }) } catch (error) { // Error must match V type reject(error) } }) } // Example using numbers const responseDataNumber = await sendMessage<number, number>({ type: "hello", subtype: "world", data: 1}) As you can see in the code above, using promises simplifies integration with the rest of your code. We've also introduced a two-level message identifier (type, subtype) to categorise handlers based on their purposes. Feel free to adapt this approach to your specific needs. ### Receiving Messages from Background and Content Scripts The good news is that both content and background scripts can listen for messages through the same `chrome.runtime` API, which includes the onMessage listener. Here's how you listen for communication between scripts: // Message and response are define in https://gist.github.com/EduardoAC/000b1e39a6ec10a892e7c6cd93730a53 chrome.runtime.onMessage.addListener((message: Message, sender, sendResponse) => { if(sender.tab) { // Sender Tab useful mostly for background script switch(message.type) { case "review": // Tab is useful for instance to obtain the url to fetch the review from reviewHandler(sender.tab, message, sendResponse) break case "language": languageHandler(sender.tab, message, sendMessage) break default: sendResponse("Error: Not found message type") break } return true // This is really important, tells the extension whether is an ASYNCHRONOUS sendResponse or not } return false // False means synchronous response }) When listening to messages, there are several important aspects to consider: - Sender Tab: Utilizing the sender tab can provide valuable information about the tab, such as its page URL. This can be particularly useful in various use cases, but some might not require this information. Consult the documentation for more details. - Message Definitions: Standardizing message definitions can help maintain consistent and homogenous logic for handling messages. - Returning a boolean: The return boolean plays a critical role in determining whether the "sendResponse" should be treated as a synchronous (false) or asynchronous (true) response. If you don't provide a return, it will expect a synchronous call to "sendResponse." While this article doesn't cover connecting this listener to the React state, I've provided a gist illustration of how you can integrate React Context with the listener. // Message and response are define in https://gist.github.com/EduardoAC/000b1e39a6ec10a892e7c6cd93730a53 interface GlobalContext { review: number language: string setLanguage: Function } const globalContext = createContext({ review: -1 language: "en" setLanguage: () => {} }) interface GlobalContextProvider { children: ReactNode } export function GlobalContextProvider({children}: GlobalContextProvider) { const = useState(-1) const = useState("en") const handleMessageListener = (message: Message) => { switch(message.type) { case "review": // Handle review, reviewHandler(message, setReview) | review = reviewHandler(message) -> state in the handler break case "language": // For simplicity, assume only message language received by content script is the new language selection setLanguageState(message.data) break default: console.error("incorrect message") // Up to you how you handle the error case break } } useEffect(() => { // Example fetching review data on context initialization retriveReview().then((review: number) => { setReview(review) }) // ... // Language and other data can be fetch together with review or separately on initialization // ... // Listening for background script message chrome.runtime.onMessage.addListener(handleMessageListener) return () => { chrome.runtime.onMessage.removeListener(handleMessageListener) } }, []) const context = { review, language, setLanguage: (newLanguage: string) => { sendMessage({ type: "language", subtype: "update", data: newLanguage}) setLanguageState(newLanguage) // Assuming that sendMessage always succes for simplicity } } return {children} } export const useGlobalContext = () => { return useContext(globalContext) } ### Broadcasting a Message to All Content Scripts The final step in implementing this flow involves sending a message to all relevant content scripts, instructing them to update the data within their local state. To achieve this, you can track each tab's status to determine whether it is actively available and loaded within the browser memory, ensuring that messages are sent to all tabs that require updates. In this article, I will present a simplified implementation of this strategy using the filtering capabilities provided by Chrome, acknowledging that some tabs may receive messages that are irrelevant to their context and will be ignored. Alternatively, you could consider tracking each time the content script is loaded on a page and only notify those tabs that are relevant to the update. I will cover this more in-depth in my following article, so make sure to follow for updates. // Message and response are define in https://gist.github.com/EduardoAC/000b1e39a6ec10a892e7c6cd93730a53 export function broadcastMessageAllLoadedTabs(message: Message) { // Get all tabs not discarded - it can be optimise further chrome.tabs.query({ discarded: false }, (tabs) => { tabs.forEach(({ id, status }) => { if(status !== "unloaded") { sendMessage(id, message) } }) }) } Within the script, you'll notice two crucial checks that require explanation: - "Discarded as false" indicates that the tab has been unloaded from memory but remains visible in the tab strip. - "Status as 'unloaded'" serves a similar purpose, although it's worth noting that there is no documentation available about this particular status. ## Conclusion In this article, we've explored the critical aspects of managing state and communication in Chrome extensions using React. We discussed the limitations of using Redux and the need to adapt our approach due to changes introduced by Manifest V3. Understanding the role of Redux in our extensions and replacing it with a React-centric architecture is essential for modern Chrome extension development. We delved into the complexities of state management within content scripts, including the distinctions between shared/global state and local/pseudo-global state. Using a multilanguage browser extension as an example, we illustrated the importance of categorising your data appropriately. By distinguishing between shared and local states, you can efficiently handle changes in your extension. Shared states necessitate broadcasting updates to all relevant tabs, ensuring data synchronisation. Our exploration of communication flows in Chrome extensions included fetching information with messages, broadcasting updates from the background script, and combining flows demonstrating message-triggered broadcasts. To help solidify your understanding, we provided code fragments illustrating implementations for each concept. These practical examples are not fully optimised but serve as valuable knowledge-building tools. Stay tuned for our next article, where we'll provide a complete implementation of the review example, connecting all the dots and further enhancing your Chrome extension development skills. Make sure to follow up for updates on future articles and practical insights into Chrome extension development with React. --- ### Managing Concurrency in Chrome Extensions URL: https://www.taboola.com/engineering/managing-concurrency-in-chrome-extensions/ Last Modified: 2025-01-14 14:37:15 Picture this: You're a developer crafting a powerful Chrome extension designed to enhance the browsing experience for thousands of users. Your extension consists of multiple content scripts, each injecting its magic into different tabs, making the web a better place, one tab at a time. Sounds promising, right? But here's the twist. These content scripts are like synchronised dancers, all performing their moves on the same stage — your user's browser. They need data to dazzle your users; sometimes, they all clamour for the same information simultaneously. This tug-of-war for resources can lead to a chaotic performance, frustrating your users and your extension in dire need of a solution. Welcome to the world of concurrency in Chrome extensions. In this post, we will dive deep into the challenges posed by concurrent data requests within your extensions and explore the architectural solutions that can help you tame this concurrency beast. Whether you're a seasoned extension developer or just dipping your toes into this exciting realm, understanding how to manage concurrency is the key to delivering a smooth and reliable user experience. So, fasten your seatbelts, and let's explore how to make your Chrome extension's multi-tab symphony a harmonious masterpiece. ## The Problem: Concurrent Resource Requests in Chrome Extensions Within the realm of Chrome extensions, a persistent challenge revolves around managing concurrency. Picture this: You have a suite of content scripts, often sharing the same codebase. This shared code can lead to a situation where these scripts simultaneously demand access to a singular resource. This challenge becomes especially pronounced when users restore multiple tabs from a previous session or engage in actions that trigger resource requests across all active tabs. The result? A potential chaos of resource requests, if not handled effectively, can lead to race conditions, sluggish performance, or even crashes, all detrimental to the user experience. ## The solution: Solving Concurrency with Service Worker Locks When it comes to addressing the challenge of concurrent resource requests within Chrome extensions, there are various approaches at our disposal. However, they all share a common requirement: the need for a lock mechanism that ensures requests from the same resource are queued until the first request is completed. ## Why a Lock Mechanism is Essential Imagine multiple content scripts racing to access the same resource simultaneously. Without a mechanism to coordinate these requests, chaos can ensue, leading to unpredictable outcomes, slow performance, or even crashes. This is where the lock mechanism comes into play. ### Understanding the Service Worker's Role The foundation of our solution lies in the nature of the service worker itself. Operating as a single thread, it processes tasks sequentially. This single-threaded architecture becomes the perfect backdrop for implementing a lock mechanism. ### How the Lock Mechanism Works In essence, the lock mechanism acts as a guardian of resource requests: - Request Queuing: When a content script initiates a resource request, it first checks for the lock's status. If a lock is active (indicating an ongoing request), the new request is patiently queued, patiently awaiting its turn. - Sequential Processing: Once the ongoing request is completed and the lock is released, the queued requests are processed in an orderly fashion. This guarantees that resource requests occur one after the other, eliminating the risks associated with concurrent access. ### The Benefits of This Approach - Conflict Resolution: By introducing a lock mechanism, we mitigate the possibility of simultaneous data access, significantly reducing the chances of race conditions, conflicts, and erratic behaviour. - Predictable Behavior: With requests being processed in a well-defined sequence, you can maintain a consistent and expected behaviour for your extension, even in the face of multiple resource requests. It's important to note that while the lock mechanism is central to managing concurrency within Chrome extensions, our focus here is primarily on its role in synchronisation. Other considerations, such as data caching strategies, can further optimise performance but fall beyond the scope of this article. In the following sections, we'll explore practical examples of implementing this lock mechanism within a real-life Chrome extension and discuss architectural considerations to optimise its use. ## Introducing the Architecture Solution: Synchronizing Resource Requests Now that we've established the significance of a lock mechanism in handling concurrency within Chrome extensions, let's delve into a practical architecture solution that utilises this mechanism. The provided code snippet serves as the foundation for this solution, enabling effective resource request synchronisation. type PendingRequests = { : Function[] } // Initialize an object to track pending resource requests let pendingRequests: PendingRequests = {}; /** * Check if there is a pending request for a given URL. * @param {string} url - The URL to check for pending requests. * @returns {boolean} - True if there are pending requests, false otherwise. */ export function hasPendingRequest(url: string) { return pendingRequests.hasOwnProperty(url); } /** * Lock a resource to prevent concurrent requests for the same URL. * @param {string} url - The URL of the resource to lock. */ export function lockResource(url: string) { if (!pendingRequests) { // Create a queue for pending requests for this URL if it doesn't exist pendingRequests = []; } } /** * Wait for a resource request to complete. This function returns a promise that * resolves when the request finishes. * @param {string} url - The URL of the resource being requested. * @returns {Promise} - A promise that resolves when the request completes. */ export function waitForResourceCompletion(url: string): Promise { // We wait until the ongoing request finishes by queuing the promise until completion const waitForRequestPromise = new Promise((resolve) => { // Add the resolve from promise to the queue for this URL pendingRequests.push(resolve); }); return waitForRequestPromise; } /** * Notify that a resource request has completed and resolve any pending requests * for the same URL. * @param {string} url - The URL of the completed resource request. * @param {Response} response - The response of the completed request. */ export function notifyResourceCompletion(url: string, response: Response) { // Notify other requests in the queue and remove the URL entry if (pendingRequests?.length > 0) { pendingRequests.forEach((resolve) => { // Resolve each pending request with a clone of the response resolve(response.clone()); }); } // Remove the URL entry as all requests are now resolved delete pendingRequests; } /** * Clear the queue of pending resource requests. */ export function clearPendingRequestQueue() { // Reset the pendingRequests object to clear all pending requests pendingRequests = {}; }   ### The Building Blocks: In the code snippet, you'll notice several functions and data structures that work together to ensure orderly and synchronised access to resources: - PendingRequests: This data structure, represented as a dictionary, plays a pivotal role in managing ongoing resource requests. It associates URLs with arrays of functions that act as promises. - hasPendingRequest(url: string): This function checks if there are ongoing requests for a given URL. It leverages the PendingRequests dictionary to determine if any requests are in progress. - lockResource(url: string): When a content script initiates a resource request, it calls this function to establish a lock for the specified URL. A new entry is created in the OnGoingRequests dictionary if no lock exists. - waitForResourceCompletion(url: string): This function is pivotal for synchronisation. It enables content scripts to wait until an ongoing request for a specific URL completes. It does so by queuing the content script's promise function in the corresponding array within the PendingRequests dictionary. - notifyResourceCompletion(url: string, response: Response): Once a resource request is completed, this function notifies other queued requests for the same URL. It resolves their promises with the response, ensuring that they proceed in an orderly fashion. Upon completion, the URL entry is removed from the PendingRequests dictionary. - clearPendingRequestQueue(): In cases where you need to clear the queue of ongoing requests entirely, this function comes in handy. It resets the PendingRequests dictionary, ensuring a clean slate. ### The Benefits: This architecture solution offers several advantages: - Synchronisation: By employing a lock mechanism and promise-based queuing, it ensures that resource requests are processed sequentially, preventing concurrency-related conflicts. - Predictable Behavior: Content scripts can rely on a consistent and expected behaviour, even when dealing with multiple resource requests for the same URL. - Resource Access Control: The solution provides a means to control access to specific resources, especially critical when dealing with shared data. ## Conclusion: Empowering Chrome Extensions with Concurrency Management In this journey through the intricacies of managing concurrency in Chrome extensions, we've explored the fundamental challenges when multiple content scripts simultaneously vie for access to the same resources. Our focus has been on providing a robust solution, and we've achieved this by implementing a lock mechanism within the service worker, ensuring synchronised and orderly resource access. Your feedback matters to us: Is there anything else you'd like to see covered, or any questions you have about Chrome extension development? We're committed to providing you with valuable insights, so please don't hesitate to share your thoughts and queries in the comments section or reach out to us directly. As you continue your path in extension development, remember the key takeaways from this exploration: - Concurrency Control is Crucial: Whether you're crafting an extension for a small user base or a large audience, managing concurrency effectively is paramount. It guarantees a seamless and reliable user experience. - The Power of Service Worker Locks: Leveraging the single-threaded nature of the service worker, a well-implemented lock mechanism can be your ally in preventing race conditions and conflicts. - Predictability and User Satisfaction: By implementing synchronisation measures, you ensure predictable behaviour within your extension. Users can interact with your extension confidently, knowing their actions won't lead to unexpected issues. - Resource Access Control: Don't underestimate the importance of controlling access to critical resources. The architecture we've explored puts you in the driver's seat, allowing you to dictate how your extension interacts with shared data. In the ever-evolving world of Chrome extension development, staying informed and equipped with the right strategies is key. We hope this exploration has been a valuable resource for you, providing insights and practical solutions to enhance your extension development endeavours. As you embark on your next extension project, keep these lessons in mind and continue to innovate, create, and provide exceptional user experiences. Your feedback and ideas drive the evolution of our content, and we look forward to hearing from you as you shape the future of Chrome extensions. Thank you for joining us on this journey, and here's to the continued success of your Chrome extensions! --- ### Breaking the Scale Barrier - Smarter Test Selection URL: https://www.taboola.com/engineering/breaking-the-scale-barrier-smarter-test-selection/ Last Modified: 2025-01-14 14:37:15 A build process is a critical procedure triggered by new code changes (Git push). Its primary responsibility is to validate the integration of new code without disrupting the existing code and ensuring the successful execution of all unit tests. However, as the number of unit tests grows, ensuring code stability becomes increasingly challenging due to the impracticality of running all tests within the constraints of time and resources. In this article, we will explore the optimizations we have implemented to achieve a faster and more efficient build process. ## The build process challenge Our development process revolves around a monorepo, a single repository housing all our code, that daily deploys its code to production. By leveraging optimized build strategies, we ensure that developers receive rapid feedback without sacrificing productivity. Integrated within our monorepo workflow, fast builds enable us to swiftly identify and address issues, ensuring that only validated and reliable code reaches production. This approach empowers our developers to work efficiently and deliver high-quality code with confidence. A build process starts once Git commits are pushed. With each addition of code, it becomes imperative to validate the project's test suite. Initially consisting of a few hundred unit tests, the test suite has now grown significantly to approximately 13,000 tests. As the number of unit tests increased, it became essential to devise a more efficient approach for determining which tests need to be rerun, instead of executing all the unit tests indiscriminately. The build process was divided into two main types: - Build of the Main branch. - Build of a Feature Branch (any branch other than the Main branch). For the Main branch, the process remains straightforward. When a pull request (PR) is merged into the Main branch, it triggers a build that runs all the existing unit tests as a precautionary measure. However, for Feature Branches, a new build is initiated when changes are pushed to Git. In this scenario, the selection of tests to run depends on the changed files from the new commits. The tests-to-run are divided into batches, which are then executed in parallel on many machines. The selection of unit tests for each batch is based on their duration, ensuring the optimal allocation of resources and minimizing the overall time required to complete the unit test runs. In the past, the unit test selection was done at the module level. If a change was detected in a file within Module A, all tests from modules dependent on Module A were executed. However, as the number of tests continued to grow, necessitating more resources, we recognized the need for process optimization. Our solution, known in Taboola as the Achilles Project, focuses on selecting unit tests to run at a finer resolution—specifically at the class level. Rather than considering the module to which a Java class belongs, we now examine the unit tests from all modules that depend on the modified class and execute only those relevant tests. It has saved us around 3 minutes from an average build time of 23 minutes. The solution: The Achilles Project comprises two parts: data collection and determining the tests to run during the build process. For Main builds we apply only the first part, collecting the tests data, and run all the tests. We use the Main data as a baseline. ## Data Collection To collect the necessary data, we employed a Java Agent. Essentially, a Java Agent is a specially crafted JAR file that utilizes the Instrumentation API provided by the JVM to modify the bytecode loaded in the JVM. The java agent gives the ability to collect loaded classes while running unit tests. Then, we implemented a listener that extends org.junit.runner.notification.RunListener, which listens to each running test. Once a test finishes, the listener utilizes Java Instrumentation to collect the loaded classes for that test. The data is then saved as a map of tests to loaded classes. After running all the tests, we gather all the test data (list of tests -> loaded classes) and transform it into a list of classes with associated tests to run. This information is stored as a JSON file in our database (HDFS in our case). Consequently, we can now select which tests to run based on the stored data in the HDFS, eliminating the need to execute all tests. ## Determining the Tests to Run During any Feature Branch build, a POST request is sent to "Achilles server", with three parameters: Current branch name, changed files, and closest Main commit. The Achilles server retrieves older data saved for the branch, and for the given Main commit. A record of data is a json file containing a map of: class name to list of tests to run. Achilles takes the data from the two records, merges it, and returns the tests-to-run according to the given changed files. ## Saving As of the time this was written, 55.3% of the tests are avoided by Achilles, something like 7150 from 13,000 unit tests. Relative to the module-level solution. In minutes - it saved us ~3 minutes from build time of ~23 minutes. A significant improvement. Note: During this article, our focus has primarily been on Java classes. However, the Achilles Project also provides the capability to track non-Java files using Java ClassLoader. This extends the versatility of our solution, allowing us to effectively handle changes and tests associated with various file types beyond Java classes. Credits: Alon Pilberg - Project lead, and Maria Saleh Naser. --- ### Recursion vs Generators: Which One is Better for Splitting Problems in JavaScript? URL: https://www.taboola.com/engineering/recursion-vs-generators-which-one-is-better-for-splitting-problems-in-javascript/ Last Modified: 2025-01-14 14:37:15 One of the day to day trade-offs that we are forced to make when writing code, is balancing between performance, efficiency and readability. For example, writing object-oriented code may be good for readability, but at the same time it may reduce our ability to minify the code and reduce code size. One of the ways to improve code readability and maintainability, is to split it into small chunks that are easy to understand. This can be done by splitting the code into small functions with meaningful names, and modularization of the code. Often we encounter problems that can be solved by recursion - assuming that the solution already exists for part of the problem and proceeding from there. This approach often results in shorter and more readable code. However, it has its drawbacks, especially when there are many stages - it is not memory efficient, and it creates a stack of function calls that blocks the main thread and can lead to stack overflow. In this blog post I will review this approach, and suggest a different way to handle similar problems using generators. ## Recursion: Breaking Down Problems into Smaller Sub-Problems Recursion is a term that originates from the field of Mathematics. It comes from the Latin verb recurro, that means to run back. The meaning is to define a problem in the terms of itself - for example, if you would like to calculate the nth fibonacci number, it is easy to describe it as the sum of the n-1 fibonacci number and the n-2 fibonacci number. script function* fibonacci(n) { let cur = 0, next = 1; i = 0; while (i < n) { yield cur; = ; i++ } } function main() { for (const value of fibonacci(40000)) { console.log("value", value); } } main(); /script In the following example I'm using memoization in order to cache the calculation result and make the code more efficient. script const mmz = let res = 0 function fibonnachi(n) { if (mmz.length > n) { return mmz } mmz.push(fibonnachi(n-1) + fibonnachi(n-2)); return mmz; } console.log(fibonnachi(40000)); script This approach has several advantages: - Recursion can make code more readable and elegant, as it allows for a more natural expression of the problem solution. - Recursive algorithms can be simpler and shorter than their iterative counterparts. - Some problems are naturally recursive, and can be more easily expressed using recursion. - Recursive functions are good for solving problems that have a recursive structure, like tree traversal and graph traversal. However, it also has significant drawbacks: - Recursive functions can be harder to debug and understand than iterative functions. - Recursive functions can cause a stack overflow if the recursion goes too deep (the example above does). TCO, or rather, Tail Call Elimination in JavaScript is an optimization that would enable developers to use recursion without causing stack overflow - as long as the function only calls itself from the return statement, and all of the information that is needed with it. script function fibonacci(n) { if(n < 2) { return n; } else { return fibonacci(n-1) + fibonacci(n - 2); } } console.log(fibonnachi(40000)); /script This optimization was originally part of ES2015, but unfortunately it was abandoned and currently it is only supported by Safari. Also, recursion isn't a good fit for every problem - some are more efficiently solved using iteration. ## Generators: Splitting Problems with Improved Page Performance Generators, AKA ES6 generators, are another technique that allows you to solve a problem by breaking it down into smaller sub-problems, but they offer some advantages over recursion. They are a type of function that can be paused and resumed multiple times. They allow you to create iterator objects, which can be used to loop over a set of data or a sequence of operations, without the need to load all the data into memory at once. A generator function is defined using the function* syntax, and inside the function, the yield keyword is used to pause the function and return a value. Each time the generator's next() method is called, the function resumes from where it left off and continues to execute until it encounters another yield statement or the end of the function. Generators are commonly used for creating iterators, working with asynchronous code and implementing features like lazy evaluation. They also provide a way to create infinite sequences, which can be useful in some cases. They are supported by the latest versions of all common browsers except IE and Opera Mini. For general support, use regenerator-runtime npm package (1.7kb). Here is an implementation of the fibonacci number calculator using a generator: script function fib(n, sum=0, prev=1) { if (n <= 1) return sum; return fib(n-1, prev+sum, sum); } console.log(fib(40000)); /script This example does not cause stack overflow, however, it does cause a long task: In the following example I use scheduler.postTask to yield into the main thread and avoid long tasks: script async function* fibonacci(n) { let a = 0, b = 1; i = 0; while (i a); = ; i++ } } async function main() { for await (const value of fibonacci(40000)) { console.log("value", value); } } main().catch((e) => console.error(e)); /script This code does not cause stack overflow, and if we create a recording in Chrome's performance tab, we can see that there are no long tasks: The scheduler API is not yet supported by Firefox and Safari. You can use scheduler-polyfill (62.4kb unpacked), or simply setTimeout(0). This approach completely removes the limit on the maximum number of steps in the calculation, and opens exciting new possibilities. For example, we can have a function that prints all of the fibonacci numbers without any exit condition. When running the following code, the page continues to be responsive, and the only limitation is that at some point the returned value settles on “Infinity". script async function* fibonacci(n) { let cur = 0, next = 1; i = 0; while (true) { yield scheduler.postTask(()=>cur); = ; i++ } } async function main() { for await (const value of fibonacci()) { console.log("value", value); } } main().catch((e) => console.error(e)); script ## Conclusion JavaScript generators offer a powerful technique to solve recursive problems. They are more efficient and can be used to implement lazy evaluation. While for small problems with few stages recursion provides a solution that requires less code to be written, for larger problems I encourage you to consider generators. ## References - https://users.cs.utah.edu/~germain/PPS/Topics/recursion.html - https://developer.mozilla.org/en-US/docs/Web/JavaScript/Guide/Iterators_and_generators - https://medium.com/hackernoon/es6-tail-call-optimization-43f545d2f68b - https://www.npmjs.com/package/scheduler-polyfill - https://www.npmjs.com/package/regenerator-runtime - https://stackoverflow.com/questions/54719548/tail-call-optimization-implementation-in-javascript-engines I'd like to express my gratitude to David Hakak, Noam Rosenthal, John Reilley and Jamie McCrindle for proofreading my post and providing useful suggestions. I got help from Chat GPT with the structure and editing of this blog post. --- ### GPU Integration Propels Data Center Efficiency and Cost Savings for Taboola URL: https://www.taboola.com/engineering/gpu-integration-propels-data-center-efficiency-and-cost-savings-for-taboola/ Last Modified: 2025-01-14 14:37:15 A year ago, we at Taboola kicked off a POC to migrate our Spark workload from the thousands of CPU cores cluster to the GPU. I started with a small test, just to get a hunch of what it would require and very quickly found myself in an intense project involving a great group of people from Taboola and NVIDIA's Rapids team. In this blog post, I'll describe what Taboola does with Spark, our motivation to move to Rapids and the insights, pitfalls, challenges and achievements so far. ## Taboola's Business Taboola is the leading content recommendation company in the world. When you surf the web and see a native advertisement, it's most likely a content served by Taboola. In order to do so, Taboola uses a very complex data pipeline that stretches from the user's browser or phone, through multiple data centers running complex Deep learning algorithms, databases, infrastructure services such as kafka and thousands of servers in order to serve the best fitting ad for the user. This blog will focus on one component in this complex pipeline - our thousands of CPU cores spark cluster and our effort to migrate it to the GPU. Data to this cluster arrives from data centers all over the world where we collect and build a unique “page view". A pageview is a very big and wide data structure identifying each user and its interaction with our system. This pageview structure, containing over 1500 distinct columns and amounts to over 1TB of hourly data, is what we process in our Spark cluster. ## Project Motivation Motivation originated from the fact that the CPU cluster scale out is very challenging in terms of both hardware costs and data center capacity. In order to cope with the increasing load of data needed to be processed, Taboola is required to increase its Spark cluster capacity quite often. We have many distinct analyzers, SQL queries, that process the incoming pageviews, 1TB of raw data, every hour with a 2, 6, 12 and 48 catchup runs. New analyzers are being created all the time and increase the load on the Spark cluster, hence the constant need for more compute power. NVIDIA's Rapids accelerator for Spark was a perfect match. ## It is POC Time We first defined what we are going to test. We took real production data from “Cyber Monday" so that we'd test and benchmark a very big dataset. The data is 1.5TB of ZSTD compressed parquet files, per hour. It has over 1500 columns, of all native types including arrays, structures, nested structures with arrays, the full monty. Hardware wise we started with a 72 CPU cores Intel server with 3 A30 GPUs, 900 GB local SSD drive for Spark to store its intermediate files, 380GB RAM and a 10Gb/s NIC card. We set a minimum bar of x3 factor so that the GPU solution would be considered a successful one cost wise. We picked 15 queries from production from multiple R&D departments that would resemble as many of the hundreds of queries we have in production. These queries are mostly very complex including many SQL operations such as: aggregations, sorts, lateral view explode, distribute by, window functions and UDFS. Figure-1 shows such query. Figure-2 gives you a sneak peek as to the factors we got. ## POC Goals We started with a single server as described above, however obviously we'd want it to scale to a multi-GPU and multi server cluster. The cluster would be managed by Kubernetes, as opposed to our current Mesos cluster. Mesos is going to be obsolete and NVIDIA supports Kubernetes. Taboola's R&D should be oblivious to the change and should not care whether their queries run on CPUs or GPUs. GPU output should be the same as the CPU, a task which is sometimes challenging when migrating to the GPU, but one that the Spark Rapids team is very aware of. Stability in production is also a crucial goal.Lastly, the GPU should outperform the CPU by a minimum factor of 3. We benchmarked several GPUs, P100s, V100s, A100s and A30 and found out that the A30 was the best price performance fit. ## Run Rapids Run I ran some SQL queries using Rapids and the results were a bit disappointing. Some of the less complex queries, mostly with lateral view explode, gave a x3 - x5 factor over the CPU. Some showed much lower factors while some queries even crashed. I started to ask around in Rapid's github repo, played with the relevant Spark and Rapid's specific parameters and started to get better results. The relevant parameters were: - sql.files.maxPartitionBytes - The CPU uses a default value of 128MB, for the GPU it is too low. We are using 1-2GB. - sql.shuffle.partitions - We found the 200 default to be good enough in most cases. - rapids.sql.concurrentGpuTasks - Determines the number of tasks that can be run concurrently on the GPU. You should try to use at least two. Tuning these parameters helped the queries to run more smoothly and perform better in some cases. ## First Bottleneck So what is holding us back? We've profiled some of the less performant SQLs and saw that most of them wasted a lot of time while parsing Parquet's footer data on the CPU. Our Parquet data has more than 1500 columns and apparently the regular Java code that is parsing the footer was not adequate for such a big footer. Figure-3 shows a snippet of NVIDIA's profiler output showing a 9 seconds Spark task, where the GPU was mostly idle and only worked for 330ms. Figure-4 shows a flamegraph of one of our queries that suffered from this behavior. The purple bars indicate time spent inside the org.apache.parquet.hadoop.ParquetFileReader class. Almost 50% of this query time was spent on parsing Parquet's footer, during that time the GPU was idle. We set off to test an idea we had. When parsing the footer, Parquet's code would iterate over the footer metadata serially for each row group. We made some adjustments to Parquet's parameters and decreased the number of row groups we had in each file. That gave us about 10-15% improvement but obviously was not enough. Remember that we have 1500 columns so each time a footer metadata is read, even though we're only asking for 50-100 columns per query, the entire 1500 metadata would be read and parsed serially. We wanted to index the footer metadata so that instead of reading the entire 1500 columns data serially, we'd just access it directly. We managed to pull this off by changing Parquet-mr public code in C++ and Java and did get nice performance results, however it was too cumbersome and complex. Luckily for us, the Rapids team at NVIDIA came with a much better idea and replaced the Java code with Arrow's C++ implementation. We now have the rapids.sql.format.parquet.reader.footer.type set to NATIVE by default for our GPU implementation. Bottleneck resolved. No more queries with the GPU idle because of footer parsing overheads on the CPU. ## Network Bottleneck The next bottleneck was caused by the fact that the network card was too weak. While the 10Gb/s ethernet card sustained the CPU load, it failed to do so for the GPU load. Replacing it with a 25Gb/s card, resolved this bottleneck. ## Disk I/O Bottleneck So two bottlenecks resolved, queries still run slow, now what? Looking in Spark's UI page, gave a clear indication as to what is happening now. See the following table.    Metric  Min 25th percentile  Median 75th percentile Max  Duration  0.4 s  0.6 s  0.8 s  1 s  1.2 min  GC Time  0.0 ms 0.0 ms  0.0 ms  90.0 ms  0.5 ms  Shuffle Read Size/Records  21.4 MB/1000  22.3 MB/1000  22.5 MB/1000  22.7 MB/1000  27.3 MB/1000  Shuffle Write Size/Records  17.5 MB/1000  17.9 MB/1000  18 MB/1000  18.1 MB/1000  18.1 MB/1000  Scheduler Delay  3.0 ms  5.0 ms  5.0 ms  7.0 ms  3 ms  Peak Execution Memory  64 MB  64 MB  64 MB  64 MB  64 MB  Shuffle Write Time  9.0 ms 13.0 ms 18.0 ms  21.0 ms 59 s As can be seen above in the Max column, the task's duration is 1.2 mins, while Shuffle Write Time is taking 58 seconds. Apparently we waste a lot of time doing shuffle work while the GPU is idle again. Figure-6 shows the appropriate event timeline graph. The orange part is shuffle times, read or write. The green parts are compute time. We're wasting a lot of time reading or writing shuffle files.Our shuffle files can get up to 500GB and higher in some queries. Obviously we can't keep this huge amount of data in the server's RAM so the shuffle files are stored in the local SSD drive. After a quick investigation with our K8s and IT teams, we figured out that the SSD drive was configured to use RAID-1, i.e. each temporary shuffle file was saved twice to the disk. This is absolutely a waste of time. Switching to RAID-0 somewhat improved the situation. The next thing was to switch to a 6TB NVME drive. That nailed it. We did have to take one of the GPUs out so that the NVME can be used but from that point we did not have any further performance issues with shuffle read or write. We also found out that, for our workload, one such NVME is capable of sustaining the load of two A30 GPUs. As with the network card bottleneck, the GPU put much more pressure on the SSD drive than the CPU to the point that we had to replace the SSD with a NVME disk.What about K8s?The move to K8s from a stand alone POC machine involved a lot of configuration work and small details but was quite straightforward. The basic idea is that Spark's driver would sit on a non-GPU machine and each K8s POD would be associated with a single GPU. A “Getting Started with Rapids and Kubernetes" tutorial can be found here. Figure-7 shows some of the major K8s relevant configurations.How fast is the GPU?The most interesting question when migrating a project from the CPU to the GPU is usually: What's the factor? For a real world cluster with multiple GPUs the answer would be: How many GPUs do I need to sustain a load manageable by my X CPU cores. That's your factor.Current StatusWe set up a system with two A30 GPUs and streamed the production data to it in parallel with our real big CPU cores production environment. Figure-8 shows two of our heaviest queries running in the production CPU cluster and on the server with the two A30 GPUs.The yellow and green lines are the hourly total time across all tasks numbers of the two queries running on the CPU cluster. The blue and orange lines are the same queries running on the GPU server. GPU factors, for total time across all tasks, are x20 and above. Figure-9 shows a zoom in view of the GPU runs. You can see that they behave similarly to the CPU in that the graph behaviors are roughly the same.The next interesting graph shows the factor of all the queries we've migrated to the GPU and their counterparts on the CPU. The GPU run is missing the biggest query, which we are still working on migrating to the GPU. It should probably add another 200 hours per day to the GPU total time. See Figure-10 to see the daily accumulated factors, they are aligned with the x20 factor we've seen in Figure-9 as well.While we emphasize “Total time across all tasks" in the above images and text, as we go deeper and deeper into migrating more analyzers to the GPU, we've realized that a more important and accurate metric is the “duration" metric. In order to better evaluate the expected performance boost from the GPU, we've set a test cluster with 100 CPU cores, ran our queries and compared the duration of each environment for each query. The end game here is to be able to say that 100 CPU cores work took 4 minutes, for example, while one A30 GPU finished the same work after 2 minutes. That would indicate that one A30 GPU performs the same as 200 CPU cores, as a matter of fact, most of our analyzers showed this performance ratio in production.What's next?Next would be to move more queries from other R&D departments to the GPU, which will result in increasing the number of GPUs in production. We've already on the verge of migrating another group of 15 heavy queries to the GPU. Soon we're also getting dozens more A30 GPUs to join our GPU cluster and enable us to better cope with current and near future load. QA and monitor the system more closely while in production.ConclusionThis has been an amazing joy ride effort during which I got acquainted with Rapids. Learn all sorts of aspects that I haven't had too much interaction with so far such as Parquet internal stuff. We had to Identify and cope with hardware limitations and push the GPU to its edge. The results, however, are extremely rewarding. This post, in another version of it, was also published by NVIDIA on their own blog site and can be found here: https://developer.nvidia.com/blog/gpu-integration-propels-data-center-efficiency-and-cost-savings-for-taboola/AcknowledgmentsThis huge effort could not have been successful without the support, assistance and patience of two great groups of people. Taboola: Andrey Gourine, Gilad Zamoscinski, Igor Berman, Keren Corsia, Lior Chaga and Michael Taranov. NVIDIA's Rapids team: Alessandro Bellina, Hao Zhu, Karthikeyan Rajendran, Robert Evans, Sameer Raheja --- ### Our Journey of Virtualization Change URL: https://www.taboola.com/engineering/our-journey-of-virtualization-change/ Last Modified: 2025-01-14 14:37:15 ## The Dark Ages We initially adopted oVirt as our virtualization platform, and it proved to be a good product with several notable advantages. Its open-source nature allowed us to leverage a wide range of features and customization options. However, despite its strengths, we encountered several downsides and problems that compelled us to seek a better virtualization solution. Two major drawbacks were that it didn't have any working DFS and inventory management issues. Additionally, we faced occasional performance issues and stability concerns with oVirt. Some resource-intensive workloads experienced latency or unexpected behavior, impacting the overall performance of our virtualized environment. These problems became more pronounced as our infrastructure grew, leading to a decrease in productivity and user satisfaction. As you understand, we needed to give a better solution to our R&D teams and an easier/better product for us to manage and maintain. (Source) ## ## KubeVirt to the rescue KubeVirt is an open-source virtualization solution that enables running virtual machines (VMs) on top of Kubernetes clusters. It allows users to leverage the benefits of both containers and VMs, making it easier to manage both traditional and cloud-native workloads in a single platform. KubeVirt uses a custom resource definition (CRD) to define VMs as Kubernetes objects, making them first-class citizens in the Kubernetes ecosystem. This means that VMs can be managed using Kubernetes-native tools and APIs, making them easier to automate and orchestrate. KubeVirt also provides a virtualization API that allows for the creation of virtual devices and drivers, giving users the ability to configure and customize their VMs to suit their specific needs. Also, KubeVirt integrates with popular virtualization technologies, such as QEMU and libvirt, making it possible to use existing VM images and templates with Kubernetes (This made our life easier to migrate from Ovirt). Overall, KubeVirt provides a powerful and flexible solution for managing virtual workloads in Kubernetes, allowing for a more streamlined and efficient approach to managing hybrid workloads. ## ## KubeVirt vs ???? KubeVirt provides several benefits over other products in the virtualization space. Here are some of the benefits of KubeVirt: 1. Seamless integration with Kubernetes: KubeVirt is designed to seamlessly integrate with Kubernetes, allowing virtual machines to be managed as Kubernetes objects. This means that users can leverage their existing Kubernetes skills and tools to manage virtual machines, making it easier to manage hybrid workloads. 2. Customizable virtualization: KubeVirt provides a virtualization API that allows for the creation of custom virtual devices and drivers. This means that users can customize their virtual machines to meet their specific needs, making it easier to run legacy workloads(ahm ahm ovirt) on modern infrastructure. 3. Open-source and community-driven: KubeVirt is an open-source project with an active community of contributors. This means that we can benefit from a wide range of features and integrations contributed by the community, as well as access to support and resources. 4. Flexibility: KubeVirt provides flexibility in terms of the virtualization technology used, as it can integrate with popular technologies such as QEMU and libvirt. This means that users can leverage existing virtual machine images and templates with Kubernetes, making it easier to migrate workloads to Kubernetes(VMs migration). In summary, KubeVirt provides a powerful and flexible solution for managing virtual workloads in Kubernetes, with seamless integration, customized, open-source community support, flexibility in virtualization technology, and efficiency in resource utilization, making it a very strong competitor in the virtualization space. ## ## It's all in the Fog After deciding on the solution, what about the design? Let's start from our needs: - Multi-regional DCs - Management cluster - Easy maintenance   We have multi-regional data centers, which means the solution should be designed to work seamlessly across multiple geographic locations. This means deploying Kubevirt clusters in each region. The management cluster should be used to manage multiple Kubevirt clusters located in different regions, providing a centralized control plane for the entire virtualization infrastructure. The Kubevirt solution should be designed with maintenance in mind. This could involve using tools like Kubernetes operators to automate common maintenance tasks, or designing the infrastructure in a way that allows for easy upgrades and updates without disrupting service availability. ## Our General Stack Won't go into details here, as there are very good articles regarding our stack: - Calico - A networking solution for Kubernetes that provides network policy enforcement, secure network communication, and network isolation. It uses BGP routing and can be integrated with other network plugins. - Rook-Ceph - A storage solution for Kubernetes that uses the Ceph distributed storage system. It provides persistent storage for Kubernetes workloads and can be used to store block, file, and object data. - Velero - A backup and restore solution for Kubernetes that can be used to protect and migrate Kubernetes workloads and resources. It can be used to create and restore backups of entire clusters or individual resources. - VictoriaMetrics - A time-series database and monitoring solution for Kubernetes that can be used to collect, store, and analyze metrics and logs from Kubernetes workloads and resources. It supports a variety of data sources and can be used to create custom dashboards and alerts. - CDK8s (Cloud Development Kit for Kubernetes) - an open-source software development framework that enables developers to define Kubernetes resources using familiar programming languages like TypeScript, Python, and Java. - Argo Events is an open-source event-based system that allows you to trigger actions in response to specific events. ## Fries and DNS Delegations In our multi-region environment, the DNS architecture consists of a main DNS server and fog nameservers for each zone. The main DNS server handles different types of DNS queries like PTR, CNAME, and A records, while the fog nameservers are responsible for handling DNS queries for a particular zone. Kubernetes uses kube-dns as its default DNS server. However, we've decided to use CoreDNS. CoreDNS is a flexible and extensible DNS server that is designed to work well with Kubernetes. It supports various plugins and middleware that can be used to customize its behavior to suit your needs. When a DNS query is made related to a Kubernetes cluster, the main DNS server delegates the query to the relevant fog nameserver that is responsible for handling DNS queries for that particular Kubernetes cluster. The fog nameserver then uses CoreDNS to resolve the service name to an IP address. (Snippet from coredns config) Managing DNS in a multi-region environment with Kubernetes and CoreDNS can be challenging, but it provides several advantages over traditional DNS architectures. By using CoreDNS as the DNS server in Kubernetes and delegating DNS queries to the appropriate fog nameserver, you can simplify your DNS architecture, improve reliability and availability, and customize DNS behavior to suit your specific needs. ## ## Fun, Games and Migrations Migrating from oVirt to KubeVirt involves moving virtual machines from one platform to another. The migration process requires careful planning, as it involves evaluating the compatibility of the VMs with KubeVirt, and ensuring that the necessary tools and configurations are in place, and that Ovirt is a production system. In order to streamline the process, we've created a custom script that automates the migration of VMs from oVirt to KubeVirt. The script takes into account our specific needs, such as integration with Foreman, templates and ensures a smooth and efficient migration process. (Snippet from migration script) ## ## Continuous Deployment? Challenge accepted It's all fun and games, but how do we manage this monster? ArgoCD to the rescue! With the growing adoption of Kubernetes, many CD systems have emerged to support containerized workloads. Among these systems is ArgoCD, an open-source project that provides GitOps-based continuous delivery for Kubernetes. ArgoCD offers several benefits over other CD systems, including: - GitOps: ArgoCD uses a GitOps-based approach to manage Kubernetes resources, which means that all changes are committed to a Git repository and automatically deployed to the cluster. This approach provides a single source of truth for the entire infrastructure, making it easier to track changes, rollbacks, and security compliance. - Declarative Configuration: ArgoCD allows users to define the desired state of the Kubernetes resources using YAML files, enabling declarative configuration management. This approach simplifies the management of complex Kubernetes environments, making it easier to manage configurations across multiple clusters. - Multi-Cluster Support: ArgoCD provides a single control plane for managing multiple clusters, making it easier to deploy applications across multiple environments. This feature is particularly useful for organizations with a large number of clusters, as it simplifies the management of resources across multiple Kubernetes clusters. - Advanced Rollback Capabilities: ArgoCD supports advanced rollback capabilities, enabling users to easily rollback to previous versions of the application or infrastructure. This feature is critical in ensuring high availability and minimizing downtime. - Scalability: ArgoCD is highly scalable, making it ideal for managing large and complex Kubernetes environments. It is designed to work with Kubernetes-native tools like Helm, Kustomize, and Jsonnet, enabling users to customize the deployment process as needed.   At our organization, we use ArgoCD to manage our KubeVirt environment and infrastructure, which includes Calico, Rook-Ceph, Velero, Victoria Metrics, and CDK8s. With ArgoCD, we can manage all our Kubernetes resources from a single control plane, enabling us to deploy applications rapidly and at scale. We can also use ArgoCD's declarative configuration management to simplify the management of complex Kubernetes environments, making it easier to manage configurations across multiple clusters. In addition, ArgoCD's advanced rollback capabilities and multi-cluster support ensure that we can maintain high availability and minimize downtime in the event of an issue. And with ArgoCD's scalability, we can easily manage our growing infrastructure without compromising performance. In conclusion, ArgoCD provides several benefits over other CD systems, making it an ideal choice for managing our multi region environments and infrastructure. Its GitOps-based approach, declarative configuration management, multi-cluster support, advanced rollback capabilities, and scalability make it a powerful tool for deploying applications rapidly and at scale. (Source) ## ## Events? What events? If you were reading closely, we are using foreman as our inventory management. Argo Events is an open-source event-based system that allows you to trigger actions in response to specific events. It enables you to easily define, route, and filter events using a declarative syntax. With Argo Events, you can create event-driven workflows that can perform various tasks, such as running jobs, deploying applications, or sending notifications. Our use case of Argo Events is deleting foreman objects.We are using sensors to listen to events and automatically delete Foreman objects when a specific event occurs. For example, when a host is deleted, Argo Events can trigger a workflow to delete the corresponding Foreman object, which helps to keep the Foreman inventory up-to-date and clean. Overall, Argo Events provides a powerful and flexible way to automate tasks and workflows based on events, which can greatly improve efficiency and reduce manual effort. (Snippet from argo events hook) ## ## Friends and templates Customizing Kubernetes objects is a crucial task in managing Kubernetes infrastructure. Kustomize has been a popular tool for this purpose, but it has its limitations. In particular, as projects grow in complexity, the resulting YAML files can become unmanageably large and difficult to work with. This can lead to mistakes and inconsistencies in the deployment process. To address this issue, many teams have turned to cdk8s as an alternative to Kustomize. cdk8s is a framework that allows us to define Kubernetes resources using familiar programming languages like Python for objects using a more flexible and modular approach. It allows for the creation of smaller, more manageable YAML files that are easier to read and edit. While cdk8s offers many benefits, one downside is that it is not officially supported by ArgoCD, to overcome this limitation, we've created a custom docker image to ensure compatibility between cdk8s and ArgoCD. (snippet from cdk8s configmap) By doing so, we can continue to benefit from the advantages of cdk8s while also ensuring that our deployment process is reliable and consistent. Now, adding VM's is easy as 1-2-3: (snippet from vm's creation with python cdk8s) This is it for now, Wanted to thank my awesome team, that without them it wouldn't be possible, Alex Bulatov Gal Beniluz Tarek Shama Maher Odeh. --- ### How I Learned To Stop Worrying And Love The Kubernetes Release Cycle URL: https://www.taboola.com/engineering/how-i-learned-to-stop-worrying-and-love-the-kubernetes-release-cycle/ Last Modified: 2025-01-14 14:37:16 ## Our Story and Humble Beginnings Taboola, a company that has more than 500 million daily active users, has been an early adopter of Kubernetes on a massive scale. In the past, the company used to run its Kubernetes clusters using Systemd-based services and deployed them using Puppet as the configuration management tool. However, as Kubernetes releases became more frequent, we found keeping up with the latest updates and new features challenging. Before Kubernetes, we used Nomad clusters to run our containerized applications. However, we quickly realized we needed a more agile and scalable infrastructure. We then turned to Kubernetes and used the "Kubernetes the hard way" project, which was written by Kelsey Hightower from Google, to initiate our clusters. ## Scaling Up Infrastructure: Challenges with Upgrades and Modifications As the company's infrastructure grew, we found modifying and upgrading using Puppet increasingly difficult. The company's infrastructure consisted of 7 large-scale Kubernetes clusters across nine on-premise data centers and tens of thousands of cores. The Puppet mechanism, which works by representing the state of the infrastructure and modifying it to match the desired state, created challenges for the platform team and the service level it provides as it could not perform gradual modifications on the cluster without causing downtime. Consequently, the team accumulated significant technical debt, and we could not keep up with the fast-paced Kubernetes release cycle. We were running an outdated version of Kubernetes while newer versions were already available. The Kubernetes release cycle has also undergone a significant change in recent times. In April 2021, the open-source Kubernetes release team shifted from four to three releases per year. This move ensured a more predictable release schedule and improved software release quality. Despite this change, it's still crucial for companies to keep up with the latest updates and new features in Kubernetes, as it's a rapidly evolving technology that requires continuous learning and adaptation. ## Revolutionizing Infrastructure Management: Our Journey from the Hard Way to Kubeadm To overcome these challenges, we designed a solution to allow us to initiate and re-initiate clusters as quickly as possible. We moved away from Puppet and adopted a more agile and scalable infrastructure management approach. The Kubeadm, a tool to initiate and manage Kubernetes clusters, had become stable and been increasingly adopted. The team created a new development cluster separated from the production infrastructure for testing and experimentation. This approach allowed us to build, break, and rebuild faster. It helped us catch up with the latest version of Kubernetes. ## Upgrades Management Process After restructuring our infrastructure, upgrades and maintenance work became vastly more manageable. We created a process to apply upgrades on our clusters with zero downtime for production services. We divided the process into three steps: - First, we gather and decide on all the new software versions that construct our infrastructure. - Second, we build and test the upgrade process with our development cluster. - Third, we schedule and upgrade our production clusters. The first step is to plan the upgrade process. It is more of an engineering job than an execution job. We list all the infrastructure applications that construct our Kubernetes platform and check the next stable and LTS version for them. We never take the latest and greatest because we prioritize stability over features. We check the compatibility of those applications with the new API versions. We also check what versions are highly adopted on other large cloud providers. In every Kubernetes version update, we must upgrade essential infrastructure applications like the Calico CNI, Rook, Istio and more, which is a challenging upgrade process to plan. We test every procedure we plan on the development cluster. We pick versions and then simulate the entire process. After that, we have an idea of what will happen in production. Usually, we need to tune versions due to incompatibility, or some applications already have new minor versions by the time we simulate. We run the process dozens of times until we perfect the technique, And only then do we schedule a maintenance day to apply on production. We call it the "Kung Fu" method, and it works. ## Automated Upgrade Service Once we start to upgrade, we actively operate and monitor upgrades of the control plane nodes, as they are inherently more sensitive than regular worker nodes. We automated the worker nodes upgrade process for the rest of the cluster with a service we wrote in Golang. We designed the service to manage the entire process from start to finish. To do it, we set the following goals in mind: a. Run the upgrade process in the background without human intervention. b. Zero downtime for the services that run on the platform. The service will continuously monitor the cluster's health and resources. It will pause if there are insufficient resources to operate the cluster or if the cluster health gets degraded. We wanted the service to have similar functionality to the rolling deployment mechanism of Kubernetes in the sense that when the upgrade is stuck, or the cluster is unhealthy, it will cease to continue. When the cluster comes back to a healthy state, it will resume the upgrade. The upgrade service can be configured with options like "max surge" and "max unavailable" as with the rolling deployment feature of Kubernetes. The service can also be configured with options like: upgrade one or multiple nodes simultaneously, the number of nodes that can be operated at once, the number of failed nodes that will halt the entire upgrade process, the waiting time between operations, and more. The service has two major components: a controller and an executor. The controller monitors the cluster and decides which nodes to upgrade. The controller marks those nodes, drains them, and initiates a single executor for each marked node to operate. The executor is a separate unit from the controller and can be replaced with different functions. We used the Ansible configuration management tool for the upgrade process, as it required running Linux commands and the Kubeadm tool commands. The executor is a Pod launched via a Job manifest by the controller. The image of the Pod contains Ansible installed on a tiny Linux distribution. The entrypoint runs an Ansible playbook to upgrade the node to a newer version. The executor can be used dynamically to run different playbooks on demand. We used Kubernetes ConfigMap as a mounted volume to pass the playbook file into the container. The executor is scheduled on one of the control-plane nodes to avoid conflicting with a node in the process of an upgrade. If the executor fails, the Job will reschedule the pod until it succeeds or has exhausted all attempts. The upgrade playbook covers all operations needed to upgrade a node, including installing new packages, modifying configurations, and restarting appropriate services. After the executor finishes upgrading a node, the controller runs tests to ensure proper functionality. If successful, the node will be returned to the cluster. The controller continues upgrading nodes until the cluster becomes unhealthy or all nodes are successfully upgraded. The executor can be replaced with other executables, making it useful for additional operations. ## Automated Upgrades for Error-Free Large-Scale Deployments One of the benefits of using an automated upgrade service is the reduction in likelihood of human error, which is especially important in large-scale deployments like Taboola's. With an automatic process, the team can be confident that each node is upgraded correctly and that the process is consistent across all nodes. Another benefit is that the team can save time and resources. Without the automated service, upgrading worker nodes would require manual labor, which is time-consuming and error-prone. With the service, the team can focus on other tasks while the upgrade is performed. ## Danger, Robot in Action Implementing an automated upgrade service has its challenges. One challenge is that the service must be carefully designed and tested to ensure it works reliably and efficiently. The team needs to consider various factors, such as the cluster's size, the available resources, and the workloads running on the cluster. The service serves us well today, and we use it often. For example, we also used the service for a "docker sunset" project. The project aims to replace the Docker runtime with the Containerd runtime on our production clusters. Initially, we were only installing Containerd on new nodes joining the cluster. However, it had to be replaced after Docker became deprecated as a container runtime in the Kubernetes ecosystem, So we used our service to ramp up the process of shifting to the new runtime app. We designed an executor that applies a reinstall function on nodes to reinstall them as Containerd machines. The result is a faster and more controlled process by using the service. Bottom line, by using Kubeadm and creating a separate development cluster, we were able to overcome our challenges of keeping up with the fast-paced Kubernetes release cycle. Adopting a more agile and scalable infrastructure management approach is crucial for companies that use Kubernetes on a massive scale like Taboola. With our upgrading processes in place, we can keep everything up to date and promote new features for our R&D teams. --- ### Insights From My First Year as Integration & Support Team Lead at Taboola URL: https://www.taboola.com/engineering/my-first-year-as-an-advertiser-integration-performance-experts-team-lead-at-taboola/ Last Modified: 2025-01-14 14:37:16 I never saw myself as the managerial type. Probably due to the manager typecast I saw while I was serving in the military and previous workplaces. Where a manager has to be assertive, loud and know how to “get things done." Therefore, when I was offered my first leadership role after 2.5 years at Taboola (and 10 years in technical roles), my first reaction was fear. Am I a good fit for such a role? How will I handle difficult situations? Wouldn't it be awkward to manage people who up until recently were my colleagues? Then again, I have a strong belief that anything can be learned and 'how to be a good leader' is one of those things. I have accepted the offer and spent the next month consulting, reading, listening and watching materials about management and leadership. I came to realize that there are many different types of leaders, and I can choose what type I would like to be. And that qualities like patience, empathy & listening can actually be very valuable for such a role. I can be the type of leader I would like my manager to be. As I end my first year in a leadership role, it seemed like a good time to reflect over the past year, and share several insights & guidelines that help me in my day to day job and remind me what I should focus on (hint: the team members). ## When you need to make a decision, keep your team members in mind Sounds pretty generic and simple but it requires dedication, perseverance and patience. For example, - When there's a task you can complete quickly, but your team member will get value by doing it, delegate it to your team member (even if it will take longer). - Insist on a certain training/documentation that will be in your team member's local language. - When a mission fails, take responsibility Instead of blaming your team member, support him and think together what should be improved for next time. ## Act the way you would want your manager to act This is an important practice I have been trying to follow since I took on the role. A simple rule that can make it much easier to decide how to act is to think how you would like your manager to react in a similar situation. - Show appreciation when your team member is working hard - Offer help when your team members are on a dead end - Inform when you are late or canceling a meeting * If you'll be late for a meeting with the VP, obviously you would inform him/her and add an apology to it. A good rule of thumb would be to treat your team members the way you would treat the VP. ## Take a deep breath before criticizing In other words, try to step into your team member's shoes. Sometimes, your team member will fail in an assignment you expected him to succeed in. In such cases, the first reaction might be blaming the employee and thinking “how the hell did he screw it up?!" On a second thought, you might recall that the employee had some personal matters lately causing him to lose focus. Or that he handled a few other tasks this week with great results and you are judging him solely based on one incident. Before reacting, take a breath, think together with your team member what went wrong and how you can help. ## Transparency Relates to the second principle - act the way you would want your manager to act. For me personally, it is hard to be engaged on a mission I don't understand, and the rationale behind it was not clearly explained to me. Therefore, when displaying a certain mission to my team members, it is very important for me to explain first why we are doing it, what is the logic behind it and the importance of it. Not just ask them to do it “because I said so". ## Give credit One of the challenging changes to make when shifting from a technical role to a leading role is that you no longer rely only on yourself. The team's success is your success. It is ok to (and you should) give the credit to your team members and let them stand out. Seeing your team members grow, improve and succeed is one of the most satisfying feelings you get as a leader and a good indication that you are doing something right. "The leaders who get the most out of their people are the leaders who care most about their people" Simon Sinek --- ### Taboola’s shared GPU experience URL: https://www.taboola.com/engineering/taboolas-shared-gpu-experience/ Last Modified: 2025-01-14 14:37:16 What happens when you set a goal to increase Taboola's critical data processing path capacity, which is in dire need of more juice? You get a x2 performance boost! In this blog post, we'll describe how Charles from the algorithm team, Gilad - our cloudification team leader and myself from Data platform, managed to double our machine learning training pipeline throughput. Taboola, processing hundreds of TBs of daily data, has many different data processing pipelines, where disciplines from various areas usually intervene. One such data processing pipeline, which we'll describe in this blog, runs many intense machine learning algorithms on huge amounts of production data. This important pipeline is used to gain business insights into how Taboola's users and sites interact with each other. Its output later translates into improving the quality of our recommendation models, the very core of our business. In order to be able to process multiple TB of new data every hour, we make use of all types of software and hardware buzzwords, such as Kubernetes, Volcano, machine learning frameworks running on heterogeneous CPU and GPU clusters. Improving the machine learning pipeline performance directly translates to increased throughput when training our different recommendation models. This also translates into reduced hardware costs. Another very important side effect is the fact that our algorithm team can now run and test more (and different) algorithms on the same hardware, thus improving our recommendation engine AI even further. So, let's dive into the technical details of how we've accomplished just that. ## Initial performance analysis Taboola trains many of its big TensorFlow networks on a heterogeneous on-prem GPU clusters consisting of more than one hundred P100, V100 and A100 GPUs. We're re-training all the networks daily on real life, live, production data. Obviously this process takes a lot of time and resources. In order to improve the cluster's performance and optimize its work, mainly so we finish the work within the desired time frame, we decided to launch a performance analysis effort and explore the short and long term optimization opportunities we can use. The initial performance analysis showed us that the TensorFlow networks did not take full advantage of the GPUs capabilities. The GPUs were relatively underutilized as shown by the nvidia-smi tool. The initial load was around 30-40% during peak time. Profiling the network's while it trained, using NVIDIA's nvprof tool, assured us that there were some significant gaps in the timeline where the GPU was idle or not fully utilized. The empty slots in NVIDIA's nvprof output in Figure-1 shows exactly how much percentage the GPU was idle out of the whole timeline. Diving a bit deeper into the code, we saw that a certain time consuming TensorFlow op was running entirely on the CPU and not on the GPU. At this point we started to look for some low hanging fruits optimizations that we could use. After all, the idea is to get as much run time within the GPU and let the CPU manage the OS, IOPs and code path. We've noticed that the default TensorFlow implementation uses almost 100% of the GPU VRAM upon initialization. We started wondering whether allocating the entire GPU's VRAM was mandatory for the network to run and whether we could squeeze two concurrent trainings at the same time on the same GPU. It soon became clear that we could decrease the memory requirement, per trainer, to just under half of the 16 GBs installed in the P100 and V100 GPUs, using the following code fragment: We let the various trainers run using this configuration for a few days in production to verify they run as expected, don't crash and yield the correct algorithmic results. Once we were able to observe this is the case, we could move on to the next stage. The next step was to launch two different networks concurrently on each GPU, each of which taking 40% of the GPU's VRAM. This is not a “text book" solution as the two trainers will be synchronized by the NVIDIA driver and things will not really run concurrently on the GPU. However, since part of the trainer's time is spent on fetching data from a remote HDFS server, CPU ops etc, we hoped that all these components would overlap each other and we would improve the overall throughput. ## Putting it all together Sure enough, we were able to achieve performance gains (almost double the performance) using this technique. That effectively meant that we were able to run twice the amount of work on the same number of GPUs without adding new hardware as well as meeting our production deadlines of running all the algorithm team's networks within the needed time frame. Figure-2 shows how the number of trainers increased once we opened the shared GPU feature in production. The orange horizontal line shows the actual physical number of the GPUs in the cluster. The left side of the graph shows the number of trainers running concurrently in the cluster, prior to our optimization. The right side shows the number of trainers running concurrently after the shared GPU mode was enabled. As can be seen, we were able to process twice the number of trainers on the same hardware as before. Going forwards we plan to investigate which parts of the network ops can be further optimized and most significantly, which ones run on the CPU and can be migrated to the GPU. Now it was time to test it in a production-like environment to test all the code and configuration changes to see if we actually gained the performance boost. ## Cloudification magic The next thing to implement was to configure the system to run as it would in production, i.e., schedule two TensorFlow networks simultaneously on the same machine. Volcano and Taboola's cloudification team to the rescue. Taboola uses a Kubernetes cluster to manage a significant part of our CPU and GPU servers. The algorithm's team ML training takes place on the various Kubernetes jobs. The Kubernetes job controller creates a TensorFlow pod that is scheduled based on an allocatable GPU to a node. When we started looking at options for implementing the scheduling of two pods running concurrently on the same physical machine and GPU, we realized we needed some extra management layer to enable us to do so. There were a couple of available open-source tools to allow just that. The first tool we looked at was Volcano, a cloud-native batch scheduling system for compute-intensive workloads. The other option we considered was Alibaba's device plugin and Nvidia device plugin with the CUDA Time-Slicing feature. After a quick evaluation process, we decided to go with Volcano. Volcano allowed us to implement GPU memory-based scheduling differently than the standard GPU compute based scheduling approach. The Volcano project was accepted to CNCF on April 9, 2020 and is at the Incubating project maturity level. It also supports popular computing frameworks such as TensorFlow and Spark, two technologies we heavily use, and this seemed to be a reasonable choice. The scheduling of two concurrent processes on the same physical machine and GPU is based on the GPU's memory size. Since we discovered that about 40% of the 16GB of the GPU's RAM was sufficient for a single standard trainer, we based the scheduling on this parameter, allowing us the flexibility to allocate a larger portion of memory to "heavier" models and schedule them alongside smaller memory usage models. So the total GPU device utilization will be higher.We use the Volcano environment variable to calculate the TensorFlow 1 memory fraction configuration or set the TensorFlow 2 virtual device memory configuration: One of the implementation issues we encounter in the process is with MIG (Multi instance GPU) enabled GPUs like the Nvidia A100.The GPU device plugins expose the resource to Kubernetes and assigns the GPU to the container based on the device index. In MIG enabled GPUs, all MIG instances share the same index of the physical GPU device. We tuned the Volcano device plugin to use the GPU and the MIG instance UUID as the unique identifier. Taboola pushed forward this effort and created this pull request which is still a work in progress, but it's stable in a large production environment for quite a long time. Figure-3 shows a typical yaml configuration file that uses Volcano to achieve our goal. In the future, we are looking into Volcano queuing mechanisms to better support backfilling workloads like large hyperparameter tuning jobs. ## Final thoughts Taboola is processing huge amounts of data every hour, multiple TB of data, and this obviously requires a complex and efficient pipeline to process all this data. One of these pipelines is running our ML models over and over to enable us to better understand what to display to the users at each specific point of time and site. We run our ML models using a complex software and hardware environment including over 100 NVIDIA GPUs orchestrated by kubernetes, Volcano, Dockers, TensorFlow and other software components. In order to squeeze more juice out of these pipelines, we went on to optimize the GPU usage, after identifying they are underperforming. Our efforts paid off and we reached a very nice x2 in throughput using the same hardware. We effectively gained x2 the number of GPUs for “free". Our longer term goals are to further squeeze the ecosystem's performance by, for example, optimizing our ML pipeline, migrating to Keras, exploring migrating CPU ML components to the GPU and exploring improving I/O operations as well. --- ### Production Incidents - 7 Practical Tips to Help You Through Your Next Incident URL: https://www.taboola.com/engineering/production-incidents-7-practical-tips-to-help-you-through-your-next-incident/ Last Modified: 2025-01-14 14:37:16 When your product or service is in downtime you lose more than just money. You are losing the trust of your users and partners. Therefore being proactive and doing what you can to prepare for production incidents is essential. Here are 7 pieces of advice to help you and your team prepare and deal with production incidents when you encounter them. ## Before the incident: ### #1 Don't let error logs fall through the cracks You probably use some kind of a logger in your service to log errors and other informative data, If not, it's time to start! For example, when an exception is raised, a lot of developers write something like this: try { //Block of code to try } catch(Exception e) { logger.error(“Got an exception, context: %s”, e, context); } If your service log is swelled with unknown exceptions you should be worried, because bad things are happening and users may be impacted. Therefore, you should monitor any error logs in your service and fire alerts. Some errors may not be so bad and some may be brutal, you should decide for yourself which ones you want to get alerted for. The ELK Stack is currently the most popular log management and logs analysis platform in the market used for monitoring. You can run a periodic query on top of Elastic Search to find undesired errors, and fire alerts when they happen. ### #2 Catch bad behaviors early by adding statistics to your service. Understanding the “status" of your service is essential for ensuring its reliability and stability. Detailed information and high visibility of the processes in your service not only helps your team react to issues but also gives them the confidence to make changes. One of the best ways to gain this “status" insight is with a robust monitoring system that gathers metrics, visualizes data, and fires an alert when things appear to be broken. A robust monitoring and alerting system will help you solve issues sooner and will minimize the damage done by the incident. Similarly to logs, you can configure alerts on your service based on statistics such as latency, queue lengths, CPU, memory usage, and so on. Prometheus and Grafana are among the most popular monitoring and alerting tools. You can also set up PagerDuty to make sure you always have an owner that handles incoming alerts. When configuring metrics and alerts, focus on the service purpose. What is its job? For example, If it's serving HTTP requests, then an important metric to set up is the HTTP return status code count. If a service job is to write something to a database, then an important metric to have can be the rate of writes to the database. ### #3 Prepare runbooks and invest in services documentation. It's possible that when things break down the owner/expert of the broken service will not be available, or the person dealing with the incident might be clueless regarding how to deal with the incident. Therefore, it's a good idea to have a prepared response plan for dealing with emergencies. (Here are some best practices to write one) The benefit of these prepared response plans (or runbooks) is you don't need an expert to be available every time there is a problem. This in turn reduces the burnout feeling when the same person has to deal with the same problems over and over again. If you can automate these plans it's even better. For example, restart the service automatically if the health check is failing. The downside with response plans is they are written for a specific case. And you can never cover all the cases. So it might be wise to train your incident responders on generic action taking, like finding a log or rolling back a deployment. Another important aspect of this is "service documentation". It doesn't have to be super detailed, but just enough to give the incident responder entry points to the service and critical things to know about it. ## During an incident: ### #4 “Read the damn error message" The first thing you should do when you want to solve the issue at hand is to understand what exactly went wrong. For example, If you get an error message from the log, read it carefully and look for details like: “What just happened", “Where it happened", and only then try to think about “Why it happened" and what you should do about it. So, the error messages in your log are valuable. This is why they should be as detailed as possible, and add important data, so anybody who reads them will know the “what" and the “where" immediately. (Stack traces are great for the “where" part) ### #5 Use data to understand the issue When you encounter a production issue, your first instinct shouldn't be to guess and immediately jump to a conclusion and try to fix the problem, which can lead to wasted time and resources. Instead, you should use the data you have at hand to devise a hypothesis and then validate that hypothesis with hard data. Think about it like playing detective. There are two ways to go about this: induction or deduction. With induction, you locate the relevant data, organize it, and then devise a hypothesis. Once you have a hypothesis, you can use data to prove it and then fix the problem. Deduction works similarly but involves enumerating the possible causes or hypotheses first and then eliminating the ones that don't fit with the data. This allows you to refine your remaining hypothesis until you arrive at a conclusion that can be proved with data. Either way, data should be at the heart of your efforts to solve the issue. ### #6 Aviate, Navigate, Communicate One of the first things pilots learn at ground school is what to do in case something goes wrong with the aircraft - “Aviate, Navigate, Communicate". Aviate means keeping the plane in the air. Navigate - to a safe location. And lastly, Communicate - let ground control know you are in trouble so they can help. We can apply the same principles used by pilots to the software domain too. In cases such as a complete downtime of our system (but not only): - Aviate - Do everything you can to bring the system up first. While saving data so you can analyze the root cause later, like taking a thread dump for example. - Navigate - Keep making decisions even when you lack complete information (avoid Analysis paralysis). - Communicate - Inform everyone that there is an issue, maybe they can help. Credit to Barak Luzon And Ariel Pizatsky who mentioned these principles in an excellent presentation they gave - When The Firefighters Come Knocking. ### #7 Avoid psychological biases and pitfalls Our brain is wired to make decision-making simpler. In doing so, it exposes itself to biases, heuristics, and other quirks that may seem like “bad decisions" in hindsight. One such example is the 'simulation heuristic'. The simulation heuristic is a psychological heuristic, or simplified mental strategy, according to which people determine the likelihood of an event based on how easy it is to picture it mentally. I.E You may think you know the incident's root cause simply because it's easier for you to imagine it. Another example is the “confirmation bias". The confirmation bias is the tendency to search for, interpret, favor, and recall information in a way that confirms or supports one's prior beliefs or values. For instance, you are more likely to find evidence that supports your existing hypothesis, and ignore evidence that disproves it. So what can you do to avoid these human biases? It's hard to say, and I don't think there is a silver bullet here. But being aware of them is the first step to mitigating them. I encourage you to watch a great talk by Boris Cherkasky that dives into how psychological biases affect incident response. ## Wrapping up Investing in a resilient monitoring and alerting system will result in more confidence when performing actions and deploying features on the system. Training your team to react to disasters, will remove the fear of breaking things. As a result, you may increase the development velocity of the now “fearless" developers. --- ### Chasing Ghosts URL: https://www.taboola.com/engineering/chasing-ghosts/ Last Modified: 2025-01-14 14:37:17 Bugs can be annoying most of the time, but if you're lucky enough (or unlucky, depending how you see it) they can also be very interesting. As part of our work in the Data Platform group at Taboola, we experienced one pretty elusive and interesting bug recently. ## Intro First, some background about Spark in Taboola. We use Spark extensively across many services and provide multiple ways to do so. One of the common ways to create a Spark Job is by using our “Analyzers" infrastructure. When someone wants to create a Spark job which is based on our main data pipe, they can usually do it by implementing it as a simple Analyzer. An Analyzer is a Java class that mostly contains only the required Spark SQL query. The rest of the concerns will be handled by our infra - triggering it when new data arrives, writing the result to HDFS in a unified way, loading the results from HDFS into our Vertica database if needed, and handling reruns, monitoring and other Spark complexities. As part of this Analyzers infra, after we get the user's dataset and before we write it to HDFS, we also repartition it, with a configurable small number of partitions. We do that in order to control the number of output files, to avoid creating too many small files and overload the HDFS NameNode which manages this metadata. ## Problem The story begins with a complaint we got from one of the R&D teams saying that something is wrong with some of their Analyzers. Their output tables contained duplicate rows, even though it shouldn't happen because the data is produced by jobs that calculate some simple aggregations using a group-by clause with a few key columns. For example, given a job with the following query: SELECT region, SUM(val) as value FROM input GROUP BY region Its output table contained more than one row per region, for some of the regions, and these rows were duplicates, including the value. So we have a problem, something is not working as expected and the results in the relevant Vertica table are incorrect, but that was basically everything we knew at that point. We had no way of knowing if the issue was somewhere in the Spark job or later in the process that loads the results to Vertica because the relevant data was no longer in HDFS due to retention. We need more data, let the investigation begin. *Side note - Our Vertica tables do not have any primary or unique key constraints because it's causing performance issues when working with relatively large tables. ## Investigation Begins Clueless about the source of the issue, we started taking steps that will provide us more information on the next occurrences of this bug: - We added monitoring on multiple Vertica tables, to alert us as soon as a duplicate row was found. For example, for the given query above, we queried the relevant table every few minutes, counting the number of rows per region and triggered an alert if the count was more than one. - We increased the retention of the relevant directories in hdfs, so we can inspect the Spark jobs output once we will get an alert for the corresponding table in Vertica. - We tried to reproduce the issue in a test environment, to see if we can get anything and understand if that's an issue in the specific executions of the relevant problematic timeframes. - We tried to look for suspicious Spark tickets that may be related (here) The tests were not successful, we didn't manage to reproduce the issue, but one day later we already got our first alert. We had new duplicate rows in one of the tables, and now that we still had the data in HDFS we could inspect it as well. We could now see that it's not an issue with the load to Vertica. The duplicate rows were already present in the data in HDFS, which means that we have a problem with our Spark Jobs. ## Collected Evidences We encountered this issue only a few times during a period of a few weeks since the first alert, but it was still enough to gather multiple observations along the way: - The issue rarely happens, and when it does - rerun fixes the data. We have hundreds of Analyzers in Taboola, each of them can be triggered multiple times per hour, and yet during a period of a few weeks we encountered the issue only a few times. - The issue affects a relatively small portion of the result rows. When it happens, only a relatively small subset of the result rows were duplicated. - The duplicates issue was a problem in multiple Analyzers, not only some specific victim. All of them were quite simple without any esoteric operations, just some simple aggregation - grouping by a few key columns and calculating the sum of a few values etc. - This is not only a duplicate-rows problem, we have missing rows too and their count is identical to the duplicate rows. In one of the first alerts, we counted the rows of the problematic execution in order to compare it to the rows count after the rerun. The rerun fixed the data, we no longer had duplicate rows in it, and yet - the total rows count stayed the same. If the rows count had the same value with and without the duplicates, it means that after the rerun we had rows that were missing from the first attempt, and their count is the same as the duplicates count. - When the issue occurs, the Spark job finishes successfully, and the duplicate rows are present in different files. When we got the first alert, we examined the data of the relevant execution in the corresponding directory in HDFS. We noticed that the job finished successfully and each single file didn't hold duplicates on its own. The duplicate rows appeared in two files, for example - some specific row was once in part-0 file, and once in part-4 file. - We saw correlation to shuffle fetch failures. Every time that we got an alert and inspected the execution we saw FetchFailed exceptions. When Spark needs to perform a shuffle of the data to redistribute it differently across partitions, the shuffle will happen as follows: First, the map side tasks will map the rows of each source partition to the new target partitions and store these shuffle data blocks locally. Next, the reduce-side tasks in the following stage will try to fetch their subset of shuffle data from the previous tasks locations in order to create the new partition and continue. When they fail to fetch this data, FetchFailedException is thrown. - We found a Spark ticket (SPARK-23207) that felt related to our issue. It was talking about a repartition operation and a failure that may lead to duplicate rows in the results. The thing is - this issue was fixed in a much earlier Spark version than ours, in Spark 2.x, while we were using Spark 3.1.2. On top of that, the ticket also provided an example code to reproduce it, and when we tried to execute it with our version, we didn't get duplicates. - Seems that we stopped getting duplicates in jobs once we disabled our pre-write repartition call (the one that we mentioned earlier, that we're adding before the write in order to control the number of output files). Despite the failed attempts to reproduce the above Spark issue with our version, this direction was still our main suspect. Why? One of the tests we were running as part of the investigation was to disable the pre-write repartition, for some of the Analyzers. We didn't experience the bug for this test group anymore. ## Theory - Intro With those observations in mind and after revisiting the fixed Spark issue, we came up with a theory that can explain what is going on and why the previous fix in the mentioned Spark ticket was not enough. Let's explain it with our per-region aggregation query example. Given such user's query, the full Spark job will be made of: sparkSession.sql( “ SELECT region, SUM(val) AS value FROM input GROUP BY region ” ).repartition(3) .write() … As mentioned earlier, we take the user's dataset, which is a result of a simple per-region aggregation query in this case, add a repartition call to reduce the number of the output partitions (to 3 in this case), and write it in HDFS. Such a job will have the following structure (The diagram is partial, we focus on the main details): It will be composed of 3 stages. - Stage 1 - Read input data from HDFS, calculate local partial sum of val per region, and map the results rows to the relevant shuffle data partitions, based on their region column that was used in the group-by, using hash partitioning. - Stage 2 - Read the shuffle data from the previous stage, calculate the final sum of val per region, and map the results rows to 3 partitions. The partitioning of rows to partitions is based on round robin partitioning, but we will get back to that soon. - Stage 3 - Read the shuffle data from the previous stage, and write the results to HDFS. The first shuffle, that separates stages 1 & 2, was created as a result of the group-by operation. The second shuffle, that separates stages 2 & 3, is a result of the repartition operation. ## Theory - When Things Go Wrong Let's assume that the job has started, stage 1 and 2 completed successfully, and we're somewhere in the middle of stage 3. Some of its tasks have finished successfully, tasks 1 and 3 in our case, but some did not, like task 2, and exactly at that point we're losing some shuffle data from stage 2. How do we lose it? A node that had stage 2 executor running on, died and is not reachable anymore, for example. In this case, task 2 from stage 3 will fail and Spark will trigger retry of the relevant tasks from stage 2, to recompute the missing shuffle data, in order to be able to retry the failed task 2 in stage 3 and finish the job. Before we will talk about the retry, let's first take a closer look at stage 2 tasks: Each task will get the relevant shuffle data from the previous stage and calculate the final sum of val per region, for the subset of regions in its partition. Next, due to the repartition call, the partition will be sorted, based on the hashcode of the rows, and each row will be mapped to its target partition using round robin. Wait, why does Spark sort the rows by their hashcode? When repartition was born, this sorting preparation step didn't exist. This local sort is the workaround that was implemented in Spark to fix the bug in the mentioned Spark ticket (see here). Since the rows order in such a partition is not guaranteed, when a retry of such a task was triggered, the rows order was different and each row was mapped to a different target partition than before. To avoid this issue, a local sort step was added per partition, to make sure that the mapping is consistent across retries. So the partition will be sorted, what can still go wrong on a retry? One more important detail about our job is that the val column data type is double. When dealing with doubles, the sum recalculation might produce a slightly different value when performed in different order (sum of floating point numbers is not commutative). Once a row has changed, its hashcode will change, its position after the sort will change, and it will shift multiple other rows even if they haven't changed. When the rows order after the sort is different, we are basically back to the original problem again. The rows will be mapped to different target partitions because the mapping is based on round robin partitioning. Ok, we might get different target partitions on a retry, but why does it lead to duplicate and missing rows in the results? Need to remember that we started from a point where tasks 1 and 3 in stage 3 have already completed. They finished writing their files and they won't be retried. Only task 2 is going to be retried now. Add it to the fact that the recalculated partition 2' from the previous stage is different than before and you'll get that: - It holds rows that were previously mapped to 1 and 3, and were written already, but now were mapped to 2 and will be written again, causing duplicate rows in the output. - Rows that were previously mapped to 2, and were not eventually written, were now mapped to 1 or 3, that are not going to be written, causing missing rows in the output. And why does the duplicate rows count equal the missing rows count? The recalculated partition in stage 2 is always the same size, because it will always hold the same subset of regions. The shuffle data partitions that are created from it are always the same size too, because the entire partition is the same size and its rows are mapped into target partitions via round robin that always begins from the same one. So, retry or not, part 2 should always contain N rows in total. If X of them are duplicates, we must have X rows that are missing as well. ## Reproducing it Now that we had a theory and we understood the situation better, we wanted to create a test version that will reproduce it consistently in our environment and see the theory in practice. The test job was similar to this: sparkSession.sql( “ SELECT region, SUM(val) AS value FROM input GROUP BY region ” ).repartition(3) .map((MapFunction<Row, Row>) row -> { if (TaskContext.get().stageAttemptNumber() == 0 && TaskContext.get().attemptNumber() == 0 && TaskContext.get().partitionId() > 1) { Thread.sleep(60_000); System.exit(1); } return row; }, RowEncoder.apply(schema)) .write() … - We implemented a simple double-value-per-region calculation as in our examples. - We added a code to kill an executor in the middle of stage 3. After we repartition and before we write the dataset, we added a dummy rows mapper that is not really doing anything but killing the running executor (calling System.exit(1)) after sleeping for one minute if we're in the first attempt and we are not in the first two partitions. We are basically exiting an executor while we are at the first attempt of the 3rd task (partitionId = 2) of the 3rd stage, after we sleep for enough time in order to make sure that the first 2 tasks of this stage will finish successfully. This is not enough yet. In order to reproduce the bug, we want that losing this executor will cause loss of shuffle data from the previous stage, to trigger the desired retry of tasks from stage 2. In order to get it we changed few configurations as well: We executed the job on a small static setup of executors, with dynamic allocation disabled, to make sure that tasks from all the stages are running on all of them. We disabled the external shuffle service, in order to lose the shuffle data when we lose an executor (external shuffle service is an external component in the cluster that can serve shuffle data even when its source executor is gone). And we disabled Spark speculation, to avoid a too-early-retry of our sleeping task, before we wanted it to happen. Success! We got the expected tests results: - We got duplicates on every execution of this job. - When we executed it without repartition, but with repartition that is based only on the region column, we didn't get duplicates. To avoid being dependent on unwanted columns when we repartition, we can provide the key columns that we want Spark to use for the partitioning. For example, instead of: df.repartition(n) We can do: df.repartition(n, col(“region”)) Which will use hash partitioning based only on the region column to partition it into n partitions, so we don't care if the val column has changed a bit on a retry because it won't affect the partitioning. And indeed, this time we didn't get duplicate rows because the partitioning was consistent across retries. - When we executed the same code, including the repartition, but instead of using a double type value column we used an integer column, we didn't get duplicates. ## Conclusion We decided to consider repartition(N) as a non safe operation that should generally be avoided because it can't really be used on an arbitrary dataset. Its result depends on the hashcode of the entire row, including non deterministic columns (e.g. floating point calculations, random values), and we saw that it can lead to correctness issues when retries are involved. We also found a related spark ticket (SPARK-38388). What are the current alternatives? - Coalesce: We said that we mostly use repartition to limit the number of output files, so why don't we just use coalesce? The reason is that coalesce happens too soon in most of our cases. It doesn't add a stage and a shuffle, it works differently by merging the partitions of the given dataset. It's not the desired behavior in most of our cases because we usually have a heavy computation that we want to keep running with the original higher parallelism and limit the partitions count only a moment after that. Using coalesce might cause performance degradation in most of our cases. - Repartition over selected columns: As we saw, calling df.repartition(n) will repartition the rows based on their position in the source partition, after it was sorted by the rows hashcode which is based on all of their columns. We showed that we can avoid the bug by choosing our own safe subset of partitioning columns, as we did in our test with: df.repartition(n, col(“region")).It can create skewed results compared to the uniform partitions created by the round robin partitioning, but in practice it's not really an issue in our case and we usually have wide enough columns to use when repartitioning, so skewness is mitigated. - Working with big decimal data types instead of double/float types: This is a good approach regardless of this issue and used in other places in our system, but it doesn't come for free (more resource demanding), and it's not always required. For every job the owner can decide if floating point calculation is enough in its case or not. In addition, in this case it required more changes on our side and would not fully protect against similar variants of this bug (because, as we mentioned earlier, the floating point issue is only a specific example of this bug, which can generally occur in any scenario with a non deterministic column). Repartition over selected columns was our chosen solution. Since we do that in the infra and not specifically per Analyzer, we implemented it dynamically based on the column types of the given dataset. We get the user's dataset, extract its schema's columns and choose a subset of columns for the repartition based on their types (float & double columns are not allowed, etc). This is the default behavior, but it can also be customized per Analyzer if needed. Note though that this approach is not bulletproof. For instance, a boolean column may be computed based on a float column, and therefore still be non-deterministic. We also plan to check additional directions. One possible alternative that we're looking into includes the new dynamic coalesce abilities that are part of the adaptive query execution in the new Spark versions. We already verified that it eliminates the problem as well, but we still need to test it because we want a solution that will have minimal impact on production. We want to avoid the concern of tuning the hundreds of different queries we have, and adaptive query execution sometimes requires it. --- ### Surviving Spark Upgrade in Production URL: https://www.taboola.com/engineering/surviving-spark-upgrade-in-production/ Last Modified: 2025-01-14 14:37:17 Going to upgrade your data pipeline to Spark3? Read about the issues we encountered while we upgraded the data pipeline in Taboola. Time has come and we decided that we need to move forward and upgrade our data pipeline from Spark 2.2.3 to Spark 3.1.2. One may think that upgrading to the new Spark version is just a matter of a simple version number change in our dependency management. Well, it's hardly ever the case, especially for a core dependency such as Spark, which affects many of our services in our monorepo project. A new version can include well documented API changes, bug fixes and other improvements. However it may also come with some other, not so documented, functionality changes, bugs and unexpected performance issues. In this blog post we will share issues from our Spark upgrade journey in Taboola. A journey that started with a dependency version number change and ended after multiple code adaptations and configuration changes that we applied in order to overcome several issues. We will dive into 8 issues, explain the symptoms that we have seen in each of them, the reason behind it including related Spark tickets. And explain the workaround that we chose to apply for each of them. ## See What Breaks The first step in our journey, and probably the easiest, is changing Spark version and dealing with the failures, be it compilation errors or failing tests. Well, we make it sound easy… As mentioned before, we have a monorepo project, and with hundreds of different production workloads, we couldn't just upgrade Spark, test it all in a couple of weeks and go on with our lives. It was a migration that was cautiously managed over a few months, in which we had to support both old and new Spark versions side by side, but that's for another blog post… Anyway, there were several changes in behavior that surfaced up in our unit tests: ### #1 - Proleptic Gregorian calendar VS Hybrid calendar Spark 3 comes with a change in parsing, formatting and conversion of dates and timestamps. Previous Spark versions used the hybrid calendar while Spark 3 uses the Proleptic Gregorian calendar and Java 8 java.time packages for manipulations. This change affects multiple parts of the API, but we encountered it mostly in 2 places - when parsing date & time data that is provided by the user, and when extracting sub components like day of week and so on. #### Parsing: We have many tests that read their input data from JSON files with schema that includes timestamps fields. These tests now failed due to new strict timestamp parsing. In our case, the parsing failed since there was no match between the provided format “yyyyMMddHH" and the input, e.g: “2019-10-22 01:45:36.0185438 +00:00". Code: sparkSession .read() .schema(getFileSchema()) .json(inputPath) .withColumn("start_ts", unix_timestamp(col("StartTimestamp"), "yyyyMMddHH").multiply(lit(1000))) .createOrReplaceTempView(tempViewName); Luckily, the symptom in this case was a descriptive exception, which had very clear cause: Caused by: org.apache.spark.SparkUpgradeException: You may get a different result due to the upgrading of Spark 3.0: Fail to parse '2019-10-22 01:45:36.0185438 +00:00' in the new parser. You can set spark.sql.legacy.timeParserPolicy to LEGACY to restore the behavior before Spark 3.0, or set to CORRECTED and treat it as an invalid datetime string. Docs: Documentation can be found in Spark migration guide in the section about the Gregorian calendar: https://spark.apache.org/docs/latest/sql-migration-guide.html Solution: As can be seen from the exception, we had two options to handle that. One option was to fix the supplied timestamp pattern in both test and production code. The only problem with this option was the amount of affected tests and jobs. We wanted to move fast so we decided to handle that later, gradually, after the upgrade, and use the other option instead, which is to force Spark to behave the same as in the previous versions by setting spark.sql.legacy.timeParserPolicy = LEGACY. #### Sub Component Extraction: As mentioned above, the time related changes also affect the sub-components extraction API. In our case, we had code that tried to extract the day of week from a timestamp using the date_format function along with the “u" pattern. This wasn't supported anymore and raised an exception. Code: date_format(col("request_time"), "u") Solution: We fixed our code to use the supported DAYOFWEEK_ISO function instead. expr("extract(DAYOFWEEK_ISO from request_time") In Spark3 the array type created by collect_list and collect_set functions is not nullable and can not contain null values. The data type itself has changed, which means that the column type of a column with such an expression is expected to change as well. We first experienced it in our unit tests. We have tests that create some expected schema programmatically and compare it with the results schema. One of our tests started failing after the upgrade due to schema mismatch. While investigating it we found that the behavior has changed in Spark3 in order to comply with Hive behavior. Code: static private StructType getExpectedSchema() { List fields = new ArrayList<>(); fields.add(DataTypes.createStructField("id", DataTypes.LongType, true)); fields.add(DataTypes.createStructField("client_data", DataTypes.createArrayType(DataTypes.createStructField("client_id", DataTypes.LongType, true), true), true)); return DataTypes.createStructType(fields); } Spark issue: https://issues.apache.org/jira/browse/SPARK-30008 Solution: The solution was quick and easy. As the nullability of the column didn't really matter to us, we just changed the expected schema accordingly and created this type with nullable=false in createStructField and containsNull=false in createArrayType. ### #2 - Random using seed produces different results in Spark3 We have code that generates pseudo random numbers by using the rand function along with a given seed. The unit tests that cover this functionality expect a predefined sequence of numbers as a result. Guess what? These unit tests failed. We discovered that the rand function result was based on XORShiftRandom.hashSeed method, which has changed in Spark3. This change caused different return values in the new version. Spark issue: https://issues.apache.org/jira/browse/SPARK-23643 Solution: The solution here was also quick and easy. We fixed the tests according to the new implementation results (as it had no effect on the real functionality of the job that wanted pseudo random numbers for sampling purpose and used seed just for deterministic test results). ### #3 - Quantiles calculation with approxQuantile function produces different results We use approxQuantile function in cases where we need to calculate percentiles, we don't want to pay the price of the exact calculation on massive data and the approximate result is good enough. In Spark 3 a few bugs were fixed in approxQuantile. In our case it was discovered again by a few tests that failed because the return value of this function was now different from the expected value with the previous version. Spark issue: https://issues.apache.org/jira/browse/SPARK-22208 Solution: We fixed the test and changed the expected result based on the new functionality. ## Try It on Production Finally, we had a green build, yay! Now it was time to test real production workloads with the upgraded Spark version. This is where things started to get interesting, and we encountered various performance issues. ### #1 - Constraint propagation can be very expensive When we started testing production workloads, we noticed that several jobs failed with OOM in the driver before any progress was made in the executors. Profiling the driver for each of them revealed that they spent most of their time on InferFiltersFromConstraints or PruneFilters. Both of these are Spark catalyst optimizations rules that rely on constraint propagation. After searching through some Spark issues we found that it's a known issue that wasn't fully resolved yet. There are query plans that can cause this constraint propagation calculation to be very expensive and even cause OOM in the driver due to the amount of memory used. Profiler result: Spark issues: https://issues.apache.org/jira/browse/SPARK-33013 https://issues.apache.org/jira/browse/SPARK-19846 Solution: As mentioned in the Spark issues, the suggested workaround in such cases is to disable constraint propagation. The flag is: spark.sql.constraintPropagation.enabled = false ### #2 - Temp view and cache invalidation For some workloads, the total time across all tasks in one of our jobs was multiplied by 4 in Spark 3 compared to Spark 2. Spark UI revealed the interesting findings. Spark 2: Spark 3: It was obvious that with the new version the job was reading the same data from HDFS several times instead of reusing the cached result. We found that this job was recreating a cached view unintentionally more than once between its calculations, which invalidated the cache with the new Spark version. Code: df.cache(); ... df.createOrReplaceTempView("viewA") Dataset df2 = ...query1FromViewA... ... df.createOrReplaceTempView("viewA") Dataset df3 = ...query2FromViewA... Docs: https://spark.apache.org/docs/latest/sql-migration-guide.html Solution: We changed the code to create the temp view just once, to allow Spark to reuse it. Fix: ### #3 - Degraded compression ratio for repeated fields One of our main Spark jobs writes massive amounts of data into parquet files, about 2TB each hour. With Spark 3 we've noticed a significant increase (roughly 10%) in the amount of data written to HDFS. Such an increase in data size surely comes with a price. In our case, having retention of a month, that's an extra 250 TB in storage (with 3 replicas), and likely slowness in downstream jobs, having to read more data. When drilling down to parquet metadata, we have noticed that while most columns maintained a similar, or even improved, compression ratio, some columns had suffered a severe degradation. It is time to state that the schema of this output contains over a thousand columns, many of them consisting of several nesting levels. One of these columns, containing a list of INT64 values, is our biggest in our schema, and occupies roughly 25% of the entire data. Unfortunately, when we sampled a few row groups, we noticed that the compression for this column severely degraded. It did not go unnoticed that both parquet and snappy versions were upgraded in Spark, and after a bunch of tests we've made with different snappy versions, we have confirmed this to be a degradation with snappy. Spark issues: More details can be found in stack overflow, and we have also opened a bug for snappy. Solution: The solution for us was to downgrade snappy back to 1.1.2.6, keeping in mind that with Spark 3 comes the prospect of using zstd compression. Which significantly improves the data compression in our data pipeline. But that's again for another blog post… ### #4 - Inflation of parquet row groups After overcoming the snappy issue, we could finally see the light at the end of the tunnel. The data size of both Spark 2 and Spark 3 was nearly the same. Now it was time to check downstream jobs over this data, mark a nice V in our checklist and move on. However, when reading the data by various downstream jobs, we noticed x3 times in input size for very simple queries (reading just a few simple type columns out of our huge schema). This was puzzling… How come data size has hardly changed, yet input size spiked? We've confirmed that the encoding and compression of the relevant columns for the query was pretty much the same. Then we noticed something different in the metadata. The data written with Spark 2 had less row groups in each parquet part, and the sizes were more uniform. row group 1: RC:15057 TS:65138984 OFFSET:4 row group 2: RC:15982 TS:60749296 OFFSET:22204117 row group 3: RC:14798 TS:62376888 OFFSET:42677725 row group 4: RC:14190 TS:69236737 OFFSET:63708850 row group 5: RC:17804 TS:61184445 OFFSET:86972402 row group 6: RC:16334 TS:74862179 OFFSET:108491642 row group 7: RC:13662 TS:72208271 OFFSET:132908516 row group 8: RC:16020 TS:55804928 OFFSET:155496590 ... However, with Spark 3 we observed 3-4 times the amount of row groups, and the sizes appeared as saw teeth - trending down gradually and then jumping again to high values: row group 1: RC:11391 TS:40761929 OFFSET:4 row group 2: RC:12215 TS:41366747 OFFSET:16972776 row group 3: RC:10702 TS:46590968 OFFSET:32677534 ... row group 10: RC:2919 TS:10232211 OFFSET:97373151 row group 11: RC:2795 TS:6880899 OFFSET:101956077 ... row group 22: RC:736 TS:1985475 OFFSET:124174705 row group 23: RC:664 TS:1878159 OFFSET:125182769 row group 24: RC:14063 TS:72394491 OFFSET:134217728 row group 25: RC:14307 TS:42351249 OFFSET:158503969 row group 26: RC:10100 TS:23834702 OFFSET:173732904 ... row group 37: RC:2225 TS:5607859 OFFSET:244260674 row group 38: RC:1859 TS:4782242 OFFSET:246965160 ... row group 45: RC:645 TS:1951772 OFFSET:259046729 row group 46: RC:526 TS:2097179 OFFSET:259958806 row group 47: RC:16164 TS:66444278 OFFSET:268435456 ... The reason for that is that Spark 3 came with parquet 1.10, while Spark 2.2 (the version we've used), came with parquet 1.8, and the `max-padding` default changed from 0 to 8MB.This behavior is explained in Parquet row group layout anomalies, but to TLDR it - when writing data at parquet format, parquet-mr tries to estimate how much data to write to a single row group, in order not to overflow from the block size (which would result in a row group split across 2 blocks and harm data locality). Setting max-padding to 0, disables this heuristic approach. In our use case we were less concerned about it, as we don't exploit data locality - our compute clusters are separated from our HDFS cluster. At worst, we're paying the cost of reading two blocks per row group, but we prefer paying it then having an inflation of row groups. Ok, but how does all that explain the growth in input size? Let's take a look at parquet format: row group 1: RC:11391 TS:40761929 OFFSET:4 row group 2: RC:12215 TS:41366747 OFFSET:16972776 row group 3: RC:10702 TS:46590968 OFFSET:32677534 ... row group 10: RC:2919 TS:10232211 OFFSET:97373151 row group 11: RC:2795 TS:6880899 OFFSET:101956077 ... row group 22: RC:736 TS:1985475 OFFSET:124174705 row group 23: RC:664 TS:1878159 OFFSET:125182769 row group 24: RC:14063 TS:72394491 OFFSET:134217728 row group 25: RC:14307 TS:42351249 OFFSET:158503969 row group 26: RC:10100 TS:23834702 OFFSET:173732904 ... row group 37: RC:2225 TS:5607859 OFFSET:244260674 row group 38: RC:1859 TS:4782242 OFFSET:246965160 ... row group 45: RC:645 TS:1951772 OFFSET:259046729 row group 46: RC:526 TS:2097179 OFFSET:259958806 row group 47: RC:16164 TS:66444278 OFFSET:268435456 ... So now imagine that we've tripled the amount of row groups for the same amount of data, as a result - the footer size of each parquet file was tripled. ## Final thoughts Upgrading Taboola data pipeline to Spark3 was an interesting journey. We deepened our knowledge in Spark and got familiar with the changes that come with the new version. We hoped that the changes, bug fixes and functionality improvements of the new version will improve the performance of our Spark jobs, but the truth is that the overall performance of our clusters has not changed dramatically. We did see improvements in some cases, but degradation in others as well. But hey - we can use some fresh new features now, and we are definitely going to try them. --- ### Why We Switched to Redux Toolkit for Internal App Development at Taboola URL: https://www.taboola.com/engineering/redux-toolkit/ Last Modified: 2025-01-14 14:37:17 ### tl;dr Redux Toolkit (RTK) saves time and reduces complexity, and the Redux stewards want it to be the industry-standard approach to architecturing Redux in apps As a tech company, when you have the opportunity to automate repetitive, engineering-heavy tasks, it's typically the smartest route to take in an ever-competitive, ever-complex landscape. These open engineering burdens waste humanpower, exhaust people's time and energy, and tend to create a growing backlog of engineering debt that snowballs quickly. At Taboola we're no different in how we feel about inefficiencies due to unnecessary repetition. To prove this, we recently created a new team, Operations at Scale, in the Publisher domain of our Professional Services department. Our team is dedicated to automating repetitive operations and workflows and to addressing unsolved inefficiencies in the day-to-day work that serves our Publishers. The Operations at Scale team tackles inefficiencies in a number of ways. One main focus is on the development of internal apps, which help remove barriers to entry for key user groups, like our Implementation teams or our Support engineers. Likewise, as a company with a large book of business, we naturally handle a lot of data, and the popular frontend stack for streamlining this data into user-friendly interfaces is React / Redux. Despite an increasing number of options for state management in React like the useState hook, the Redux library remains a common choice for developers due to its reliable performance, powerful debugging capabilities, and scalability. There are some downsides however. These include: - Redux often requires additional installation of other libraries, such as reselect (which can be used to create custom selectors) - Implementing Redux requires a large amount of boilerplate and repetitive code (again not big fans of this at Taboola) - A Redux store is complex to configure, with a steep learning curve Enter Redux Toolkit (RTK). It describes itself as SOPE, which stands for “Simple, Opinionated, Powerful and Effective". By abstracting over the familiar setup process that Redux requires and including common utilities like Immer (which greatly reduces the complexity of depending on spread operations re: state management), RTK solves a lot of the usual issues with Redux raised above. In Operations at Scale, we have found our code has become DRYer (Don't Repeat Yourself), vastly more maintainable, and moving through faster development cycles since we've switched to RTK. Let's look at some examples to illustrate the difference RTK can make to your codebase. Using a simple counter example, an action creator function would look something like this in the standard Redux library: RTK has a helper function that combines this into one clean declaration called createAction. The above could be written like this: Already, the visible difference between the “legacy" implementation of Redux and the RTK implementation of Redux shows not only simplified code, but also the simplification of the domain knowledge necessary for understanding what Redux is doing under the hood. Additionally, the reducer for this example counter in standard Redux is even more complex and requires a greater level of understanding for the engineers who work on the project. It would be like so: RTK provides a utility called createReducer that simplifies creating reducer functions: You may notice that the createReducer RTK code directly mutates the state, which is a violation of the Redux rule to keep reducer functions pure. One of the big advantages of RTK is its inclusion of the Immer package. This is used under-the-hood to drastically simplify immutable update logic by writing “mutative" code in reducers, thus removing the need for excessive JS spread syntax, often a necessary evil in standard Redux to ensure state is not mutated. The RTK utilities we have looked at so far are useful, but you start to see the real power of switching from the standard Redux library when using the createSlice function. This combines createAction and createReducer into one, allowing you to define reducer functions, initial state, and generate the corresponding action creators and types, all from calling one function. Here's how this would look with our counter example: Hopefully, you can now begin to see how creating slices can dramatically simplify standard Redux syntax, with everything in one place, making developers lives much easier. The reason we appreciate Immer and RTK in general is not only for the cleanliness it offers, but because it directly influences the architecture of our state and, likewise, the overall architecture of the app's codebase. Regarding how we develop our ever-growing state object, we no longer have to spend (waste) time debating which layer of state a new layer should go into, how much of the current code base requires a refactor as a result of new state additions, etc. As noted, the inherent difficulty of a complex app's Redux store can easily create a barrier of entry for any junior devs looking to onboard and contribute quickly to an app's development, which is an issue we've encountered many times. So we see the reduction of complexity not only benefitting the velocity of our development process, but also our capacity since it allows us to onboard more junior devs due to the code's intuitive characteristics like simple key-value assignments. For large and complex apps, like the ones we work on at Taboola, this makes a substantial difference in development time and keeping our codebases clean. RTK has many other advantages and functions that have not all been covered here. Make sure to check out the official documentation to see what else RTK has to offer when moving from the standard Redux library. --- ### Detecting ANRs In Your Application URL: https://www.taboola.com/engineering/detecting-anrs-in-your-application/ Last Modified: 2025-01-14 14:37:18 If there was a way to make any of my applications ANR proof, I would do it. They (ANRs) always seem to sneak up on you and the problem is that you have nothing you can do about them. You might see them pop up when seeing the overview of your application in Google Play Console. But there won't be a lot of information there to understand how the ANR happened and what you could do to fix the situation. Furthermore, if a user happens to experience an ANR, you have to rely on their willingness to invest time and effort to tell you about it. And we have all been on the other side. If you happen to be using an application that gets stuck, the one thing you do is go straight to uninstalling it. So, how can we as developers, do everything in our power to protect the users of our applications from experiencing ANRs? Let's find out. ## Be Proactive Before we delve into solutions that can help us detect ANRs, let's understand what we can do to avoid getting them in the first place (or minimizing the chance they can happen). These points might sound obvious, but in a large enough application, it might be easy to overlook things: - Check to see if you have places in your code that are doing extensive work on the UI thread. The work done on the UI thread should be short and relating to something to do with the UI of your application. If you are doing any other logic there or perhaps even asynchronous work, delegate it to a background thread or a service - If by any chance you have threads that hold locks or certain blocks of code that need to be synchronized, make sure you are not creating a deadlock or a certain state of your application reaches it - If your application deals with broadcast receivers, you must verify that the execution of the onReceive method is short and ends in a timely fashion. If there is work there that can take some time, delegate it to a background thread Another way that you can detect places that might cause ANRs is by using StrictMode. You can use it while developing your application as it catches accidental disk or network usage on the main thread. ## Be Smart You have gone over your application, you think it isn't at risk for any ANRs and so you release it for public consumption. Lo and behold a couple months go by and then you start seeing reports of ANRs. What could you have done differently? As we said earlier, those crash reports hardly provide any information regarding the ANR. To allow your application to give you the most amount of detail that it can when an ANR occurs, I will detail two options: - Run a thread which polls the UI thread to see if it is stuck - On API level ≥ 30, you can use getHistoricalProcessExitReasons There is already a library called ANR-Watchdog which takes care of detecting ANRs and providing you with all the details. In case you don't want to use it or want to have something of your own, here is a rough outline of what it does: - Create a thread which runs on the main thread (it doesn't have to do any actual work) - See if the thread's execution is completed after a few seconds - If it did, then no ANR took place and you run the thread again - If it did not, some other thread is blocking the main thread and causing an ANR Below is a rough outline of such a class: package com.tomerpacific.anrdetection import android.os.Handler import android.os.Looper import java.lang.Exception class ANRHandler: Thread() { val TIMEOUT: Long = 5000L private val handler: Handler = Handler(Looper.getMainLooper()) private val worker : Runnable = Runnable { } override fun run() { while (!isInterrupted) { handler.postAtFrontOfQueue(worker) try { sleep(TIMEOUT) } catch (exception: Exception) { exception.printStackTrace() } if (handler.hasMessages(0)) { //worker has not finished running so the UI thread is being held val stackTrace: Array<StackTraceElement> = currentThread().stackTrace var output: String = "" for (element in stackTrace) { output += element.className + " " + element.methodName + " " + element.lineNumber } print(output) } } } } ⚠️ The execution of the runnable is always on the main thread, but since it doesn't do any work, it is not supposed to impact your application's performance. You could also decide to run it every desired time interval Option #2 can make your life simpler as its API gives you a lot of information. Introduced in Android 11 (API level 30), getHistoricalProcessExitReasons, does exactly what you think it does. It returns a list of recorded objects that account for the most recent application terminations. This method is called on the ActivityManager and accepts three arguments: - The package name - of String type (can be null) - The process's id that belonged to the package - of int type - The maximum amount of reasons you want to get back - of int type It's important to note that all of these arguments can be substituted with default values. I.E., you can pass null as the package name and get the entire exit reasons for the caller's UID So what do these recorded objects contain? Well, these objects are of ApplicationExitInfo type and they can provide you with a lot of useful information. For starters, you could call the getReasonmethod to find out why the process terminated. This method returns an integer marking the code for the exit reason. If the value returned is6, that means the application was terminated because it was unresponsive due to the fact that an ANR happened. That's neat, but how can we see where the ANR happened from? For that we can use getTraceInputStream. Like the name implies, the returned value is an InputStream of bytes which needs to be read like any other InputStream. An example output looks like the following: I/System.out: ----- pid 2738 at 2022-04-26 17:48:12 ----- Cmd line: com.tomerpacific.anrdetection Build fingerprint: 'Android/sdk_phone_x86/generic_x86:11/RSR1.210210.001.A1/7193139:userdebug/dev-keys' ABI: 'x86' Build type: optimized I/System.out: Zygote loaded classes=15746 post zygote classes=728 Dumping registered class loaders #0 dalvik.system.PathClassLoader: [], parent #1 #1 java.lang.BootClassLoader: [], no parent I/System.out: #2 dalvik.system.PathClassLoader: , parent #1 Done dumping class loaders Classes initialized: 302 in 19.361ms Intern table: 31490 strong; 543 weak JNI: CheckJNI is on; globals=637 (plus 31 weak) I/System.out: Libraries: libandroid.so libaudioeffect_jni.so libcompiler_rt.so libicu_jni.so libjavacore.so libjavacrypto.so libjnigraphics.so libmedia_jni.so libopenjdk.so librs_jni.so libsfplugin_ccodec.so libsoundpool.so libstats_jni.so libwebviewchromium_loader.so (14) Heap: 91% free, 2330KB/26MB; 67022 objects Dumping cumulative Gc timings I/System.out: Average major GC reclaim bytes ratio inf over 0 GC cycles Average major GC copied live bytes ratio 0.738176 over 4 major GCs Cumulative bytes moved 11482280 Cumulative objects moved 217937 Peak regions allocated 28 (7168KB) / 768 (192MB) I/System.out: Start Dumping histograms for 1 iterations for young concurrent copying ProcessMarkStack: Sum: 26.311ms 99% C.I. 26.311ms-26.311ms Avg: 26.311ms Max: 26.311ms ScanImmuneSpaces: Sum: 5.625ms 99% C.I. 5.625ms-5.625ms Avg: 5.625ms Max: 5.625ms VisitConcurrentRoots: Sum: 1.121ms 99% C.I. 1.121ms-1.121ms Avg: 1.121ms Max: 1.121ms I/System.out: (Paused)ClearCards: Sum: 375us 99% C.I. 7us-235us Avg: 28.846us Max: 235us GrayAllDirtyImmuneObjects: Sum: 329us 99% C.I. 329us-329us Avg: 329us Max: 329us VisitNonThreadRoots: Sum: 327us 99% C.I. 327us-327us Avg: 327us Max: 327us I/System.out: InitializePhase: Sum: 306us 99% C.I. 306us-306us Avg: 306us Max: 306us (Paused)GrayAllNewlyDirtyImmuneObjects: Sum: 164us 99% C.I. 164us-164us Avg: 164us Max: 164us (Paused)FlipCallback: Sum: 144us 99% C.I. 144us-144us Avg: 144us Max: 144us SweepSystemWeaks: Sum: 142us 99% C.I. 142us-142us Avg: 142us Max: 142us I/System.out: ScanCardsForSpace: Sum: 125us 99% C.I. 125us-125us Avg: 125us Max: 125us ThreadListFlip: Sum: 96us 99% C.I. 96us-96us Avg: 96us Max: 96us ClearFromSpace: Sum: 78us 99% C.I. 78us-78us Avg: 78us Max: 78us CopyingPhase: Sum: 76us 99% C.I. 76us-76us Avg: 76us Max: 76us I/System.out: FlipOtherThreads: Sum: 58us 99% C.I. 58us-58us Avg: 58us Max: 58us ProcessReferences: Sum: 54us 99% C.I. 19us-35us Avg: 27us Max: 35us SweepArray: Sum: 53us 99% C.I. 53us-53us Avg: 53us Max: 53us I/System.out: EnqueueFinalizerReferences: Sum: 38us 99% C.I. 38us-38us Avg: 38us Max: 38us RecordFree: Sum: 37us 99% C.I. 14us-23us Avg: 18.500us Max: 23us ForwardSoftReferences: Sum: 25us 99% C.I. 25us-25us Avg: 25us Max: 25us FlipThreadRoots: Sum: 21us 99% C.I. 21us-21us Avg: 21us Max: 21us I/System.out: (Paused)SetFromSpace: Sum: 19us 99% C.I. 19us-19us Avg: 19us Max: 19us ResumeRunnableThreads: Sum: 12us 99% C.I. 12us-12us Avg: 12us Max: 12us EmptyRBMarkBitStack: Sum: 8us 99% C.I. 8us-8us Avg: 8us Max: 8us SwapBitmaps: Sum: 7us 99% C.I. 7us-7us Avg: 7us Max: 7us Done Dumping histograms young concurrent copying paused: Sum: 750us 99% C.I. 750us-750us Avg: 750us Max: 750us I/System.out: young concurrent copying freed-bytes: Avg: 1052KB Max: 1052KB Min: 1052KB Freed-bytes histogram: 960:1 young concurrent copying total time: 35.641ms mean time: 35.641ms young concurrent copying freed: 8956 objects with total size 1052KB I/System.out: young concurrent copying throughput: 255886/s / 29MB/s per cpu-time: 179578666/s / 171MB/s Average minor GC reclaim bytes ratio 0.742269 over 1 GC cycles Average minor GC copied live bytes ratio 0.276211 over 2 minor GCs Cumulative bytes moved 1410368 Cumulative objects moved 26626 I/System.out: Peak regions allocated 28 (7168KB) / 768 (192MB) Total time spent in GC: 35.641ms Mean GC size throughput: 28MB/s per cpu-time: 169MB/s Mean GC object throughput: 251284 objects/s Total number of allocations 75978 Total bytes allocated 3382KB Total bytes freed 1052KB I/System.out: Free memory 23MB Free memory until GC 23MB Free memory until OOME 189MB Total memory 26MB Max memory 192MB Zygote space size 3040KB Total mutator paused time: 750us I/System.out: Total time waiting for GC to complete: 80.600us Total GC count: 1 Total GC time: 35.641ms Total blocking GC count: 0 Total blocking GC time: 0 Histogram of GC count per 10000 ms: 0:1 Histogram of blocking GC count per 10000 ms: 0:1 Native bytes total: 15621964 registered: 98204 I/System.out: Total native bytes at last GC: 15537168 /system/framework/oat/x86/android.hidl.manager-V1.0-java.odex: quicken /system/framework/oat/x86/android.test.base.odex: quicken /system/framework/oat/x86/android.hidl.base-V1.0-java.odex: quicken I/System.out: Current JIT code cache size (used / resident): 0KB / 32KB Current JIT data cache size (used / resident): 4KB / 32KB Zygote JIT code cache size (at point of fork): 45KB / 48KB Zygote JIT data cache size (at point of fork): 33KB / 36KB Current JIT mini-debug-info size: 26KB I/System.out: Current JIT capacity: 64KB Current number of JIT JNI stub entries: 0 Current number of JIT code cache entries: 48 Total number of JIT compilations: 6 Total number of JIT compilations for on stack replacement: 1 I/System.out: Total number of JIT code cache collections: 0 Memory used for stack maps: Avg: 35B Max: 52B Min: 28B Memory used for compiled code: Avg: 125B Max: 257B Min: 69B Memory used for profiling info: Avg: 70B Max: 188B Min: 20B Start Dumping histograms for 48 iterations for JIT timings Compiling: Sum: 385.780ms 99% C.I. 0.556ms-25.610ms Avg: 8.037ms Max: 25.610ms I/System.out: TrimMaps: Sum: 44.431ms 99% C.I. 2.400us-5148us Avg: 925.645us Max: 5643us Done Dumping histograms Memory used for compilation: Avg: 83KB Max: 322KB Min: 8560B ProfileSaver total_bytes_written=0 ProfileSaver total_number_of_writes=0 ProfileSaver total_number_of_code_cache_queries=0 I/System.out: ProfileSaver total_number_of_skipped_writes=0 ProfileSaver total_number_of_failed_writes=0 ProfileSaver total_ms_of_sleep=5000 ProfileSaver total_ms_of_work=0 I/System.out: ProfileSaver total_number_of_hot_spikes=5 ProfileSaver total_number_of_wake_ups=0 I/System.out: suspend all histogram: Sum: 11.468ms 99% C.I. 0.018ms-10.658ms Avg: 1.042ms Max: 11.094ms DALVIK THREADS (15): "main" prio=5 tid=1 Runnable | group="main" sCount=0 dsCount=0 flags=0 obj=0x72107008 self=0xe7d05410 | sysTid=2738 nice=-10 cgrp=top-app sched=0/0 handle=0xf6267478 I/System.out: | state=R schedstat=( 5812106631 1041760011 2536 ) utm=535 stm=45 core=0 HZ=100 | stack=0xff7cb000-0xff7cd000 stackSize=8192KB | held mutexes= "mutator lock"(shared held) at com.tomerpacific.anrdetection.MainActivity$onCreate$1$1.onClick(MainActivity.kt:18) at android.view.View.performClick(View.java:7448) at android.view.View.performClickInternal(View.java:7425) I/System.out: at android.view.View.access$3600(View.java:810) at android.view.View$PerformClick.run(View.java:28305) I/System.out: at android.os.Handler.handleCallback(Handler.java:938) at android.os.Handler.dispatchMessage(Handler.java:99) at android.os.Looper.loop(Looper.java:223) at android.app.ActivityThread.main(ActivityThread.java:7656) I/System.out: at java.lang.reflect.Method.invoke(Native method) at com.android.internal.os.RuntimeInit$MethodAndArgsCaller.run(RuntimeInit.java:592) at com.android.internal.os.ZygoteInit.main(ZygoteInit.java:947) This is only a partial snippet of all the output, but you can see that it provides a ton of information, including: - Free memory/Total memory/Max memory - Heap diagnostic (percentage free, size and amount of objects allocated) - The stacktrace of the main thread Other useful methods include: - getTimestamp - the timestamp of when the process terminated - getDescription - a system description of why the process terminated ## Be Resourceful If your application does suffer from ANRs, solving them can be quite tricky. Whether it is not getting a complete stacktrace (or not having one at all), not being able to reproduce it or it happening on some esoteric device. Well, what can you do? In Android Studio version ≥ 3.2, you have a utility called CPU Profiler. This tool lets you inspect your thread activity during the runtime of your application. With it, you might find out which threads are running, for how long and where they are running. To use it, inside Android Studio, go to View → Tool Window → Profiler A window will open at the bottom of the screen and once you attach a process to it, you will see three timelines: - Event timeline - CPU timeline - Thread timeline You want to focus on the Thread timeline to see if anything is out of the ordinary there. Each thread's activity can be identified by three colors: - Green - indicates that the thread is running or in a runnable state - Yellow - indicates that the thread is waiting for the execution of some I/O operation - Gray - indicates the thread is sleeping Hopefully, by now, you have gained some confidence in making your applications as ANR proof as you can. Using the tools and techniques listed above may help prevent your application's next ANR. You are welcome to check out some of my other articles below: https://github.com/TomerPacific/MediumArticles --- ### 7 Questions Product Leaders Must Ask to Discover the Right Problems to Solve URL: https://www.taboola.com/engineering/7-questions-product-leaders-must-ask-to-discover-the-right-problems-to-solve/ Last Modified: 2025-01-14 14:37:18 Anyone working as a product analyst or product manager has a crucial role in ensuring product success. A successful product addresses one or many problems, helping customers overcome obstacles or solving a problem they struggle with. That's why one of the initial steps in the product life cycle is defining the problem the product will solve. But how do we define a problem properly? — By asking the right questions. Analyzing without asking the right questions wastes time, although it's a common occurrence for most product analysts and managers, myself included. In my 3+ years as a Product Analyst at Taboola, I've developed a better understanding of asking the right questions and defining a problem statement that hits the mark every time. In this post, I'll share seven questions that'll improve your approach to defining problems, whether you're a product analyst or manager. ## Start by collaborating with the right people Albert Einstein once said, “If I had an hour to solve a problem, I'd spend 55 minutes thinking about the problem and 5 minutes thinking about solutions." He understood that defining it is most of the challenge for problem-solving, no matter the domain. That's why here at Taboola, product analysts and product managers work closely to define them. The two roles may have different responsibilities and may work in different business units, but they collaborate closely when it comes to problem identification in any project. To make sure our problem definition phase is 100% finished before moving forward in the project, I created a question toolkit. It uses very specific questions to get to the heart of the obstacle the project aims to solve. ## Use these 7 questions to better define problems Let's go over the seven questions using a classic pain point we face at Taboola all the time: the sheer volume of data our platform generates. The Taboola platform creates millions of content recommendations daily based on a machine learning algorithm that fits each user's interests. We gather massive amounts of data per user to train the model to predict a user's click. ### Question 1: What's the goal of the analysis? The first pain point of handling all this data is the cost. It can be expensive to store every piece of data, so we need to know if the value of the data is higher than the storage costs. By analyzing user segments, we'll discover there's no value in storing more than a certain number of segments per user. If that's true, how can we discover that threshold number? In this scenario, the goal of the analysis is to estimate the impact of limiting the number of segments per user on Taboola revenues (which take storage costs into consideration.) The lower the storage costs, the higher the revenues. ### Question 2: What decision do we take based on the analysis? Answering this question helps focus on key insights. It is even more important to understand the potential decision-making because it's hard to decide with no significant analysis results. In this scenario, a possible decision is setting a threshold of maximum data Taboola stores per user. ### Question 3: What metric can support the chosen decision? Your decision needs supporting metrics, so you need to pick the right ones that support your chosen goal. In our scenario, you could conduct A/B tests and compare each group's percentage revenue per user. ### Question 4: What are the optional values and benefits of the analysis? If your questions have offered measurable values that are clear to you and your product team at this stage, congratulations! That's pretty rare, in my experience. Over the years, I've found it hard to expect every analysis to deliver clear and measurable results that show how to proceed going forward. Sometimes the measured value is only one part of the overall project's value and may be ignored by product managers or other business leaders. In our scenario, we might find that an optional value of the analysis is that we can reduce data storage costs by $10,000/month, making it likely that it'll be included in future project planning. However, if we only found it reduced costs by $3,000/month, that data might be ignored. ### Question 5: Who are the stakeholders? Knowing how the stakeholders are for a problem is essential as it'll dictate the most relevant metrics and can inform how the project is defined later. For example, customers, business leaders, and particular roles in particular business units may be stakeholders for one problem, but only customers may be relevant for another. ### Question 6: How accurate should the results be? Sometimes, complete accuracy isn't achievable, so you've got to decide how to proceed based on the accuracy you have. Do you need high accuracy, or can you proceed with low accuracy and still get the same results or answers? In our scenario, you may not need to analyze all the stored segments. Analysis of a portion of it might give you a representative answer, and your conclusions will probably be the same. ### Question 7: Is this a one-time analysis, or will you do it regularly? A Taboola dashboard can be a great way to support decision-making, but you don't do the same analysis every time, right? There are plenty of one-off decisions and analyses you do, like if you're validating a new product or ideating a new feature. If you choose to make a dashboard, be sure to include the required time you'll need for it in your planning. In our scenario, you decide not to create a dashboard in Phase 1 of the project, but you know you'd like to monitor the savings more regularly later on. So you set up an A/B test in Phase 1 and track the results over time in a dashboard. ## Final thoughts Generating actionable insights throughout the product life cycle is key for product analysts and managers. Here at Taboola, product analysts and managers often collaborate early to hone in on the right problems that will have the most impact for stakeholders. This post outlined the seven questions you can ask to help your product team focus on those valuable problems to create products that lead to more success. --- ### Product Analyst's Disneyland: A User's First Session URL: https://www.taboola.com/engineering/product-analysts-disneyland/ Last Modified: 2025-01-14 14:37:18 Newsplace, the news app I'm working on, was born 2 years ago. Just like a newborn it was constantly growing with new features and ideas and I met it when it already had tens of thousands of users on the one hand, while still searching for its core value proposition on the other. Currently the app offers a stories experience and a content categorized feed of aggregated news items which lead the user to the publisher's website, where they can read the full article they saw in our app. As a junior product analyst, for me Newsplace seemed like Disneyland, since it was in its early stages and I could influence the core value it offers users . Despite that excitement, I didn't know how to expand my touchpoints with the product beyond the weekly value proposition meeting when a whole history existed before me and there was (and still is) a strong and talented team which was already running things, let alone to be proactive with initiatives of my own. While retention and engagement are the obvious things to optimize when trying to increase LTV, they are big ambiguous metrics and trying to improve them was too vague of a goal for me at that point. Together with Ran, our Product Manager, I found out that breaking down these words into the journey steps that users go through in our product, and choosing only one step to focus on, makes things much more clear and exciting. The step that got me excited was a user's first session. A user's first session with any product is critical to the value they'll eventually get from it. This obvious fact quite literally hit me when I opened Tik Tok for the very first time and I HATED IT! Without onboarding, without easing into the app and what it holds, one innocent click on the TikTok icon led to a violent hijack of my precious phone with a full screen video combining music. For me, the instant urge was to shut the app down and delete TikTok from my phone forever. The only reason I continued watching videos was the initial reason I downloaded it in the first place- to finally understand what my colleagues are talking about when they are referring to the experiences in TikTok. But like Newsplace, most products don't have the luxury of retaining users beyond the first session solely on their solid reputations, and in order to do so, we need to optimize for the best user experience possible. This moment - a user's first session- in which I found the ownership that I wanted to take upon myself in this product was the moment I started to feel passion and excitement and most importantly, started to love this product. ## A user's first session So once I had a clear and bounded playground - a users' first session, many questions could be asked to optimize the user experience in it: - What does a 'good' first session look like (one in which users discover the value proposition and return to)? What is its flow? - What is the biggest moment of disappointment a user encounters? (like the intimidating feeling I had from opening TikTok for the first time) Is it a long loading screen? Many clicks to get somewhere? A general feeling of being 'lost' in the app? Lack of content? Cognitive overload? - What is the last action a user does before churning? As the reference to my TikTok experience might suggest, I decided to focus on the reasons that cause users to churn, specifically in the first session.(In) To find these reasons, I worked together with Ran the PM and used Mixpanel to create a cohort of users who churned on their first day, in order to look at their flow in the app and hopefully discover what went wrong- what was the last thing they did/saw in the app that made them leave (or worse- hate us). That cohort with the flow chart were my ticket to the Newsplace Disneyland. ## Diving into the data Digging through the flow chart was a great opportunity for me to get my hands dirty and to really understand how and when we report each event. Things that initially didn't make sense such as a 'visible story' event before the impression on the first screen in the app, became the chance to sit down with Amit, The Mobile R&D Team Leader, for the first time, and get explanations about the technical way these screens are loaded. After eliminating irrelevant events that are sent automatically or those that don't reflect the actual progress of the user in the app, something super interesting caught my eye- I noticed that 17% of users who see the user selection screen (which soon will be vastly elaborated upon) drop off. Additional 33% drop off after exiting the app while this screen appears or doing some other esoteric events that come before the stories. That means that more than 50% of the users who churn during their first day, do so during the first screen they see. 50% (!) of users leave us forever before even taking a sneak peak at our value proposition! The second time I came to Amit with questions was much shorter as it came from a problem that was very easy to spot- a very long loading time for this simple screen that enables a user to choose which content they'd like to read today- top news or lifestyle content. This first screen that caught my interest so much was added when Our UX Designer, Naama, came up with the idea for it when a user test she conducted revealed that the mere choice of content type showed an uplift in the ratings that users gave to the app. After this idea turned into an actual screen and was rolled out in the latest version- there was an uplift in other metrics as well which aligned with the conclusions of the user test. There was an increase of 13% in meaningful interactions and an increase in time spent in the app (in Mixpanel) among users who actively made a choice (clicked on either news or lifestyle) over those who were auto-transferred to the news. Despite the fact that this screen was intentionally shown for a long time to allow the app to load the stories, we failed to measure the friction this feature created and its impact on churn. This revelation brought up very interesting questions - is the user drop worthwhile for us? Do we get more engaged users afterwards thanks to this feature? Does this screen increase engagement only on the first session(s) or does it also affect retention? Does this screen increase our KPIs only when a user is actively choosing the content or is just encountering it is enough? If we find out that it contributes to engagement but not retention- is it worth optimizing and if so- what to change in this screen? How to prioritize the ideas? If the user selection screen doesn't increase measures- how do we integrate choice of content in another way with less friction? Since I wrote the first draft for this post, many actions have been done to answer these questions, including re-evaluating and altering our north star metric, However many questions are still left to be explored. For me, this screen which I had no contribution to up until this point, was the gateway to the team, to being able to work independently in a proactive manner and to love this product. --- ### Practical Guide to Collecting Feedback and Data URL: https://www.taboola.com/engineering/practical-guide-to-collecting-feedback-and-data/ Last Modified: 2025-01-14 14:37:18 Today is the big day - you're finally launching an exciting new feature that you've worked on for the past few weeks. The feature's released, so you test it yourself, and it's working! Customers will be able to use it now! But soon you begin to wonder How many people have started to use it? How many times? How did they use it? Did they get any errors? Does it save them time? Does it improve the bottom line? The last question requires further analysis and diving deeper into the data, but the first ones can be answered by looking into usage data and asking users for feedback. As a UX lead, who's also a UX researcher, it's extremely important I understand the impact each feature has on the user experience and how our users interact with our platform. When we add new features, we only have assumptions on how they'll be used, so we need to check if they are true after they're released. This post will describe how we get user reaction data about our platform by using these four recommended channels for product feedback and usage data gathering. ## Google Analytics Usage Data We first need to measure our platform before we can know what to improve. We've implemented Google Analytics tags across our platform, and each UI component fires an event for each user interaction. Google Analytics can help us answer these questions and take appropriate action: - What are the most common form errors users get? We are continuously analyzing top errors and taking action to resolve them. - What are the most common UI selections? For example, by looking at the usage, you can reorder the items in a drop-down by popularity and make the UI more intuitive and easy to use. - How long does it take to complete a task? We track the time it takes to complete tasks and the success rate and try to improve them. - How fast is the application? How long does it take to load pages? We try to improve the page or report load time if it's too slow. - How many people interacted with a feature, and how often? When we add a new feature, we can measure how often people use it and adopt it. Here's a Google Analytics report example: ### How we collect Google Analytics data When we created our new platform, - Taboola Ads, we mapped a tracking plan with a Google Analytics expert and implemented the events into the platform code following this structure: - Category: page name - Action: UI component - Label: user interaction - Custom Event Value: what the user has selected For example, this is what is sent when a user selects “last 30 days" in Taboola Ads: - Category: Day Report - Action: Date Picker - Label: Date Preset - Custom Event Value: Last 30 days Implementing events across an entire application can be time-consuming for the UX and development teams, so I have a suggestion: Instead of trying to map everything in a structure that seems reasonable and makes sense, start with only the basic usage questions you have in mind. This way, you'll avoid over-complicating things and gathering data on events that no one will ever look at. You can always add event codes later to answer more specific questions. ## In-App Surveys In-app surveys are a great way to connect with your users and involve them in improving your product. The benefit of using in-app surveys is that you're connecting with users in the right context while they're already using the app, giving you more valuable insights and higher response rates. There are many tools for creating in-app surveys, such as Typeform and Momentive (formerly SurveyMonkey.) At Taboola, we use Hotjar to gather user feedback since it also has an easy-to-use survey feature. Here are a few things to consider when surveying users: - The first question will have the highest response rate, as there is a natural drop-off after each question, so ask your most important questions upfront. - Instead of asking a general question, ask more specific questions about each feature separately. The answers will give you more actionable insights. - Keep your survey as short as possible. Here are a few examples of questions we can ask: - How satisfied or dissatisfied are you with ? Answers can range from “Extremely Satisfied" to “Extremely Dissatisfied" on a scale from 1-7. - How simple or complex do you find it to perform ? Answers can range from “Very Simple" to “Very Complex" on a scale from 1-7 - I have everything I need to create X or to do Y. Answers can range from “Strongly Agree" to “Strongly Disagree" on a scale from 1-7 It's always good to add an open-ended question at the end, such as “Is there anything we can do to improve your experience?" or “Do you have additional feedback or ideas on improving feature X for you? This allows people to give additional comments, suggest new ideas, or air any frustrations they might have related to this feature or product. ## Open Feedback Channels In addition to promoting surveys on-demand or pulling usage numbers, it's also helpful to provide users with an open channel for feedback. Internal Slack Channel If you use Slack at work, a dedicated Slack channel for the product or feature can be helpful for internal users to send questions, report bugs, ask for new feature requests, and suggest improvements. This channel can also be a way for you to communicate new releases and upgrades to people. Hotjar Feedback Widget If you want to hear from external users and find out what bothers them in real-time, you can use a feedback solution on your product in production. Since we use Hotjar, we've implemented a Hotjar feedback widget where Taboola users can tell us their pain points, bugs, or whatever else they want to share. Users can also add their email addresses so we can reply with valuable feedback or bug reports and ask clarifying questions as needed. # ## Additional Channels Besides using online analytics and surveys, there are other channels users can send their feedback. User Interviews It's always interesting and insightful to meet users and understand how they use your product, and learn what difficulties they have that your product can solve. We use this often at Taboola as it helps us learn about our customer's challenges and helps us engage more with our users. Support TicketsSupport tickets are another channel that can tell you a lot about your product. Use them to find out what product areas need improvement, what's broken today, and what users misunderstand, so they have to contact support. Quality Dashboards Users want fast applications, so it's essential to monitor application speed and know how many times you've had downtime or slow server responses. These are more technical monitoring tools but provide another perspective to the overall user experience. ## Wrapping up A good mix of quantitative data (like Google Analytics) and qualitative data (surveys and interviews) provide a better picture of your product experience and usage. However, only looking at the numbers won't tell us what the user experience is like. Those are important but are only one part of the overall picture. Imagine a friend coming back from a live music concert, you ask them how it was and they replied, “It was 2.5 hours long, I danced for 32 minutes, drank two beers, and ate one hotdog. The drive home took 52 minutes. These are only the numbers, but what was it like? Did they have a good time? Can they describe it on a scale of 1 to 10? We'll only know by speaking to them. So, if you're looking to collect practical data and feedback about your product, you can pull usage data, but don't forget to ask for feedback too! And don't forget to share it with the entire product team! Everyone will learn more about the value of the new feature you just released, whether it's being used, and how well it's been received. --- ### 5 Product Management Lessons You Can Learn from Ted Lasso URL: https://www.taboola.com/engineering/lessons-you-can-learn-from-ted-lasso/ Last Modified: 2025-01-14 14:37:19 Every once in a while, you stumble across a TV series that reflects aspects of your personal and professional life. That's what happened when I saw Apple TV+'s Ted Lasso. Ted Lasso is a spin-off of a priceless and hilarious NBC commercial about the life of Ted, a fictional American football coach hired to coach an English Premier League football club in London. The critically acclaimed show, starring Jason Sudeikis as Ted, has racked up a ton of awards. https://youtu.be/3u7EIiohs6U It took me 4-5 episodes to “get it" but eventually, I had this big a-ha moment coming. Being an American football coach training an English soccer club is no different than being a product manager. Let me explain. A product manager and a head coach are both leaders, mini-CEOs of their product or team. Both positions need to manage many moving parts and pull team members together to achieve a common goal. The goal and the way success is measured might be different - a product manager needs to deliver key results, and a coach needs to win games - but the process of getting there is much the same. To win a game or resolve locker room conflicts, Ted Lasso had to earn the trust of his players and the other club stakeholders, bringing them together to achieve a mutual goal. The same goes for a product manager. No one just follows you because you're the product manager. You need to earn the trust of your fellow engineers, account managers, sales and marketing teams, and others. Let's look at some examples of how Ted earned that trust and how you can apply the same strategy to your job as a product manager to achieve the same. Because I faced many of the same struggles as Ted, I'll also offer my real-world product management examples to make the lessons more tangible. ## Lesson 1: Be kind - it's a challenging time for everyone ### Season 2, Episode 10 “No Weddings and a Funeral" “And I knew right then and there that I was never gonna let anybody get by me without understanding they might be hurting inside, you know. 'Cause life, it's hard. It's real hard." - Ted Lasso Ted Lasso talks to the club's psychologist about his father's tragic death in the episode. He then explains, plain and simple, why he's exceptionally kind to everyone, attributing it to his understanding that life is complex. The takeaway: It's a challenging world out there, and the world is getting crazier and crazier. We're fighting a war, pandemic, climate change, increasing living costs, heavy commutes, or Zooming all day long. Your peers, like you, need to perform under an increasing amount of pressure and stress. While managers need to maintain a professional environment, we need to realize that people are human, and we don't always know the other issues they are coping with. A real-world example: I am an extremely nice person. Sometimes, my wife calls me naive and reminds me that there are people who will exploit that. When I joined Taboola, I decided to be crazily kind. I started this as an experiment because I wanted to prove that being nice pays off. I came across an account manager who bombarded me with questions. I knew this person wasn't trying to ruin my day and that they were under pressure to achieve their goals too. The answers I could provide would clarify the issue they were working on, so why not sit with them and help? Of course, I know you can't help everyone all the time as you have your own tasks, but you never know how an extra act of kindness will pay off. I can tell you that that account manager was grateful, and when I needed help securing meetings with clients, they were there for me. Showing people compassion in their moment of need is always the way to go. Source: AppleTV ## Lesson 2: Listen, hear others' opinions and echo them ### Season 1, Episode 7 “Make Rebecca Great Again" "Ted Lasso: Thank you. Also, I read through your thoughts. They're great. And I agree with every last one of 'em. But I can't say this to them. Nathan Shelly: But they need to hear it. Ted Lasso: I agree. That's why you're gonna do it. Nathan: What? Ted Lasso: You're givin' the pre-game talk. And you're readin' em this. C'mon, it'll be fun." In this scene, Nathan, the team's quartermaster, leaves Ted a note with tips each player should hear before a crucial away game. Unlike Ted, who is new to the team, Nathan is a diehard fan and knows each player's strengths and weaknesses, giving him a unique perspective. Knowing this, Ted insists that Nathan reads the letter to the players himself in the critical pre-game talk, usually a coach's duty and one of the most ceremonial things any coach does. By deferring this honor to Nathan, Ted empowers and uplifts him, even though he is “just" the team's housekeeper. The takeaway: Leave space for others to speak their minds and keep ownership of their ideas. Yes, you're the product manager, the “CEO of the product" but you don't know everything. Someone else's perspective could completely turn things around. You should leave room for others' opinions, and when you spot a chance, give them the space and the spotlight to express themselves publicly. You'll be amazed at how this empowers the person, builds confidence, and in the long run, improves the person's work and the people around them. A real-world example: At Taboola, I had a similar experience and opportunity with a dedicated employee from the India office. Formally, he worked in our Professional Services department, serving as the first point of contact for a customer in need - but he wanted to do more than that. He excelled at seeing the bigger picture and identifying valuable feedback from our customers because he looked at things through a different lens, one that was colored by the fact that he was fluent in mobile development. He identified a procedure we could amend, reducing a specific integration phase time. We then gave him the room to announce this change, present the solution, and train the other team members. Taboola earned a more motivated employee, who later also got promoted. Source: AppleTV+ ## Lesson 3: It's the small things that make a difference ### Season 1, Episode 2 "Biscuits" This one played out during an entire episode, so I'll summarize the situation. Ted and his assistant, Coach Bird, put a hand-crafted suggestion box in the locker room to give players a chance to anonymously share their suggestions and voice their complaints. Most of the players used it as an opportunity to call the coach “wanker," only giving a negative response to this “too nice for men" method. But on one note, a of the player complained that the water pressure in the showers was weak. It was a small thing and had nothing to do with the game, but Ted and Coach Bird took care of it, earning some important and initial trust from the players. The takeaway: Regardless of whether you're joining a new team or have been working with the same group for years, don't assume things are perfect as they are. There are often easy solutions to minor problems, and solving them can have a significant impact. However, if team members don't have a way to comfortably point them out, they will fester and play out in more detrimental ways. A real-world example: When I started working at Taboola, I amended an operational workflow that drove one of the experienced engineers crazy. Unfortunately, not everyone on the team could add a new component to our JIRA dashboard. Thinking there must be a solution, I reached out to a coworker on the Information Systems team. We tracked down the right person and found a way to add the new components. The new process made our working environment more straightforward for everyone, and I'm sure solving the problem scored some goodwill from that engineer. Source: AppleTV+ ## Lesson 4: Make sure you're working as a team ### Season 1, Episode 4 “For the Children" “When it comes to locker rooms, I like 'em just like my mother's bathing suits. I only wanna see 'em in one piece." - Ted Lasso There's a noticeable and ongoing conflict between two key players, and they often argue in the locker room. The young and gifted Jamie Tartt can't seem to get along with team captain Roy Kent, an older player in the final years of his career. The men's disdain for each other creates negative energy, affecting the whole team and causing them to perform poorly. It takes Ted some time, but eventually, he finds a way to bring them together, completely changing team dynamics for the better. The takeaway: Personal differences are inevitable, but they need to be dealt with when they begin impacting the team, work performance, and atmosphere. Meddling can backfire on the person trying to be the peacemaker; however, resolutions are possible if the situation is approached with thought and care. The greatest risk is when an imbalance occurs, disrupts the team, and is left unaddressed. A real-world example: If the situation is dire, I suggest taking a formal approach and reporting the issue to a supervisor or HR. But another option I've personally found helpful is to address the situation in an informal, non-threatening environment. I've taken team members out for coffee or lunch and, without going into detail, explained that the current situation was palpable and interfering with the team's goals. Don't take sides. Instead, focus on delivering the message to both peers and make sure they understand they are both responsible. In my experience, once it's pointed out, adults realize how much their petty differences are interfering with the team, and they'll find a way to work it. Source: AppleTV+ ## Lesson 5: Believe in BELIEVE This lesson unfolded over both seasons. One of the first things Ted did when arriving at Richmond AFC was to hang a sign with the word “BELIEVE" in the locker room. The motivational sign is meant to inspire them in the short-term to win a game and in the long term to achieve any dream. Believing in BELIEVE is a core value for Ted. He passionately rails against those who lack faith in his team throughout the show, and he never gives up because he believes in BELIEVE. His attitude is contagious, and by focusing on this message, his team fights to achieve the same goal - winning. The takeaway: The same lesson can be applied to a sports team, product team, or any other team. Everyone must abide by company rules and serve the company's needs, but if the team is bonded as they work towards a mutual goal, each person feels they have a stake in the outcome and something to work towards. A real-world example: In my early days at Taboola, I was part of the mobile team. As a product manager, I needed to bring people from different departments (R&D, marketing, product, and professional services) to achieve a common goal. I found it helpful to remind everyone of the technical and business milestones we needed to accomplish at daily meetings and during monthly planning meetings. Keeping the goal front and center and making everyone believe in believing they can achieve anything is what led to our success. ## We can all learn from Ted I believe that a product manager's heart and strength are working with people and driving them. Implementing the latest product methodologies can help, but the people you work with will be the ones who help you move the ship. Consider onboarding and applying these lessons fully, but if not, at least try to understand their roots, the conflicts they represent, and the fact that having the right attitude can make all the difference. You can follow in Ted's footsteps and solve problems the Lasso way, or you can come up with your own solutions. While I revealed some things about the show, and there are some spoilers here, there's a lot more to learn from the fearless Ted Lasso. Go watch the show! --- ### How to Kill Underperforming Features and Why You Should Do It Now URL: https://www.taboola.com/engineering/how-to-kill-underperforming-features/ Last Modified: 2025-01-14 14:37:19 ## Can you create value by not building features? In this blog post, I will share why it is important to remove underperforming features and how my team systematically removed such features to dramatically improve the product's key metrics. This process helped us grow our Day 1 retention rate by over 20% while reducing the number of features we support. But it wasn't always like that.. ## The Capped Potential Problem At the beginning of 2021, I took ownership of a legacy mobile app that displays personalized news on the lockscreen of mobile phones. Over the years, the product developed a strong presence, mainly in Latin America, and currently has over 8M Daily Active Users receiving daily news updates from leading publishers directly on their lockscreen. Until then, our product team had been consistently launching new features, but they weren't having a significant impact on our KPIs. The idea that the product's potential is capped has become "common wisdom" in the organization. This is the ultimate frustration for every product team, but the problems didn't stop there… ## “We have a problem with Hungarian" During our release process, our QA team found a bug when users switched their device language to Hungarian.I was surprised as I didn't realize the app supported Hungarian. I was even more surprised to learn we supported over 36 languages that were used by only 0.9% of our users. I didn't think that fixing the problem would actually make a difference. Instead of fixing the bug, I suggested we remove the redundant languages and see what happens. What “happened" amazed me - our guess was right, users did not complain.What surprised us was that our app size decreased and our product performance improved. But then it made me think... ## “What will happen if we remove more features?" At first, the mere thought of removing something the team had worked so hard to build was scary. In fact, when thinking about removing features we usually have many excuses, such as the team's investment or that the request came from a specific customer. So maybe we should just keep them? ## Why keeping underperforming features is bad for your product Obviously our beloved features aren't performing as we expected, but why should we get rid of them? We've already made this effort, and we can even see users using the feature in our analytics platform ("Finally, after 5 years, we have early adopters!"). Deep in our hearts we know that this feature isn't strategic in any way. We can't justify investing more in it, because we know it won't move the needle. This is exactly the point where, by not removing the feature, we're damaging our product. First of all, the 1% that actually use the feature are going to suffer, mostly from a lack of attention. But the 1% by definition should be the least of your problems, you should worry about the 99% who “just don't get the feature". Having an underperforming feature will constantly hurt your core user segments. The damage derives from both the direct and indirect costs of keeping this feature alive. Let's take a closer look. ## The Direct & Indirect Costs of Underperforming Features The direct costs of keeping an underperforming feature are usually obvious to the product team, but might not be measured. For example, if there's a minor bug, we'll fix it. But sometimes that time spent fixing can pile up into days and weeks that could have gone to more important features. In my opinion, however, the direct costs are just a small tax to pay, compared to the indirect costs of having underperforming features. The first and most important cost is the lack of strategic focus. The lack of strategic focus makes the product harder to sell, longer to implement and makes it slower to be adopted by users. You also need to account for the maintenance costs of keeping these features “future-compatible" with every improvement, refactor or even external change you make to the product. will require you to adjust these features as well. Okay, but how do you know which features to remove? ## Finding the Deadweight in My Product Now you realize that having underperforming features comes at a (very high) cost. The hard part is that you now have to choose which features should go. In this section I'll share our methodology for making the difficult decision to remove a feature. The decision is difficult because it's emotionally tied to the level of investment the team has poured into this feature, otherwise known as your Sunk Cost. To overcome this bias, we propose using the same framework of prioritization to de-prioritize your underperforming features. For example, at Taboola we use the ICE methodology to prioritize new initiatives based on a 1-10 ranking for Impact, Confidence and Ease. We compare the ICE scores of all features in the backlog to decide on the priorities. ## Reverse Engineering ICE The same framework can be applied to reverse engineering our selection process for removing underperforming features. Since these features are already live, we usually have much more confidence in the impact they are (not) making. The R&D team can then estimate the complexity of removing these features. It's important to gather your stakeholders and enable them to share their views on your prioritization. By systematically applying this process we'll have a prioritized list of features that we can then start to remove from bottom to top. But where do we start? ## The Usual Suspects The “usual suspects" are often found in areas that are not in your product's “golden flow", or the path users receive value from your product. - Settings: In this area we're likely to find things we developed “just in case" or as a response to a very specific use case. Our goal here is to determine whether these use cases are frequent enough to justify their existence. This can be achieved by either your product analytics platform, or by looking at support tickets and feature requests in that area. - Legacy Customers: Look at your customer list and see if you have built custom-made features for old customers. Are these customers active? Do they still represent your core user segment? - Analytics Events: Go to your product's analytics platform and extract the events that have had the least usage in the previous 3-6 months. Do these events originate from specific areas of the product? These areas could potentially host underperforming features. When applying this approach, we identified many features that were just “sitting there" waiting to be removed. ## When to Iterate and When to Kill Even after we've gone through the entire process, and identified potentially underperforming features, we still had a difficult time deciding to remove these features. This is because we have our inner product voice that says to us: “Maybe just another iteration could turn this around". The dilemma of when to iterate and when to kill a feature is something that every PM comes across at least once in their product career. My suggestion to solve this dilemma is to use the same method we use for setting our goals, - the OKR method. In a nutshell, we define ambitious and measurable goals every quarter and rank them according to our performance with 3 colors. When I think about whether to iterate a feature or not (regardless of if I defined it as underperforming or not), I ask myself what could've been the best case scenario for this feature? To be more precise, if the feature worked like magic (exactly as I planned), by how much would it drive my metrics up? 5%? 10%? more? After doing this hypothetical exercise, which I usually do before building the feature, I go back and measure the impact the release has actually had. Going back to the OKRs: - Red Zone (0%-40% of my goal) - additional incremental iterations won't work - Yellow Zone (40%-70% of my goal) - good chance I'll get there with further incremental iterations For example, let's say I build a new sign up form and expect my conversion rate to grow by 10%. - Red Zone Scenario: After the release we see that we've only grown by 3.5%, which is 35% of my target of 10% - I would rather try out a new concept than iterate on the existing one - Yellow Zone Scenario: We managed to grow our conversion rate by 6.5%, which is 65% of my target of 10%. I am much closer to my intended goal and will decide to iterate on the existing concept The same concept could be applied to underperforming features as well. We can estimate the best case scenario for the feature's adoption and then measure the actual results of the feature after it's been released. If we don't see it moves the needle for them, it's definitely not worth another iteration. ## Wrapping Up In this article we discussed why keeping underperforming features could damage your product. We shared where and how to find the usual suspects for such features and our method for solving the “Iterate or Kill" dilemma. Since applying this process, our product team has carefully evaluated which features we can remove and which we must keep. This has led us to reduce the product's scope by 25% and has subsequently improved our key metrics, such as our day 1 retention by over 20%. As Product Managers we strive to bring more value, but we sometimes get confused between delivering more value and delivering more features. Or to paraphrase a famous Bill Gates quote; Measuring programming progress by lines of code is like measuring a plane's quality by weight. --- ### Attentive Audiences: Taboola Cuts Median CPA by 34% and Boosts Median CVR by 87% URL: https://www.taboola.com/engineering/attentive-audiences/ Last Modified: 2025-01-14 14:37:19 Advertiser success is one of Taboola's main goals as we want to help our advertisers gain the most out of their investment with us. Two metrics advertisers use to measure that are Cost Per Action (CPA) and Conversion Rate (CVR). A high CPA means they're spending too much for each audience action and a low CVR means they're not getting a good return on that investment (ROI). We wanted to see if there was a way Taboola could enhance our solution to make it easier for advertisers to lower their CPA or increase their CVR. That's why we created the Attentive Audience program. It's a new way that Taboola collects, aggregates, and filters audience data to be more meaningful for advertisers looking to improve these two metrics. Keep reading to discover how we developed the program and rolled it out to select advertisers. ## The basics of the Attentive Audiences project Before we get into the details of the project, let's go over some relevant definitions and concepts. Attentiveness Metrics: These metrics measure user attention on the advertiser's website and predict the user's likelihood of converting on the advertiser's website. They're based on signals like time on site and the number of returning visits, etc.. Attentiveness Event: A request from an advertiser's website to Taboola's servers containing a distinct name and attentiveness metrics value. Conversion and Attentiveness Conversion: In Taboola, conversions are based on specific rules and aligned with the advertiser's business goals. Each type is unique to each advertiser and are set in the conversion and attentiveness conversion rules, respectively. Attentiveness Rule: This rule is a unique set of conditions based on the attentiveness metrics set for each advertiser. For example, time on site > 30 seconds and number of returning visits > 2 visits Conversion Rate (CVR): The total number of users who converted divided by the total number of visitors to an advertiser's website. Cost Per Action (CPA): The amount of money an advertiser pays for a single conversion. It's calculated by dividing the total budget spent by the number of conversions. ## Our goal for the project Our business goal for this project was to increase CVR or reduce CPA, as they are both dependent on the total number of conversions. We planned to do it by offering advertisers a unique retargeting group of attentive users who have a higher chance of converting on their website. ## How we implemented the project To create a unique group of attentive users for each advertiser, we set up a dedicated workflow to handle the data collection, digestions, aggregation, filtering, and presentation. Project Architecture ### 1. Collection We used the Taboola Pixel to collect the users' required session-related attentiveness metrics when they visited an advertiser site. The data was then sent to the Taboola servers by an attentiveness event at a defined time interval. ### 2. Digestion Attentiveness events go through two filters on the Taboola server-side: rule matching and user attribution. These filters are used to find only events that matched the advertiser's attentiveness rule by users that reached the advertiser's website through one of Taboola's publisher partners. All other events are ignored for this workflow. ### 3. Aggregation The attentiveness conversions are then written to a Vertica cluster for further aggregation. The data aggregation is performed offline daily and looks for two types of data per advertiser: - Basic data, like the number of conversions, attentiveness conversions, clicks, and so on. - Performance metrics, like attentiveness conversions that preceded other non-attentiveness conversions (precision) and for the percentage of non-attentiveness conversions that came after attentiveness conversions (recall). ### 4. Filtering By filtering the aggregated data, we're looking to find which advertisers would benefit from targeting attentive audiences. More specifically, we're looking for audiences that would have a high positive impact on the advertiser's campaign goals. We perform several basic and performance data verifications, including: verifying the thresholds for the number of attentiveness conversions and the percentage of attentiveness conversions and the percentage of attentiveness conversion precision and recall. The most important filter happens at the end of this stage by verifying multiple-day results of the earlier filters to determine the attentive audience viability group for each advertiser. If the data passes this filter, Taboola offers the Attentive Audience enhancement to the advertiser. ### 5. Presentation Advertisers that are offered Attentive Audiences find it in three areas of the Taboola Ads campaign management platform. #### As a recommendation It will appear as an option for the advertiser on the Recommendation screen. #### When setting up a campaign Attentive Audiences is a targeting option when setting up a new campaign: #### When viewing their audience types It appears as an audience type on the Audiences screen. ## The Attentive Audience project results We currently have thousands of advertisers using Attentive Audience, and they've enjoyed significant success. Advertisers enjoyed a median CVR increase of 87% and a median CPA decrease of 34%. Here are just a few examples of advertisers who enjoyed great success with attentive audiences: - VAHA doubled their CVR across all channels and increased site visit engagement by 338%. - Pionier Developers increased their CVR by nearly 44% and decreased their cost per lead by 35%. - Look After My Bills lowered their CPA by 60% and gained 10,000 new memberships in nine months. - The Motley Fool Canada decreased CPA by 11% and generated an average of 25,000 leads per quarter. We also noticed two main drawbacks to the Attentive Audience project: only a few advertisers qualified for it, and we couldn't target audiences outside of advertiser site visitors. ## Areas of improvement for the Attentive Audience project The project or feature is only available to a small portion of Taboola advertisers in its current form. That means most of our advertisers don't have access to the feature and cannot increase their CVR or lower their CPA through Attentive Audience. We'll be looking at ways we can offer it to more advertisers, but for now, it's only available to a select few. Secondly, the project only looks at an advertiser's website visitors, but Taboola offers many more audiences than that. We need to find a way to discover the users who have a high chance of converting but haven't necessarily visited an advertiser's site first. As a first attempt, we are pleased with the Attentive Audiences project. It was a good exercise for our engineering teams to learn how to help advertisers in a specific way and reach more of their business goals. We were able to find, aggregate, filter, and present relevant targeted audiences to advertisers so they could hit more of their business objectives, like increasing conversions, lowering advertising costs, and growing revenue. We look forward to enhancing the project in the future to scale it up to the level where we can offer it to all our advertisers. --- ### Dynamic Security Operations in Kubernetes URL: https://www.taboola.com/engineering/dynamic-security-operations-in-kubernetes/ Last Modified: 2025-01-14 14:37:19 In a dynamic world, your Kubernetes security infrastructure should also be dynamic. Many Kubernetes applications only add security features after deploying or are ready to be deployed. No wonder 55% of application rollouts were delayed, and 94% experienced a security incident last year. Kubernetes actually offers several security mechanisms that could be provisioned, such as NetworkPolicies and PodSecurityPolicies. Kubernetes NetworkPolicies acts like a basic firewall by controlling the traffic to and from each pod. Calico NetworkPolicies lets us extend Kubernetes network policies even further with additional security features such as referencing sets of IP subnetworks, creating non-namespaced GlobalNetworkPolicies, and allowing or denying external services access. The manual work of managing the sets of external IP addresses used by the Kubernetes deployments is complex. With the size of an environment like Taboola's, it's impractical and takes too much time. It can also introduce additional risks to the Kubernetes cluster. We wanted to see if there was a way we could sync our Kubernetes NetworkPolicies dynamically with tools we already use. Keep reading to see how we use Consul and Calico to do it. ## Dynamic security equals scalability For the Taboola Engineering team, scalability is a high priority. We manage seven Kubernetes clusters with more than 100,000 cores on-premise. Plus, we have more than 6,000 Physical Nodes and other VMs that run other workloads. Running infrastructure of this size means we're making dozens of changes per day, sometimes even dozens per hour. To keep the k8s workloads isolated and operational, we've got to update the NetworkPolicies regularly. Every deployment, auto-scaling, and maintenance task can lead to an IP address update in the NetworkPolicy. That's impractical to do manually with such a large Kubernetes deployment. We knew there had to be a way to do this dynamically, so we started investigating. ## Searching for the best dynamic security option In our infrastructure, we use Consul to discover services running outside of k8s. We had three critical requirements to use Consul and Calico to dynamically enforce security: - If an IP address is added to a Consul service, it should be added automatically to Calico GlobalNetworkSet. - If an IP address is deleted from a Consul service, it will be removed from Calico GlobalNetworkSet only after a defined grace period to prevent a flapping node. - The update process should not overload the Calico datastore or the Consul API. With our requirements set, we looked at three options to enable Kubernetes security with Consul and Calico. ### Option 1: A bash script and cron job combination We could run a bash script in a k8s cron job that uses Curl and Calicoctl command line tools to fetch the Consul service catalog and update Calico GlobalNetworkSet. ### Option 2: A Consul template and Calico command line combination We could create a Consul template for each Calico GlobalNetworkSet that would apply the updated GlobalNetworkSet whenever a Consul change was made. ### Option 3: A sync process using Consul and Calico APIs We could use Consul API to watch for service changes and directly update Calico GlobalNetworkSet with lid calico-go. ### The Winner: Option 3 We rejected the first two options as they didn't satisfy all our requirements. They both cannot delete an IP address from Calico after a grace period. They could also potentially overload Consul with too much LimboAPI traffic. We discovered that running a Consul to Calico sync process was the winning option to enforce dynamic Kubernetes security. It allowed us to add IP addresses dynamically while giving us the ability to add a grace period for all removals/deletions. It prevents bottlenecks on the API and scales easily, no matter the size of the Kubernetes deployment. To use this process, we started by running a blocking query for each Consul service configured for the operator. The process will store an updated list of IPs from both Consul and Calico. Upon every event from Consul, the process compares the IP lists in Consul and Calico by getting GlobalNetworkSet from Calico and the specific catalog from Consul. Then, before it makes any changes in Calico, the process will check: - If the IP address exists in Consul, but not in Calico. If not, it will add the IP to GlobalNetworkSet. - If the IP address exists in Calico but not in Consul, it will wait 30 minutes before checking again. If the IP is still not in Consul after those 30 minutes, it will delete it from GlobalNetworkSet. ## Real-time scenarios: How we're using dynamic Kubernetes security The dynamic approach we use eliminates manual operations and reduces human error while allowing better agility, scalability, and delivery times while reducing human error. Here are two scenarios that often happen at Taboola. ### Scenario 1: Scraping pods from outside Prometheus We use Prometheus outside our clusters to collect metrics from our pods. To do this, we need a secure way to allow incoming Prometheus traffic to scrape the servers for the data. We configure the new consul-calico-sync process for the Prometheus Consul service and the correlated GlobalNetworkSet. Now, any time the service catalog is changed, the relevant GlobalNetworkPolicy is updated dynamically so the Prometheus server can successfully scrape the pod. ### Scenario 2: Querying external databases Some of our applications that run on k8s regularly use databases outside the cluster. To allow this traffic, we constantly sync all IP addresses in the Consul database service with those in Calico. This way, we only allow traffic from pods with the allow_db label and reject everything else. In both these scenarios we see how the use of dynamic Kubernetes security negates human intervention, which saves time and reduces errors - both extremely important when we're considering scalability. ## Over to you And there you have it, a way to dynamically manage security in your Kubernetes deployments at scale. Try adding the Consul2Calico sync process to your next Kubernetes application and stop making manual IP address changes to save time and effort. For more details on how to run this solution in your Kubernetes cluster check out Consul2Calico on GitHub. --- ### How Taboola Powers the Conversion Data Pipe URL: https://www.taboola.com/engineering/how-taboola-powers-the-conversion-data-pipe/ Last Modified: 2025-01-14 14:37:20 When you set out to build the world's largest content recommendation platform that performs to high service level agreements, you need a lot of computing power, responsiveness, and reliability. That's why Taboola uses nine data centers worldwide, each with hundreds of servers. Users send and receive data from the closest data center to ensure the fastest and most efficient performance of the system. Behind the scenes, however, the Taboola system needs to access data across all the data centers to find, analyze, and serve up the right data at the right time. To coordinate all the traffic, data, and information, we use a single data center as an aggregation point for all the rest. This technology helps Taboola digest, transform, and aggregate the data at scale, so it's valuable to our customers and users. In this article, we'll explain what conversions are, how we handle billions of daily events at scale, and how it all presents meaningful data to customers. Ready to take a look under the hood of Taboola? Let's go. ## Quick Definitions Before we go any further, here are a few definitions that'll help you understand the data path in Taboola. - Action: The desired action a user takes on an advertiser's website, such as a product purchase, page view, or email sign-up. Each action is reported to the Taboola server by a corresponding event using the Taboola Pixel. - Taboola Pixel: A Taboola script embedded in an advertiser's website that gathers specific data about user behavior on the site. The script sends events to the server for consumption and analysis. You can learn more about the Taboola Pixel here. - Rules: A set of conditions defined for a conversion for a specific advertiser. E.g., if a page contains the phrase “Thank you" or “Sign-up successful." - Action Conversion: When a user takes a desired action on a landing page they reach through a Taboola recommendation. The two main types of conversions defined by rules in Taboola are event-based conversions and URL-based conversions. Event-based conversion: A conversion based on an event sent from the advertiser's website through the Taboola Pixel. E.g, Add a product to the shopping cart, install an application, or sign up for the newsletter. - URL-based conversion: A conversion based on rules created for a specific URL. E.g., event signups from a specific landing page. Now, back to the conversion data path in Taboola. ## The Data Path in Taboola The data path architecture in Taboola is divided into two main parts: - In each data center, where the data is consumed and processed internally. - In a central processing data center, where the data is mirrored from all data centers and written to a Vertica database for further analysis. ### Part 1: Data inside the Data Center Data Path Inside Each Data Center #### Step 1: Creating an Event When the Taboola Pixel is triggered, it sends an event to our back-end servers. The event is processed and transformed into a ProtoEvent using Google's Protobuf format for easy and lightweight serializing and deserializing. The ProtoEvent is sent to the Action Event Topic and is consumed by an Apache Kafka consumer, where it's prepared for rules matching. #### Step 2: Rules Matching Each event contains contextual account information, so this step checks for a relationship between the business rules that the advertiser defined in the Taboola Ads platform and the current event. It looks to match the event data with the advertiser's defined rules, which are fetched from an updatable cache reading from a corresponding MySQL database for reduced time and increased performance. An event can match a single rule, multiple rules, or no rules. They will still be sent to Kafka. Once passed the rules matching phase, the event, and any matching rules, are sent to the Fired Events & Matched Rules Topics in Kafka to be consumed in a later stage. Matched events are asynchronously queued to move to the next stage of the process. #### Step 3: User Attribution Once matched, the event must be analyzed to determine if it's attributed to a user that reached the current website through one of Taboola's publisher partners. Our platform tries to match the user's event to their clicks and relevant visible events on publishers' websites. It's checked to see if it's from a direct ad click on a publisher's page or if the user was presented with an ad on the publisher's site and they reached the advertiser's site some other way. As we deal with a very large scale of events, in order to have high performance and availability at scale we store the data above in an Apache Cassandra cluster. Once the event is attributed, it's reported, counted, and stored as an action conversion. #### Step 4: Topic Storage in Kafka The final step for the data in the single data center is transformation and storage in Kafka. The conversion is transformed into a ProtoActionConversion using Protobuf protocol and sent to the relevant topic in Kafka for the aggregation part. Once the conversion data passes through the individual data center, it's ready for aggregation. ### Part Two: Data is Sent to the Aggregation Data Center Aggregation Data Center Data Path Conversion data is mirrored from the individual data centers into the aggregation data center before being analyzed and consumed by the Taboola platform. - The final Kafka consumer reads and writes the data to the aggregation Hadoop Data File System (HDFS.) - Offline Apache Spark jobs read the data on HDFS for further filtering and aggregation into multiple tables in Vertica. - The data is sent to Taboola, where it's used in various visualizations and reporting for advertisers, such as the Campaigns Report with Conversions Statistics. And there you have it! Now you know how conversion data moves through our servers, data centers, and systems to provide Taboola customers relevant conversion information. Do you have any questions about how we handle this? Let us know, and we'll try to answer them. --- ### Sneaky Peak to The Secrets of Kafka Assignment Strategy URL: https://www.taboola.com/engineering/the-mystery-of-kafkas-assignment-strategy/ Last Modified: 2025-01-14 14:37:20 Something strange happened while I worked with Kafka. While adding a new consumer from Kafka to one of our services, the service stopped consuming from ALL other existing consumers. As part of my job at Taboola as a team leader on a production team in the Infrastructure group, we're supposed to remove bottlenecks, not create them. This post will describe how I investigated the issue, explain what I discovered, and share my insights into the whole situation. ## Some background Before I get into the rest of the story, here's some background on how we use Kafka at Taboola’s events handling pipeline and why it's critical to our infrastructure. Taboola's recommendations appear on tens of thousands of web pages and mobile apps every second. As users engage with the content, multiple events are fired to signal that recommendations are rendered, opened, clicked, and so on. Each event triggers one or more Kafka messages, which translates into a lot of Kafka messages for every recommendation. ### Event handling pipe Our Kafka clusters reside in eight data centers around the globe and handle more than 140 billion messages a day. That's more than 100TB of raw data daily, and approximately one-quarter of those messages are related to the events. Taboola serves over half a million events per second. The servers that handle those events need to be as fast as possible (p999 < 1ms fast,) so they don't ruin the user's experience. To achieve such a quick response time, we split the work between two types of services, HTTPS request handling services and data enrichment services. Kafka helps pass messages between them. ### Event handling servers Event handling pipe HTTPS request handling services: The events web servers handle the HTTP requests and run a light and quick process for responding to HTTP requests. They convert requests into a proto and send it to Kafka. Some events need to be processed faster or backed up for a longer time. We have several topics based on event types to keep track of this. Kafka can then tailor the retention and prioritization of processing events based on multiple factors, including event type, topic, and more. Data enrichment services: The data enrichment servers consume the messages from Kafka and send them to another Kafka topic, which is the next part of our data pipe. (The Backend Data Pipeline in the previous diagram). Each consumer reads messages from a specific topic, and there are multiple consumers on each server. Since the data enrichment can be CPU- and IO- intensive, we considered splitting each topic's consumers onto dedicated servers. This would isolate problems like CPU pressure from when handling one topic affects the consumption of another. But, because we run these on physical machines, it's hard to maintain such topology. It requires a larger number of physical servers from different types, and it is hard to scale up and down those servers. ### Message processing affects servers differently In our scenario, the partition assignment between data enrichment servers must be balanced. Processing messages for each topic requires different resources because each topic represents a different event. And some event processing requires more CPU power, others require data to be read from a database, etc. It can create partition bottlenecks if too many similar events are consumed from a specific server and can delay processing for all the events. Additionally, we want to have one consumer per topic partition so the message processing will be as parallelised as possible. We have more servers than we actually need, so we still have a consumer per topic if one or two servers go down. Therefore, when all servers are running, there are few idle consumers. The topology of the consumers' servers Now back to the story of what happened when I added a new Kafka consumer to one of our services. ## The strange thing that happened when we added the new consumer We add new types of events to our infrastructure all the time. We usually just add the new topic and relevant consumers, test everything locally with different CI procedures, and then on production servers running existing consumers. This time, something strange happened when we added a consumer for a new event type. When we added the new consumers on one server in the server pool, other consumers on that server stopped consuming from all other topics. The new consumer had a different group ID and consumed from a new topic, so this shouldn't have happened. We were surprised that it affected other groups and other topics. Here's a visual of the consumer assignment after adding the new topic (Topic 3, or T3) to Server3. You'll see that Server 3 should have active consumers for Topics 1, 2, and 3 (the new one,) yet when we added T3, Server 3 stopped consuming T1 and T2, making them inactive. Consumer assignment diagram after adding the new topic What was going on? We started investigating immediately to figure it out. ## We needed help answering why At first, nothing we looked at could explain why this was happening. We searched online to see if other data teams and developers had suffered this but came up empty. So, we looked at Kafka's code as we thought it might've been an issue with how the group coordinator works in Kafka's clients code. We needed a little help from the Kafka instruction manual (Kafka: The Definitive Guide) to help us understand what was going on: When a consumer wants to join a group, it sends a JoinGroup request to the group coordinator. The first consumer to join the group becomes the group leader. The leader receives a list of all consumers in the group from the group coordinator (this will include all consumers that sent a heartbeat recently and which are therefore considered alive) and is responsible for assigning a subset of partitions to each consumer. It uses an implementation of PartitionAssignor to decide which partitions should be handled by which consumer. - Kafka: The Definitive Guide ### Log files to the rescue By looking at the code and enabling all logs on the ConsumerCoordinator event, we finally found the source of the problem. ConsumerCoordinator's logs on debug level “Aha!" I cried. Looking at the consumer's naming, I noticed that the names have some kind of incremental part which is not related to the groupId nor the topic. It all made sense now. ## The solution was in the partition assignment Looking at the logic of RoundRobinAssignor#assign we discovered that before assigning consumers for each topic in the partition, it sorts the consumers by their member ID. The same logic happens also on RangeAssignor#assign, so the issue I described can happen also if you are using RangeAssignor as your assigning policy. Now, a member ID is composed of two parts: - The consumer ID, which is incremented as the consumer is created, and - a UUID, the universally unique identifier, which is assigned by the group coordinator when a new member joins a group, as per the Apache Kafka Rebalance Protocol. By default, consumer IDs are numbered sequentially regardless of their consumer group. You can see this in ConsumerConfig#maybeOverrideClientId. (Note: consumer IDs can be overridden by a specific value in a configuration file, but we'll get into that further down.) ### Why the consumer ID mattered here When we added the new consumers to the server, they were added to a new consumer group and given low sequential ID numbers. We expected them to be placed at the end of the ID list since they were new IDs. Instead, they were added to the front of the list, which then moved all existing IDs on that server forward. Since the assignor sorts the consumer IDs for each group and we have more consumers than partitions, it assigns all partitions to consumers in servers running without the new consumer groups since their sequential IDs are lower. As a result, the affected servers are assigned only to new consumer groups. Partitions were assigned to the new consumers first, so the old consumers no longer had partitions to consume from the old topics. As a reminder, here are the assignments before and after adding the new consumer again. Before - Server 1: Consumer 0 = {T1, Partition 0} - Consumer 2 = {T2, Partition 0} - Server 2: Consumer 0 = {T1, Partition 1} - Consumer 2 = {T2, Partition 1} - - Server 3: Consumer 0 = {T1, Partition 2} - Consumer 2 = {T2, Partition 2} After - Server 1: Consumer 0 = {Topic 1, Partition 0} - Consumer 1 = {Topic 1, Partition 2} - Consumer 2 = {Topic 2, Partition 0} - Consumer 3 = {Topic 2, Partition 2} - Server 2: Consumer 0 = {Topic 1, Partition 1} - Consumer 2 = {Topic 2, Partition 1} - Server 3: Consumer 0 = {Topic 3, Partition 0; and Topic 3, Partition 1} - Consumer 1 = {Topic 3, Partition 2} Remember, Server 3 stopped consuming from Topic 1 and Topic 0 and that incremental consumer IDs are per server. All consumers on the same server affect the lexicographical order regardless of their group ID or topic. And that was ultimately the cause of the issue. The order in which consumers are created affects the assignment of partitions to consumers. Now that we knew what to fix, it was time to edit the new consumer to avoid affecting the IDs this way. ## Implementing the fix To fix it, we needed to add the new consumers for Topic 3 on Server 3 at the end of the consumer list. This would leave the consumption of T1 and 2 balanced as before, but Server 3 can now additionally consume the new T3. Consumer assignment diagram with the fixed T3 consumption on Server 3 We investigated several ways to implement this fix, but we knew we'd have to create short- and long- term fixes. Given how we could introduce this issue into our infrastructure at large, we wanted to be sure to have a good fix available. ### Our short-term fixes We had two options here: - Stopping and starting all consumers simultaneously. - Decreasing the number of idle consumers on each server. Stopping and starting all consumers simultaneously resulted in adding the new consumers to all the servers at the same time. That rebalanced the old topics across all servers and fixed the issue. However, it's a risky move since there may always be problems if any server doesn't restart as expected. As we investigated the issue, we discovered that even though we increased the server numbers over time due to increased traffic, the configuration of the number of consumers per server hadn't changed. That meant there were a lot of idle consumers, no matter how many servers there were. We could mitigate the balancing problem by playing around with the number of idle consumers on each server. But we still needed that long-term fix to truly solve the issue and ensure we didn't have it again in the future. ### Our long-term fixes We came up with four different long-term solutions that would work, depending on the situation. 1. Force the order of bean creation We discovered that we could achieve deterministic assignment of consumer IDs by controlling the order in which Kafka consumers are created for each service. At Taboola, we use Spring to create the consumers, so we can use the @order code to ensure the order of consumer ID creation. 2. Use client.id consumer configuration to control the order of consumer IDs We can control the lexicographic order of the consumers by adding the consumer configuration client.id to Kafka consumers. It replaces the incremental consumer ID and assigns an incremental predefined identifier to all consumers on a server. We can enforce this usage by commenting the code to ensure that all developers specify an ID when adding a new consumer. 3. Implement Kafka PartitionAssignor We can override the ID assignment method to create the sorting we need by using a custom PartitionAssignor in Kafka. By implementing the assign() method, we can control the exact way we want to assign topic partitions between the consumers' servers. 4. Modify the ConsumerConfig code We can modify the client ID creation code inside ConsumerConfig#maybeOverrideClientId, so that the consumer client ID sequence is per consumer group instead of using a global sequence. This solves the issue because it creates a random balance of consumers through the UUID component of the consumer ID. ## Key takeaways While consumer issues aren't ideal, this one helped my team be better developers, and truly understand Kafka's assignment to the further extent. We learned that: - There's a relationship between consumers on the same service and the assignment of topics on a partition, regardless of the group or topic ID. - Each server increments consumer IDs unless the order is explicitly overridden. - All consumers on a service affect the lexicographical order of consumers on the same service . Knowing all this will make it easier and safer for us to add new consumers to servers and topics, and deliver more information more efficiently to our Taboola customers. --- ### Our Deployment Process: How Taboola Manages Servers for Your Recommendations URL: https://www.taboola.com/engineering/our-deployment-process-pillars/ Last Modified: 2025-01-14 14:37:20 There are more than 350 developers at Taboola, and each day dozens of changes are deployed to thousands of servers in seven different data centers around the world. Yet, our systems serve nearly 500,000,000 unique users every day and rarely experience downtime or a significant bug — all without any backend QA teams. How do we do that? Keep reading to find out the secrets to how Taboola deploys and manages the thousands of servers that bring you recommendations every day. ## Our deployment process pillars At Taboola, we have two main process pillars we use in deployment to be efficient and ensure everything we release is high-quality. The pillars are divided into two main phases: pre-deployment and post-deployment. ### What we do pre-deployment We test every project, change, and code update pretty heavily at Taboola. We use a combination of automated, canary, AB, and side-by-side testing to ensure we only promote the highest quality code forward. #### Automated testing We use over 80,000 unit and end-to-end tests before each deployment. That's 73,000 unit tests executed automatically before each deployment and more than 6,000 end-to-end tests developers execute before each merge to master. Almost every change we make, or every line of new code we add, must be unit tested in Taboola. Our CI/CD platform runs most of these tests automatically when changes are pushed to any feature branch, but it's not the only testing we do. Automated tests are a good first line of defense, but additional safeguards are needed. #### Canary testing Wouldn't it be great to test a new code version on a small fraction of users or traffic, so you don't disrupt anything else in production? And if this small test only took a few hours to run, making it easy to do? That's what canary testing is all about (named after the mining tradition of using canary birds to detect carbon monoxide in underground mine shafts.) Taboola's infrastructure is ideal for canary testing since it allows us to: - Deploy new code versions on a few servers that get only a small fraction of traffic (usually 1% of all traffic.) - Rollback the new code easily if there are any issues, or once the time is done. - Generate automatic comparison reports that identify latency, resource and bandwidth usage, errors, etc. We enable canary testing at Taboola with a single click, making it an easy, efficient, and effective testing method for our developers. #### A/B testing To discover insights beyond the performance-related metrics revealed by canary testing and automated testing, Taboola uses A/B tests. After all, new code versions may perform well technically but don't meet business requirements or objectives. A/B testing uses higher traffic volumes and longer time frames to obtain the data needed to be statistically representative of business-level performance. For example, an AB test may use 10% of overall traffic and run for a full week or two instead of 1% of traffic and a few hours like canary testing. It allows Taboola to measure both system-level performance and business-level performance at the same time. And that gives us a better understanding of how the change would affect business metrics like revenues, clicks, and so on. Even though canary and A/B testing are similar, we don't always do A/B testing because it takes longer (a few weeks versus a few hours.) We tend to do more canary tests than A/B tests, but it's easy for us to do both when needed. #### Side-by-side testing Sometimes there are instances where automated, canary, and A/B testing will miss a coding error. For example, you might refactor a method to change its behavior but not its implementation and unintentionally introduce different outputs than was previously expected. The previous testing types I mentioned may not catch the bug unless the new method throws more exceptions than usual. A/B testing might catch it if the bug impacts a business metric like revenue, but running one for a small refactoring change isn't appropriate. To catch these types of refactoring errors, we use side-by-side testing. It's a type of testing with a short feedback loop and a high confidence level that can be done quickly. It works like this: - Duplicate the piece of code that needs to be refactored, whether it's a method or an entire class. - In the places where you invoked the old method, invoke both methods, the old and the new. - Compare the results of the two methods, which should be the same since the behavior shouldn't have been changed. - Log every discrepancy between the two results. - Deploy this feature branch to a few servers for a few hours, so enough traffic passes through it, and it can be properly analyzed. - Look for discrepancies in the log files. If there are none, you're good to go. If there are, fix them and repeat steps 1-6 until you find no discrepancies. Side-by-side testing is a powerful tool but only works if: - You're refactoring a method AND - The refactored method is a query and not a command. It won't work if you're changing or adding behaviors or refactoring commands. ### What we do after deploying A Taboola developer's work is not done after pre-deployment testing and rolling out the change in production. Post-deployment, we do a few more tests and checks to ensure new changes aren't causing new issues. #### Feature flags For whatever reason, if a new change is causing all sorts of problems in production, developers need a way to disable it quickly or roll back the code to a previous version. Doing a full code rollback is time-consuming and costly since it affects thousands of servers and all traffic. In a large-scale operation as we have at Taboola, every second counts, so a full rollback isn't always the best option. We use it only in very serious cases and really, only as a last resort. The other option is to use feature flags to turn off any new changes in production code. We wrap the code changes in feature flags and keep track of which ones apply to the new code. That way, if something goes wrong, we can quickly turn off that feature flag alone without touching the rest of the production code. (And before you ask, feature flags are a cumulative thing in code, so managing them in a large codebase as we have at Taboola is a big task and a topic for another post.) #### Data verification Another task we do after merging our code to master is tracking and verifying that our code didn't cause any regressions in production. We don't verify every change since we do so many every day, but we try to plan for the projects we know should be verified and what tools we'll use to do that. We store most of our raw data in BigQuery, Google's cloud database, so we run queries there to verify the content we serve after deployment. But that's not the only tool we use. We also use: - Grafana to track unlimited metrics (CPU and memory load, total exceptions, number of requests, etc.) on various aspects of our systems to verify how our changes have affected different aspects of our systems. - Tableau to generate reports on our business performance (number of page views, clicks, revenue, user engagement, active users, etc.) - ELK to aggregate all server logs so we can search them to verify behaviors or investigate problems. ## Final thoughts Keeping large systems, as we have in Taboola, alive and healthy while deploying dozens of changes daily is challenging. The size of our infrastructure environment and developer teams means we need to work collaboratively, test frequently, and monitor our changes even after deployment. Taboola developers have a strong sense of ownership of their work, so they don't just merge their code and forget about it. They use and participate in the various testing layers we have before deployment to ensure their code doesn't break anything. They also actively monitor their changes and quickly act when alerted to a problem. Our post-deployment processes ensure developers know how to efficiently fix a problem, so it doesn't significantly impact the rest of the environment or ecosystem. And that's how we keep the Taboola servers up and running efficiently while still deploying all the changes and enhancements our customers love. --- ### The Challenges of 3rd Party Scripting URL: https://www.taboola.com/engineering/the-challenges-of-3rd-party-scripting/ Last Modified: 2025-01-14 14:37:20 Many years have passed since the first time Javascript entered our web pages, turning them from static forms into the dynamic, vivid creatures they are today. Every year, the amount of scripts, styles, and plugins in every popular site keeps increasing. Inevitably, you cannot expect the site maintainers to develop everything they need on their own. Consequently, they come to rely on external content, tools, plugins, and services to handle many of their growing needs, like buttons, comment sections, ads, chatbots, and more. These tools, usually generalized and used by many different consumer sites, are all 3rd party services. At Taboola, such 3rd party services are a critical part of our product. Namely Rbox, our recommendation product, is a 3rd party service embedded in publisher sites. A dedicated team of top skill front-end developers and web experts are responsible for developing and maintaining the Rbox. Supporting the Rbox is a hard, complicated job, but a rewarding one. When we observe the differences between writing a 3rd party service to writing for the host application, we can find many points of interest: Feature/Platform Host 3rd Party CORS conflicts Usually same origin site and content, so no problems. Can face problems trying to get resources from another source. Main window control Can access whatever variables and API's they want, no objections (unless no browser support). May face clashes with global variables, clash with host API usage, sometimes living in an iframe. Web resource usage Can use as much network bandwidth as they like, whatever libraries they want (as long as their customers live with that). Better not bring too much code and use too much network at the expanse of their host. Content positioning Always aware of where their content is relative to the page. May be placed at every point (may change due to agreements between the 2 sides) and must adjust to every circumstance. Security and compatibility Security very important for company and client interests. Host code should strive to be safe from security breaches, malformed code, and compromised dependencies. Security is even more crucial. In 3rd party scripting, there's no such thing as "too much" security, since 3rd parties act on someone else's site. If there is any doubt about using certain libraries or code aspects, the default answer is "don't" until proven otherwise. AMP Can use whatever AMP components they want and add the amp-scripts they want. Need host consent in order to bring the components they need, have limited availability to the page. Performance Even though the host can suffer from performance issues, it does not always address them. This may be because there is not enough impact, more focus on making content, or no fitting tools to find the problem. Must monitor all traffic and identify performance constantly since host sites can see which 3rd party services cause issues and need to be able to shut them down. Most challenging to 3rd party providers is the fact that most of the scripts are generic across all of the hosts, and those scripts might expose different performance impacts on specific types of traffic. As the writers of a 3rd party script, we cannot enjoy the advantages of using whatever we want, knowing everything we need, and having access to everything a page offers. Worse, since we may “live" in many different sites, each with their own styling, privacy conventions, and state of global javascript context, we need to adapt to every possible situation, sometimes even surreal ones. Let's go over some of the more potent and common challenges and see how we at Taboola handle them: ## This seat is already taken, sir Imagine you're working on your code. You're toiling endlessly on polishing every bit, ousting every bug, solving any edge case, brandishing every feature, just to have everything crash and burn the moment your script is uploaded on the client's site. And for what reason? Because you wrote something like this: // myAwesome3rdParty.js And it just so happens that the host site has a script that looks like this: // HostCode.js And guess what happens? An exception is thrown on your side, and things break (or worse, an exception is thrown on their side and host site breaks), and next thing you know, your customer is on the phone and is furious because you stepped on their toes. So, what can we, as good citizens, do to avoid this chaos: - Namespacing: First, it is beneficial to have our variables with their own prefixes to reduce naming clashes. Most coders use descriptive names for their variables, and they usually wouldn't use our unique prefixes in their variables. This way, we greatly reduce the harm potential - Encapsulate into private “modules": namespacing is easy and fast, but not absolute. Being implemented in so many sites, you can never know what someone may do to alter them. Hence, it is imperative to have our code as isolated as possible. Have all our variables hidden inside code blocks, functions, and modules (which in classic vanilla javascript, would look like this): In the example above, any declared variable will only be visible to this secluded scope and will not clash with the host code scope.Unless someone does something careless… ## Who moved my API? Do you think overriding someone's variable is bad? Imagine what happens when you override a webpage API. So, let's imagine you would like to use an API like IntersectionObserver in order to know when certain elements are in the viewport. IntersectionObserver is a native API, so you don't even need to import it from anywhere! And yet, it is not supported on IE, so a bit of caution is necessary. You decide to go with this approach: window.IntersectionObserver = window.IntersectionObserver || myintersctionObserverPolyfill; Seems legit, right? In case the browser supports IntersectionObserver, you will use native implementation. If not, you can have your own polyfill, and everything's fine, correct? But, what if you're not the only one who thought about this? What if the host code does just this: window.IntersectionObserver = window.IntersectionObserver || theirIntersctionObserverPolyfill; This means that either you're going to use their polyfill, or they will use yours in any browser that does not support native. Since there may be various versions to even native APIs, things can get tricky.And this is just one example. Caution is needed when doing something like overriding anything that sits on 'window.' A better method would be: const myintersectionObserver = window.IntersectionObserver || myintersctionObserverPolyfill And use myintersectionObserver in your code instead. One must be careful when changing something that the host site may employ. This is also true to external library clashes, prototype overrides, and more. ## Floating red boxes are a tradition here Ok, so you were careful. You put every variable, every class, every style in isolation. You know your code and the customer's code live in parallel universes. There shouldn't be any problems now, right? They'll do their thing, and you do yours, and you'll all live happily ever after, until... But how??? We have no similar classes. We made sure to get everything specified to the max.Alas, CSS doesn't care. Consider the following code segments: // myAwesome3rdParty.css And their code: // HostStyle.css What a mess! These are some predatory rules, and they might work fine for the host since they built everything around it. But it sure doesn't work right for you. Worse, you may think to get your styles hyper specified, something like: here is the code: This may be useful in some cases, but then they hit you with the worst offender of all: !important Which nothing can beat. ## Is there a solution? One of our basic solutions for publishers is custom CSS hooks, which means that we can add additional style rules in order to reconcile the gaps between us and a specific publisher . We can use higher specificity rules, change the way we uphold our styles, and in some rare cases, even use an !important or two just to set things right. Using these should be very subtle, but it allows a generic service to be more user-tailored. ## Live in the pod (and eat the bugs) There are many types of 3rd party services. Some of them as simple as showing a small ad, a little panel with some content, a button that sends ajax, or even a nice slideshow. But some 3rd party services can get very complicated. Taboola, in particular, offers some features that may count as very “invasive" to the page, such as: - Next up - a floating ad fixed at the bottom of the page - Read more - shortens and hides part of the article - Explore more - takes over browser history to affect “back" functionality - Stories - small, ad circles at the top of the page What if: - We are in an (unfriendly) iframe and can't access the page or page APIs. - AMP - scripting is very limited, and we can't use custom UI outside AMP components. - PWA/SPA - apps built in a structural, restrictive way that may not integrate well with how we do things. Were these brands more mainstream, it would have been severely damaging for us. In these cases, many of the previously mentioned features would break. In fact, even our base functionality can suffer greatly. How do we work against this? In the case of AMP, we have our dedicated AMP component in Google, which loads our widget inside an iframe while retaining most of our abilities. As far as iframes are concerned, we use messaging and events between the hosts and us, enabling us to do several UI actions even without control outside our iframe cage. For modern frameworks, we have already started planning for special wrappers written in respective frameworks (such as React) that will enable us to be more in line with the surrounding implementation. ## When byte gluttony goes bad The web is a fast, dynamic place, and whoever gets too slow goes broke. Research shows that even mere milliseconds more in page load can drastically reduce the number of users a site receives. For such reasons, site maintainers do their best to reduce their loading and rendering times and ensure that UX stays as fluid as possible. It is also one of the cornerstones for initiatives like AMP. But what happens when said site owners use 3rd party providers, and those 3rd party providers are bogging down the site with many kilobytes (or even megabytes) of their own content? This content can be slow loading, harm UX, and even reduce SEO for the site. There are several ways in which we make our product more friendly for the host site: - Lazy loading: our core scripts come in large chunks, which weigh on the host site. However, any module that is not required immediately for the operation of service will be lazy loaded instead to balance the load on bandwidth and page load. In the future, we plan to make it such that any piece of code that is not required for a publisher will not be loaded. - Monitoring: as part of our performance improvements, we are tracking core web vitals, CLS (Cumulative Layout Shift), FID (First Input Delay), and LCP (Largest Contentful Paint). We also work on features that improve performance on these metrics. An entire stream of developers is dedicated to solving problems in these domains and providing solutions to publisher complaints. ## IE for life Javascript has gone a long way since it first appeared. After the big jumps of ES2015,16,17, we have so many great language features, so much syntactic sugar and shortcuts to make our life easier and the code more readable. CSS has gone a long way too, and every new invention helps us build a more robust experience. Many new apps and sites may choose to neglect old browsers and direct their attention to the vast majority of users on modern browsers - they live in the Chrome world, and for them, life feels like an idyllic dream. However, we at Taboola still work with many publishers worldwide who serve users with a wide variety of systems and user agents. We cannot neglect all legacy browsers and devices and must act accordingly while developing and maintaining our code. ## It's always your fault! After all is said and done, and you have made sure your work is covered, tools and good processes are in place, and everything is working just as you intended. You sit back, relax, and see if your customers are happy with your product. Then, your customers call to complain that things are not working. You're baffled, and start double-checking everything you did, considering the most absurd possibilities and any mistake on your side. Just a few hours later, you find out it was an integration done wrong. Mainly because your client got the wrong notion of what exactly they can do with the service you provided them. Does it sound frustrating? Yes. Does it seem outrageous that they still blame you, and you are the one who needs to help them out? Yes. But that's how this works, and if the customer is not pleased, the customer will not keep using your service. So indeed, it is always your fault, and even though no product is fool-proof, you need to make sure you do everything in your power to avoid them making mistakes. At Taboola, employees are always on call in case of an emergency. Our experience finding and solving production issues is very high, sometimes mere minutes after a customer complains (or even before). Our product is not bulletproof, but we find and fix it very quickly if there is a problem. ## After party There are many challenges in 3rd party scripting, but the possibilities and benefits are vast. Being able to easily integrate into many products, make and gain value wherever your code lies, and spread that code around the web like wildfire is a huge opportunity. It is no wonder that in order to achieve this, one must be careful and considerate. Of course, being an expert in web development doesn't hurt! There are more points to consider, but I hope the above gave you enough material to understand the basics of this wild, wild web. --- ### Samplex: Scale Up Your Spark Jobs URL: https://www.taboola.com/engineering/samplex-scale-up-your-spark-jobs/ Last Modified: 2025-01-14 14:37:21 # Using Samplex to Scale Up your Spark Jobs When I encountered Taboola's data a few years back it struck me that everything I've heard and learned about Big Data wasn't really Big Data. ## Understanding Scale Taboola's big data environment is on a massive production scale, with petabytes of data flowing through data pipes. And with thousands of different workloads running constantly over this data, Taboola generates enormous amounts of raw data each day, ~100TB, not including all further aggregations created for the sake of reports and analysis. Such data is constantly used in various applications throughout the company, in applications such as Deep Learning, business intelligence (BI) analysis, and reports. When working with big raw data, we often seek to optimize downstream jobs and prevent unnecessary crunching of the entire dataset. This is achieved by creating subsets of the raw data typically needed for different jobs or common queries. A classic example is applying the same filter over data for different downstream pipelines. ## Multiple Reads Wastes Resources Over time, we at the Data Platform Group identified a growing number of Spark jobs that produced different subsets of our full dataset for later use in downstream queries. Each of these queries required substantial resources to query the full dataset, filter it, and save it back to HDFS. It seemed like a waste of resources... if only we could read the full dataset once and produce all these subsets in a single Spark job. The naive and straightforward approach is to produce different subsets of the data by reading the source data frame and then save the data multiple times with multiple predefined filters, as in the following example: Spark is lazy, each write action would trigger the entire DAG (a workflow on Spark's scheduling layer) for that action, so it will require Spark to scan the whole data several times with a corresponding number of write actions. We could rather try caching the input data. Unfortunately, we are dealing with huge amounts of data that is too big to fit in memory, even with dozens of executor nodes. In addition, each subset we produce might benefit from predicate pushdowns and other storage optimizations to avoid reading the entire data. But with many different filters, and over very complex schema, we can't really store the full data optimized for every read path, without keeping different projections of the full dataset. ## Read Once Write Many We decided to accept the challenge and find a solution that would involve a single Spark action, with no caching (and without reading the data multiple times). Our solution, which is described hereinafter, saved us an abundance of resources, and we even open-sourced it. We realized that we needed to return to basics and take ourselves out of our Spark SQL comfort zone. We used Spark core functionality to go over each Parquet row, apply a series of different predicates to decide which subset(s) of data it should be written to, and write to multiple output locations the rows that met the predicates' conditions. ## Samplex Usage The Samplex application programming interface (API) implements two interfaces for every output dataset. The first is `SamplexJob`, which defines the destination path for the output written to, and a `SamplexFilter` that defines the predicate used. SamplexJob - in which you should provide destination path, Record, and Schema filters. SamplexFilter - which will tell to the samplex which record should write to output SchemaFilter (optional) - in which the user should define the fields to keep or remove based of Block/Allow lists. When you have your SamplexJobs ready just create a SamplexExecuter and provide your jobs. Please look at the full examples at GitHub. While testing Samplex, we encountered a bug (as described in PARQUET-1441), in order to solve it we used the latest version of the parquet-avro library and used maven-shade-plugin to relocate it under Samplex open-source. ## Things to Note Using Samplex, there are few considerations that one should be aware of: - Parallelism - With Samplex, each parquet part is fully processed within a certain task, and data cannot be split and repartitioned to achieve higher parallelism. Basically, Samplex is capped in parallelism by the number of input files. This means that if a parquet part contains multiple row groups, they cannot be processed in different spark partitions as they might be with SparkSQL. - Number of output files - The output of Samplex is also determined by the parallelism used. If a certain output subset is very small, it cannot be coalesced to fewer parquet parts, i.e., all outputs are written with the same parallelism. - Predicate Push Down - Samplex cannot benefit from any predicate pushdown mechanism, as this is counter to the idea of Samplex where each output uses a different filter. Applying predicate pushdown would require intersecting the predicates for all subsets of data. - Tuning - Additional configuration that may help to boost performance is to add more cores per task (spark.task.cpus) as writing of outputs performed in parallel in the same task. Some of the concerns above will be resolved as Samplex development continues. Note that all developers can contribute to future Samplex versions, whether it's by identifying and resolving bugs, suggesting new features or improvements or simply by brainstorming. ## Results Our benchmark of Samplex vs naive usages of Spark SQL to produce sampled data, included 4 sampling filters over 1.1TB of parquet data. In both cases, we allocated 800 CPUs for Spark. Moreover, to avoid any bias of data locality, Spark executors were not running over HDFS cluster machines. During our proof of concept (POC), we found a significant improvement in execution times with the same resources as used before, with over 50% reduction in overall duration and even greater reduction in total bytes read. The overhead of adding another filter for Samplex is much smaller when compared to adding another independent Spark job. Furthermore, by using threads for writing the output we were able to exploit our resources more effectively. Our benchmark was conducted as follows: - Producing four different subsets - Number of Input Files: 4500 - Total Input Size 1192.0GB - Number of Records: 337,693,147 - Number of Columns: more than 1600 (with deeply nested schema) - File Format: PARQUET - File Compression: SNAPPY ## Conclusion We love Spark, it is one of our core technologies at the Data Platform Group and a key technology in our stack since it’s early releases. It has never failed us. But with our scale, we often need to think outside the box, and find creative solutions to leverage the technology at hand and do things more efficiently. With over 700 compute nodes (~30000 cores) in our Spark clusters, we strive to maximize the efficiency of available resources. As new initiatives in R&D require more computing power, we always prefer to scale up, rather than blindly scale out before pursuing other alternatives. --- ### Building a Puppy POC: How Product Methodologies Guide My Work and Home Decisions URL: https://www.taboola.com/engineering/building-a-poc-for-a-puppy/ Last Modified: 2025-01-14 14:37:21 Many R&D buzzwords and acronyms can seem like complex jargon — unnecessary shortcuts for concepts that are already pretty basic. But really, the good ones are put in place to help processes flow more smoothly. And they're so simple, even a child can understand them. I learned that first-hand during the pandemic. As the VP of Tech Operations for Taboola, I specialize in accelerating teams to develop products. And I've always believed that an agile and iterative approach leads to a successful and innovative company. Now, my ten-year-old daughter understands that, too. Here's how it happened. ## The Pitch In March 2020, my family was doing what many others around the world were doing: staying safe and quarantining in our home. As many others were experiencing, nothing seemed to be going well. Then, to make matters more stressful, my daughter started asking (a.k.a. begging) for a dog. Despite my off-hand joke about her little brother already being a sort of pet, she made a face and went to ask my husband instead. His response? “Yeah, maybe." “Oh no," I thought. “Big mistake!" ## Assessing the Product-Market Fit Then my business side kicked in. While handling my toddler son who was trying to “help" with the dishes by dangerously perching himself on the counter, I turned to my daughter and said, “Sure, you can have a dog. But first, let's do a POC." I didn't even think before I said it. I just blurted it out as if I was in a meeting with a colleague. Here's how that conversation went: Kid: A PO what? Me: A POC, dear. It means "proof of concept." If you really want a dog, you need to prove that it is feasible. Kid: Ok, let's get a dog and I will prove it to you. Me: No, no, you're missing the point. I won't get you a dog until you prove you're capable of having one. Why waste time and energy on finding a dog if you might not be up to it? Finally, we agreed on the following terms for our POC: My daughter needs to take an imaginary dog with an imaginary leash for a walk twice a day around 6:00 AM and 6:00 PM. If she can do it, the MVP, or minimal valuable product, would be to have a four-legged animal at home for a few days. (I figured the neighbors would be happy to let us sit with their cat for a short period of time.) My daughter said she understood, but she still thought it would be a waste of time because, in a few days, we'd be back where we started — with her asking for a dog. “Ok," I said. “Let's see in a few days." ## The Results Two days later… Kid: Mom, will you go for a walk with "the leash"? It's cold outside. Me: No way! Are you saying the POC failed? Kid: *Grunts and slams the door on her way out for the walk.* The next day, at a family barbecue… Me: It's 6:00 PM. Enjoy your walk, sweetie. Beware of cats on the way. Kid: Oh, no. I forgot to tell you. The POC failed. We won't be doing the MVP after all. Pass the kebabs, please. Me: *Smiling, knowing what victory feels like.* Thirty seconds later… Kid: By the way, mom, what's the POC for hamsters? Me: *Not smiling anymore.* ## The Takeaway Looking back, I think this experience had three positive outcomes: - Sharing more of my work life with my daughter - Helping my daughter learn about the responsibilities of owning a dog - so that I won't find myself taking care of her dog - Experiencing how useful and valuable it is to use concepts like POC and MVP in real life In fact, I already have my next lesson planned! If we must go into lockdown again as this pandemic continues to unfold, I think I'll use the opportunity to explain how my kid's Nintendo Switch affected the EBITDA and cash flow of the household. You know, good old family fun! --- ### From Home Support: Our IT Transition to Work-From-Anywhere URL: https://www.taboola.com/engineering/work-from-anywhere-support/ Last Modified: 2025-01-14 14:37:21 During the pandemic, most companies quickly adapted and moved to a work-from-home model, as a sudden necessity of the lockdown restrictions introduced by efforts to combat the spread of COVID-19. For most workers, this was a new way of thinking. But for others, more specifically, in IT and Support departments, this was more than an expansion of their existing arrangements. From an IT and Support perspective, what makes working from home more challenging than working from the office? ## Hey there! Can you help me with this issue? In some countries, employees are returning to their offices, even if only part-time. But office life, as we knew it, has changed dramatically. For a corporate IT team, the list of work-from-home challenges is long. It includes the shutdown of IT logistics, the primary support location, internal client connectivity, protection, and more. For many companies, the office serves as a mini-hub for IT logistics. If something doesn't work with your computer set-up, you can call up an internal IT expert, and they will help you out, whether it's a software upgrade, new cable, or broken hardware. The center of IT services isn't just around the corner from your desk anymore, waiting for you to ask for help. We now have to bring logistics to the employees. Delivery services provide an alternative for the need to visit the office, although a cost is attached. Yet, the most significant change is the barrier to receiving personal support. Many larger support organizations have remote desktop support or call centers. For many smaller IT shops, this isn't the case. You can't just drop by when you have an issue. You can't just ask someone at an adjacent desk to open a ticket for you. ## Does your home need an IT upgrade? Your home IT infrastructure might not meet the minimum requirements needed to operate effectively and securely. Anything from home WiFi coverage, ISP speeds, ISP performance issues, and VPN connectivity problems (for the few resources that still require VPN access) can be deficient. There is no corporate-level support for home networks, and now there are more users on your home network, such as children in remote schools or spouses and partners also working from home, both eating up bandwidth. From the perspective of the IT department, the single, reliable corporate network suddenly became hundreds of home networks. Multiple home networks are now part of the company IT realm from a single network under close IT control and support. From the perspective of the IT department, the single, reliable corporate network suddenly became hundreds of home networks. From a single network under close IT control and support, multiple home networks are now part of the company IT realm. Every home network issue becomes a productivity issue. Multiple users with the same problem might become a shared ISP problem or an extremely chatty application consuming too much bandwidth or causing a DNS issue. The list goes on. Because of this shift to home offices, there are many new problems to consider and address. ## We found our “why" The ability to help direct our employees and improve their work experience is part of what we do in IT. Our goal isn't to manage corporate networks, but to provide business enablement, empowering our employees to do their work effectively. This is why we come to work, and this is what the business needs from us. Enabling the team to focus on improving the day-to-day work and productivity of the organization was a project everyone enjoyed working on. We decided to build our own "Home-IT" proactive monitoring system. This needed to be a light and agile solution that would provide our IT team and employees with clear visibility of their working environment. ## Improving the user experience with data How can we improve our users' experience in this new, sometimes chaotic environment? Of course, being IT geeks, our IT and Support Taboolars first looked for data, statistics, and observability to help solve our problems. Being data-driven is what allows us to learn and improve daily at Taboola. We looked for a way to bring in data from every laptop and device connected to our applications and systems. From that data, we would try to find patterns and help support users in a meaningful way by proactively identifying and resolving individual user issues alongside regional or home network-related problems. When reviewing months of "work from home" support tickets, we discovered which statistics would be the most useful: - OS statistics (CPU, memory, disk capacity, running apps, versions, uptime) - Network data (connection type, signal strength, DNS, and ISP) - Network statistics (packet loss and RTT to multiple targets - over VPN, fixed points, and 2nd hop) This data could help us provide users with quick answers to many of the questions regarding the state of their connection and the problems they may be facing — anything from a weak WiFi signal to a saturated home Internet uplink. ## The project The solution we set out to build had to solve the issues in our new working environment. Also, we wanted to use what we already had to avoid deploying new tools. To that end, we defined the following: - The solution should support both Windows and Mac workstations. - No new agents should be added to the OS. - No installations of any tools into the working environment. - VPN should not be a factor when collecting the data. - There should be a minimal impact to users' devices. To our benefit, we did have the advantage of a managed OS environment. All our workstations are centrally managed and have some agents on them already, so anything we did had the ability to use our central management system. ## Solving the data collection challenge Following a brainstorming session, we found that creating a local script using only the existing OS tools could actually work. The data was readily available with command-line tools, which was the "easy part." Checking for connectivity, CPU levels, OS version, or IP information would be no problem! Checking the packet loss statistics for the second router up the chain, again, no problem; there is also a tool for that. The main issue was the data collection. How do we bring in all the data in a secure fashion yet avoid the use of VPN? Our choice was to upload the files to a cloud repository. We decided to use the cloud as a go-between and provide the end-user with a destination unique to that user's machine for convenience. Creating a write-only key that doesn't have read permissions was part of the security design. If the user can only write and never read, no data will be exposed even if the key is lost before the next rotation. On the local drive, we keep only a limited size log and manage the log rotation (as we didn't use any log tool, but a script, that logic needed to be taken into account as well). ## How did this help us? (and how can it help you, too!) We did this project with the belief that it would make a difference. Little did we know how much of a difference that would be. Now, when a user calls in for support, we already have lots of baseline and historical data about the user's environment. This data is well beyond the local OS and crosses into the working environment of the user. Packet loss on the local network? Weak WiFi signal? VPN connected with no DNS update? All of this is, and more, now available in data that we can easily access. Support calls have reached an exponentially faster resolution due to the data being already available for the support engineer to read. For the more technical users that want to check on their own, there is now a local log of multiple potential issues they can see. The impact on the user's support experience is profound. There is no longer a need to open a ticket and provide a call back with "check this/do that." The calls now start from an insightful point, and we already see the time to resolution of support cases shorten. We are also seeing the comprehension of the support reaching deeper into the user's work environment. ## A note on privacy We take user privacy very seriously and collect only operational data that can help solve problems. The data collected is limited to system monitoring data, the same data normally collected from servers (on-prem or in the cloud). The data collected from the laptops is hardware and operational related, with no user actions being collected in any way. ## Conclusion Although we were suddenly plunged into a challenging situation with the onset of the COVID-19 pandemic, the Taboola IT team was able to rise to that challenge. Supporting both staff in the office and at home during different phases of the virus outbreak meant that we needed to quickly develop tools to help us diagnose problems across our new universe of multiple home-based networks and hardware. By working together, leveraging the power of the cloud, and using existing tools, we were able to smooth the path for better IT solutions for those at home and continue to support the business and our colleagues effectively. --- ### We Have Only Just Begun: A Message from Our SVP of Research & Development URL: https://www.taboola.com/engineering/we-have-only-just-begun/ Last Modified: 2025-01-14 14:37:21 *Excerpt from email originally sent by Aviv Sinai to Taboola's R&D team There have been so many new, exciting changes happening at Taboola. We became a public company, a huge milestone for Taboolars everywhere. We also announced our plans to acquire Connexity, bringing eCommerce recommendations to the open web. And this is just the beginning. I've been getting messages from friends, family, and colleagues wishing me congratulations, echoing that this must be amazing closure for almost 14 years of work. By definition, closure (the end) couldn't be further from what I feel. For me, this is just the start of something grander. In 2004 Google went public with fewer than 2,000 employees. I bet Larry Page and Sergey Brin, CEO and president of Google's parent company, Alphabet, respectively, didn't feel closure when they IPOed, as they were so far from realizing their vision. We all know what Google has accomplished since then! For all the years I've been at Taboola, I can't think of another time when we've had so many opportunities (and challenges) in front of us, while also kicking-off numerous initiatives, projects, products, and technology advancements. We are reinventing how we think about our current products while considering the need for specialization. We are adding new advertising sources, from new verticals to new customer segments. We are working on new user touch points on publisher sites, apps and mobile devices. We are revamping our algorithmic and AI stack, focusing on making both more scalable and safe while changing the basic concepts on how we train, serve and model our algorithms. And that's not all. Taboola has completely new web platforms, including Taboola Ads. Our team will continue to innovate and build new Ad Formats and UI Innovations. We have built our own NLU stack and content lake infrastructure. We've completely changed how we conduct experiments at scale, and how we create infrastructure to build, deploy and test the software and models we build. For example, we're building our next generation front-end generation engine, with new tools and infrastructures that would increase our scale and performance of all systems. We're reimagining what we can do with Taboola Newsroom and insights, revamping our bidder to become a platform used by many products, and rearchitecting our core advertising platform and stack. We're reequipping our already state-of-the-art data platform to support our needs internally, our users' needs, and our new systems. This, and more, will allow us to succeed in our previously mentioned goals and many other new beginnings that we predict are just around the corner. I don't feel closure, because our products, technology, and platforms are just the foundation of what they could be for our users and our partners. Looking at the hard work we've already mastered - now is the time to take our creations to the next level. While being mindful of the amount of (cool) work ahead of us, it is an excellent time to think about what we have accomplished, and to me, the proudest and most unique achievement at Taboola is our team. Together, we've built a strong Research & Development (R&D) team that comprises hundreds of engineers, working from around the world: Israel, US, Taiwan, Ukraine and Greece. Writing code is the easy part. This is the basic undertaking which is expected of us, engineers. However, this is not enough. If one wants to build a strong team that can accomplish anything, the most important thing is one's culture. We've built a team with one of a kind culture: a culture based on collaboration and helping one another, owning our tasks end to end, knowing our business while accomplishing our to-do list with amazing craftsmanship and professionalism. Having Taboola's R&D team gives me the ability to know that we can do anything. I look forward to working on all these new beginnings, and I look forward to seeing how our products continue to improve and grow. But mostly, I look forward to working with my team. Celebrate each and every day, as tomorrow will continue to bring even more exciting beginnings. If you're interested in a career at Taboola, we're hiring internationally! Please visit our website to check out our open positions. Disclaimer – Forward-Looking Statements ###### Taboola (the “Company") may, in this communication, make certain statements that are not historical facts and relate to analysis or other information which are based on forecasts or future or results. Examples of such forward-looking statements include, but are not limited to, statements regarding future prospects, product development and business strategies and our projections for future periods. Words such as “expect," “estimate," “project," “budget," “forecast," “anticipate," “intend," “plan," “may," “will," “could," “should," “believes," “predicts," “potential," “continue," and similar expressions are intended to identify such forward-looking statements but are not the exclusive means for identifying such statements. By their very nature, forward-looking statements involve inherent risks and uncertainties, both general and specific, and there are risks that the predictions, forecasts, projections and other forward-looking statements will not be achieved. You should understand that a number of factors could cause actual results to differ materially from the plans, objectives, expectations, estimates and intentions expressed in such forward-looking statements, including the risks set forth under “Risk Factors" in our Registration Statement on Form F-4 and our other SEC filings. The Company cautions readers not to place undue reliance upon any forward-looking statements, which speak only as of the date made. The Company does not undertake or accept any obligation or undertaking to release publicly any updates or revisions to any forward-looking statements to reflect any change in its expectations or any change in events, conditions or circumstances on which any such statement is based. --- ### ScORe - Schema On Read for Spark SQL URL: https://www.taboola.com/engineering/spark-sql-score/ Last Modified: 2025-01-14 14:37:22 ## The world is not flat, it’s highly nested With over 4 billion page views per day and over 100TB of data collected daily, scale at Taboola is no joke. Our primary data pipe deals with masses of data and endless read paths. Could we optimize our schema for all these read paths? Guess not... Our schema is HUGE and highly nested. After digesting the data, we keep it in hourly Parquet files on HDFS, where each hour consists of about 1-1.5TB of compressed data. Our schema roughly looks like this: root |-- userSession: struct | |-- maskedIp: long | |-- geo: struct | | |-- country: string | | |-- region: string | | |-- city: string | |-- pageViews: array | | |-- element: struct | | | |-- url: string | | | |-- referrer: string | | | |-- widgets: array | | | | |-- element: struct | | | | | |-- name: string | | | | | |-- attributes: map | | | | | | |-- key: string | | | | | | |-- value: string | | | | | |-- recommendations: array | | | | | | |-- element: struct | | | | | | | |-- slot: integer | | | | | | | |-- type: string | | | | | | | |-- campaign: long | | | | | | | |-- clicked: boolean | | | | | | | |-- cpc: decimal(10,4) It contains over a thousand columns, with repeated fields and maps interwound together. Parquet is a columnar format, allowing one to query for a certain nested column without reading the entire structure containing it. So in the example above, one could execute the following query without fetching the entire userSession data to the client: select userSession.geo.city from user_sessions We use Spark widely to produce reports, carry out our billing process, analyze A/B tests results, feed our Deep Learning training processes and much more. But Spark turned out to be crude when pruning nested Parquet schemas. It surprised us to see that running the above query in Spark SQL would result in reading the entire data into Spark executors. Redundant fields are omitted only on client side, leaving us with the required fields for the query (highlighted): root |-- userSession: struct | |-- maskedIp | |-- geo: struct | | |-- country: string | | |-- region: string | | |-- city: string | |-- pageViews: array | | |-- element: struct | | | |-- url: string | | | |-- referrer: string | | | |-- widgets: array | | | | |-- element: struct | | | | | |-- name: string | | | | | |-- attributes: map | | | | | | |-- key: string | | | | | | |-- value: string | | | | | |-- recommendations: array | | | | | | |-- element: struct | | | | | | | |-- slot: integer | | | | | | | |-- type: string | | | | | | | |-- campaign: long | | | | | | | |-- clicked: boolean | | | | | | | |-- cpc: decimal(10,4) In this scenario, given that the other elements in the schema are repeated, we may very well end up throwing away 99% of the data that we fetched from Parquet. Image 1 depicts the total input size and execution time for one of our actual (but not heaviest) queries in production. It reads into Spark about 350GB (out of 1.1TB of hourly data). Total time across all tasks: ~60h, with 36MB input per task on median. Image 1: stats for stage reading input for production query ## Prune it yourself We decided to make it easier on Spark, and “flatten” our schema as much as possible. That is - prune any nesting level for non repeated or map fields. So our above example schema then looked like this: root |-- userSession_maskedIp: long |-- userSession_geo_country: string |-- userSession_geo_region: string |-- userSession_geo_city: string |-- userSession_pageViews: array | |-- element: struct | | |-- url: string | | |-- referrer: string | | |-- widgets: array | | | |-- element: struct | | | | |-- name: string | | | | |-- attributes: map | | | | | |-- key: string | | | | | |-- value: string | | | | |-- recommendations: array | | | | | |-- element: struct | | | | | | |-- slot: integer | | | | | | |-- type: string | | | | | | |-- campaign: long | | | | | | |-- clicked: boolean | | | | | | |-- cpc: decimal(10,4) As you can see, if one wants to query now for the city column, Spark doesn’t need to read the entire data, only the queried column will be read. But what if we wanted to compute the revenue (the sum of cpc for all clicked items)? We still had to read a lot of data, the entire userSession_pageViews instead of the fields participating in the Spark SQL query below. select sum(rec.cpc) from user_sessions lateral view explode (userSession_pageViews) as pv lateral view explode (pv.widgets) as widget lateral view explode (widget.recommendations) as rec where rec.clicked = true We came across the notorious SPARK-4502 bug. Though there was an open pull request for this, it was not progressing as quickly as we needed... We realized that if we provided Spark with the required schema, making it “blind” to the unnecessary payload, we could reduce input size for our queries. We tried it for a couple of queries and it showed promising results. We even used it in production for our heaviest queries, providing the schema manually. But with dozens (and nowadays hundreds) of routine queries, we realized manual pruning just wouldn’t scale. ## Giving it our best effort While the heroic thing to do was to buckle down and help with Spark PR, there were few reasons to take a different course of action: - It’s still out of our hands - months could still go by until the issue is fixed in some future Spark version, along with all of the risks involved in upgrading. - Even if a fix was slated to be released, we didn’t want to constrain ourselves to upgrading. We have a mono repo with multiple services and different use cases, and upgrading is challenging (but that’s for another post). We decided to tackle it from a different angle, one that is based on the fact that Spark is using lazy evaluation, and so we developed ScORe (open source) for Spark SQL queries. Given any Spark SQL query, we can take its logical plan, and gauge which columns from the underlying data source are participating in it. We can then create the required schema, and then recreate the dataframe with the schema that we generated. It is only then that we perform any action. So we basically take the original (full) schema, and try to nail the perfect ScORe (schema on read) out of it. ## Traverse the woods Deriving the required schema for a Spark SQL query starts with passing the LogicalPlan of the query to SchemaOnReadGenerator. The examples in this section will assume the following schema: root |-- nestedStruct: struct | |-- childStruct: struct | | |-- col1: long | | |-- col2: long | |-- str: string |-- someLong: long |-- someStr: string #### Simple Enough Let’s consider the following Spark SQL query: df.select(“someLong”, “someStr”) .filter(“someLong = 5”) .select(“someStr”) You probably noticed that the query is very simple, over top level columns. This would help us understand how ScORe works, and later on we’ll get into a more complex example. Back to our query. This is what the logical plan looks like: 'Project +- Filter (someLong#52L = cast(5 as bigint)) +- Project +- Relation parquet A LogicalPlan is a TreeNode object. Each TreeNode contains a sequence of TreeNode children, and according to its concrete type class, other properties. For instance, the Project is by itself a LogicalPlan, having a single child LogicalPlan, and a ProjectList. In our example, the ProjectList contains `someStr`, and the child LogicalPlan is the Filter. We iterate over the LogicalPlan bottom up, using TreeNode.foreachUp, as illustrated in Plan 1 (note: tree presented according to traversing path, so bottom is up and vice versa) , keeping a SchemaOnReadState object. Plan 1 - iterating the plan bottom up (note the tree is upside down) For each TreeNode in the LogicalPlan we apply different logic. For example, for the LogicalRelation, which holds an HadoopFsRelation, we store the relation reference in the state, and mark it as currentHadoopFsRelation. Upon storing the relation reference, we also create a RelationSchema object for it, with full schema being the LogicalRelation schema of the underlying data, and an empty StructType for schema on read. Moving on to the parent Project (Project_2 in Plan 1), we iterate the ProjectList and update the RelationSchema’s schema on read with the columns we identified. In this case our partial schema would look like this: root |-- someLong: long |-- someStr: string Next, visiting the parent Filter, we iterate on the condition and update the RelationSchema with the newly observed columns. In this case, someLong is already present in schema, so there is nothing to update. Finally, we get to the uppermost Project (Project_1), which already holds someStr, so again, no need to update.. Now that we’re done traversing the plan, we can use the created schema on read: Dataset<Row> df = sparkSession.read().parquet(“/path/to/parquet”); df = df.select(“someLong”, “someStr”) .filter(“someLong = 5”) .select(“someStr”); StructType schemaOnRead = SchemaOnReadGenerator.generateSchemaOnRead(df).getSchemaOnRead(“/path/to/parquet”); Dataset<Row> reducedSchemaDf = sparkSession.read().schema(schemaOnRead).parquet(“/path/to/parquet”); // repeat the same query reducedSchemaDf.select(“someLong”, “someStr”) .filter(“someLong = 5”) .select(“someStr”); Notice that the schema on read for our DF is accessed by the path to the data. We have to assume there’s no single schema to be provided for the query, as the query might represent a join between several data sources, each with its own schema. If we chose to provide an alias in the query, we also could have retrieved the schema by alias. #### What Are You Implying? Let’s review this simple Spark SQL query, but without the bottom project: df.filter(“someLong = 5”) .select(“someStr”) While we didn’t explicitly instruct to select someLong in the scenario below, it’s implied by the filter node. Plan 2 – field appears only in Filter node As we traverse the plan here, we again populate the schema on read for this query, but this time we only add someLong when we encounter the Filter, and then add someStr as we get to Project. The resulting schema is the same. #### Things Are Complicated... Let us consider a more tricky Spark SQL query with nested fields: df.select(df.col(“nestedStruct”).as(“myStruct”)) .select(“myStruct.str”) The LogicalPlan illustrated in Plan 3 Plan 3 - nested fields and aliases In Project_2 we query for nestedStruct. But we don’t really need the entire nestedStruct. In Project_1, we only take the str field from this nestedStruct. But to make things more complicated, we also have an alias for nestedStruct ? We need to keep track of this alias, and exclude the redundant columns from the nestedStruct schema. For each Relation we’re handling, we keep a map of SchemaElement objects. The SchemaElement can represent a concrete column from the relation schema, or it can represent a shadow instance of such a concrete SchemaElement, with the given alias. So when we encounter myStruct.str, we know that myStruct is an alias for nestedStruct, and that the str column is required. Eventually, when deducing the required schema, we know that all we wanted from this nestedStruct is str, so we can toss the rest of its inner columns. Another pleasant complication -- given the above query, suppose we sort it by nestedStruct (even though it seems to make no sense). df.orderBy(“nestedStruct”) .select(df.col(“nestedStruct”).as(“myStruct”)) .select(“myStruct.str”) Obviously, when we order by nestedStruct, we imply ordering by the entire content of it. If we had reduced our schema to have only nestedStruct.str, that would have changed the entire semantics of the query and yield incorrect results. So how do we decide if we need the entire nestedStruct schema (for ordering), or just part of it (based on projection)? We introduced the notion of conditional vs mandatory full schema. When visiting a TreeNode element while traversing the plan (see Plan 4), we mark whether the column requires the full schema (Sort node) or not (Project node). All of these attributes were taken into consideration while wrapping up the final schema on read. Handling orderBy is also one of the most dangerous manipulations we can apply. Unlike many other clauses in queries, if we omit childStruct out of nestedStruct, we will not get a runtime exception, but we will get wrong results. So by all means, test coverage was one of the top most concerns in this project. Plan 4 - Mandatory and conditional full schema Even with these 4 examples, we have still only covered the tip of the iceberg. Things get more complicated with aggregations, window functions, explodes, array and map types etc. ## Where trees are felled, chips will fly ScORe is not flawless, it’s a best effort solution. No doubt we have corner cases in which it will fail to generate a correct schema, or even fail to generate any schema. In fact, we do face such issues occasionally, and we address them as they come. A classic example is a Spark SQL query that involves invoking a UDF on some column with a nested schema. The UDF is a blackbox, and so ScORe must assume we need the entire schema, even if the UDF needs only a subset of the schema. There are other edge cases that we might encounter in production, as the syntax we support is mostly based on real production use cases. If tomorrow a developer or an analyst would use the PIVOT clause in a Spark SQL query, ScORe will probably fail. As any other production code in Taboola, metrics are our eyes and ears. Nothing gets to production without proper visibility, and the scale of our metrics pipe alone (~100M distinct metrics per minute) is a BigData operation in itself. At the Data Platform team, we also provide the infrastructure to execute Spark SQL queries over our data lake. ScORe is used by this infrastructure, so any query submitted, automatically benefits from it. Normally a Spark SQL query is implemented as a Java class implementing the SqlQuery interface, and executed at least once an hour. This interface includes the execute method that returns the DataFrame that we base the action on (normally saving the output to HDFS). Occasionally, the execute method itself is expensive even without performing an action. So we added the getBaseQuery method to the interface for the sake of ScORe, whereas the default implementation is just invoking execute. When an SqlQuery implementation is first encountered, we use getBaseQuery to create the schema on read, and cache the result in memory. The price of generating the schema is ‘cheaper’ by several orders of magnitude compared to performing the action (and yes, we measure it as well). We can then use the cached schema whenever we encounter the same SqlQuery implementation. If we fail to generate a schema on read for a Spark SQL query, we are notified by alert. We also monitor for incompatible schemas. This might happen, for instance, if the query structure is based on some feature flag, and someone has changed its value. In this case, we would attempt to automatically regenerate the schema on read, and replace it in cache. The getBaseQuery method is also useful when we face queries that SCoRe is not capable of handling properly. In the case of query using a UDF on a complex structure -- we can use getBaseQuery to provide the exact columns from this complex structure and bypass this limitation. For sql queries executed outside the scope of our datapipe infrastructures, developers can use ScORe directly in their code. For ad-hoc queries we’ve also modified zeppelins SparkSqlInterpreter to apply ScORe with a hint provided in the query. ## Quod erat demonstrandum Recalling the production query we presented earlier, reading nearly 350GB of data. Image 2 shows the effect of ScORe running the exact same Spark SQL query over exact same data. Total input size has decreased to less than 60GB (nearly 85% reduction), and total time across tasks was reduced from 60 to 15 hours, WOW! Image 2 – stats for stage reading input for production Spark SQL query using ScORe With ~700 compute nodes running Spark jobs in our on-prem clusters, ScORe was a game changer. I hate to imagine the waste of resources if we did not develop this tool. And as an engineer, I'm lucky that the scale in Taboola constantly poses new challenges, forcing us to look for further optimizations. At Spark, already significant progress was made for pruning nested schemas. These days we are working on upgrading our system to spark 3, so ScORe is still relevant to us as well as to anyone who is still using earlier spark versions. But even once we're fully upgraded, pruning in spark 3 still doesn't work for some syntax elements (e.g. window functions), so we expect ScORe to keep serving us for quite some time. You are welcome to tweet us for any question! --- ### TEST in PRODUCTION - should you? URL: https://www.taboola.com/engineering/should-you-test-in-production/ Last Modified: 2025-01-14 14:37:22 You wrote your code. You even tested it. And now, you are eager to git push it. But how can you verify that it really works? In Taboola, we test our code in production! In this article, you will see how every software engineer, even on the first day in the company, can test in production - all thanks to a dedicated Jenkins pipeline job and lots of metrics. ## How hard is it to test in production? Quite hard. You probably already knew that. Everybody fears that moment when they need to test changes in production. The main reason is that not everyone has the required IT skills. Moreover, people have to repeat error-prone, manual tasks - which might result in downtime and revenue loss. For our release engineers, it was also an unmanageable headache - a “thundering herd”  of developers eager to test their features in production. ## The problem - tedious, manual, error-prone tasks The process of testing in production was manual and tedious. Each developer had to deal with (at least) the following: - Judiciously choose a server in production - Make sure that the datacenter is not under stress - Take the server out of the load balancer - Install the feature branch - Perform functional and non-functional tests - Somehow verify that no performance degradation was introduced - Rollback to the previous version - And plenty more… We can all agree that this is too much for a human to memorize! ## Solution - Jenkins pipeline does it all Our release engineers built a fully automatic pipeline for verifying and testing - rapidly and easily - in production. In order to use it, all you need is a feature branch, and then - the magic begins: ## PREPARE - Jenkins judiciously picks a pair of available servers - Installs the feature branch - on the first server - Installs the release as baseline - on the second server ## TEST - During the test, both servers record all their logs and metrics - The developer tests the code with real production traffic ## WRAP UP - A script compares the logs and metrics - The script decides - ✔ pass / ✘ fail - Success criteria - below X% deviation from baseline After a few hours, the Jenkins pipeline ends and rolls everything back, like nothing ever happened! ## You get an email with the results The comparison script reads from Prometheus and Elastic. It verifies that the basic KPIs didn’t drop and that error logs didn’t peak above the allowed threshold. Below, you can see examples of these email messages, for successful and unsuccessful test results. Success Email (in green) vs. Failure email (in red) The comparison failures are highlighted - so it is very simple and straightforward to decide if your feature is ready, or needs more work. ## Drill down metrics in Grafana and Kibana To visualize the comparison results in one place, ad-hoc Grafana and Kibana dashboards with the relevant test time frames are created automatically. The links are sent by email and #slack. Example drill down Grafana dashboard ## Main branch is tested in production every 2 hours Actually, our release engineering teams use this method as a mandatory phase for validating new releases, and as an ongoing validation, every 2 hours. We continuously verify that the main branch is healthy. By doing that, we prevent bad code from entering production as early as possible. ## What did you just read? You just met a real continuous deployment enabler. The benefits of such a process is a fast, stable delivery process with improved quality. This is as close as it gets to the ”real thing”. This is a win-win for developers and release engineering teams, and of course, for the customer. ## Do you also need such a solution? Would you agree that now is a good time to start automating your testing on production? Feel free to reach out to us, we will be happy to share more details. - Co authored with Tidhar Klein Orbach --- ### The Challenges Of Uploading 150TB/day From Spark To BigQuery - Part 2 URL: https://www.taboola.com/engineering/challenges-uploading-150tb-day-spark-bigquery-part-2/ Last Modified: 2025-01-14 14:37:22 In part 1 of the series we shared the architecture of Taboola’s PV2Google service which uploads over 150TB/day to BigQuery. In this article (part 2), we’ll share the challenges and lessons we’ve learned over the course of a few years. ### Lesson 1: queries might be (extremely) expensive We continuously upload pageviews to BigQuery and keep them for six months. This translates to over 13PB of pageviews in BigQuery. Querying the entire dataset would be extremely expensive, about $65K/query (assuming $5/TB). We apply a few methods and guidelines to substantially reduce this cost: - Never use `SELECT *`: BigQuery’s query cost is based on the size of the data scanned. Most queries actually need only a few fields. Hence, selecting only the relevant fields will dramatically reduce the cost of the query. - Cluster tables: clustering is a neat BigQuery feature that reduces the scanned row count. With clustering, BigQuery optimizes the data layout to reduce the scanned data size when filtering on a clustered field. For example, with clustering on ‘siteName’ field, a query containing ‘WHERE siteName = example.com’ will only scan rows matching that filter rather than the entire table, reducing the cost significantly. - Partition/shard tables by day: most queries use a timespan of week or a month. Daily partitioning of tables enables scanning only the relevant days, eliminating the cost of scanning redundant data. - Sample tables: many queries calculate statistical information on the pageviews (e.g. click ratio). Such statistical data can also be determined from querying a sample of the pageviews. This is much more cost effective. Hence, in addition to the complete pageviews tables, we also have 0.5% sampled pageviews tables. Using the sample tables reduces the cost by a factor of 1:200. Figure 1. A query demonstrating: (a) Selecting only relevant fields. (b) Using publisherId clustered field to reduce bytes billed from 85.1GB to 500MB. (c) Querying a single day table ### Lesson 2: multi-layer retries are essential Ideally the PV2Google service functions without human intervention. On the other hand, the service is complicated and depends on other internal and external services. These services might temporarily fail or be unavailable. We do our best to contain and automatically recover from these problems. Airflow operators execute their tasks by submitting jobs to the PV2Google engine. Temporary Internet issues or Google service interruptions might cause these jobs to fail. To mitigate this, we leverage the Airflow operator retry feature. Airflow will automatically retry the failed operators a few times before giving up. Airflow operator retires are a simple solution that work well. However, it is an expensive solution because operator execution can be long. In most cases only a small portion of the operator work fails, and it can be recovered by retrying only the part that has failed. Retries at lower layers do the trick: - Spark task retry: a spark job builds and uploads the pageviews to Google Cloud Storage using thousands of Spark tasks which create thousands of files. It is likely that a few of those Spark tasks will fail due to Internet failures. To avoid failing the entire Spark job along with the Airflow operator, we use the Spark task retries feature. Each Spark task may retry a few times. In most cases this will save the job. If the problem is persistent, Spark will give up and the entire job will fail. - REST API retry: Airflow and the engine communicate over REST API. The API is vulnerable to internal networking problems. Hence, API retries are important. To retry at the REST API layer, we extended Airflow’s HTTP hook with a state-machine to gracefully recover from such problems. For example, retry on timeout, but fail (without retrying) on 4XX responses. These retries are transparent to the Airflow operator. Figure 2. Spark task 785, attempt #0 failed, retried and succeeded on attempt #1 ### Lesson 3: appends might corrupt your data In some PV2Google’s workflows, we append an hourly stage table into a daily production table. Appending the same stage table twice will lead to duplicate data. For example, Airflow might retry the append (copy) operator, which can lead to double append. Obviously, this must be avoided! The bright side is that Google jobs are atomic. They’re either completely successful, or in case of failure, do nothing at all. This is the key to avoid double appends. All that’s left, is to make sure we run Google’s append job exactly once. We do that by checking the status of the existing job before submitting a new one: - Job does not exist: this is the typical case (first append), when we definitely did not append the data before. Hence, we can safely execute the append job. - Job successful: we already appended the data. There’s nothing more to do. - Job faile: Something bad happened at the last attempt. The data was definitely not appended, because the job is atomic and failed. So we submit the job again. - Job running: We “adopt” the running job and wait for its completion. How do we know which BigQuery job ID to look for? The trick is to calculate a unique job ID ourselves rather than letting BigQuery generate it randomly. Before submitting the job, we attach the job ID to the destination table in a label to ensure we always can retrieve it. Figure 3. Table labels keeping job ids ### Lesson 4: mind the uploads duration A few hours after the end of each day, we re-upload the pageviews of the day to BigQuery to ensure BigQuery has the most complete and up to date pageviews, including data that arrived recently. The load size is 75TB and it takes about 10 hours. Initially, we uploaded the entire day in one step. This turned out to be problematic. If an upload fails just five minutes before its completion, we would have to retry and repeat the entire 10 hours upload. Furthermore, the longer the step, the higher the probability it would encounter a problem and fail. To mitigate that, we split the day upload into 24 independent hourly uploads. This aligns well with Airflow’s periodic processing. We just changed the Airflow 24 hour cycle to a one hour cycle. Figure 4. PV2Google Airflow DAG, data is uploaded every hour while it is copied to production only daily ### The devil is in the details The right architecture is important but not enough, the devil is in the details. Paying attention to the lessons described above makes the service scalable, reliable and cost effective. --- ### The Challenges Of Uploading 150TB/day From Spark To BigQuery - Part 1 URL: https://www.taboola.com/engineering/challenges-uploading-150tb-day-spark-bigquery-part-1/ Last Modified: 2025-01-14 14:37:23 Have you ever tried building an infrastructure to upload 150TB a day? Have you ever tried querying over 13PB without going bankrupt? These are some of Taboola's PV2Google (pageviews to Google) service scale challenges that we deal with in our day to day. In this blog series, we’ll share how we do it, and the challenges we face. In this article (part 1) we’ll focus on the architecture. Part 2 covers the lessons we’ve learned over the years. ### Hello, Pageviews! Taboola’s goal is to power recommendations for publishers and advertisers. Our platform serves over 360 billion content recommendations and processes over two billion pageviews a day. Pageview is a record describing recommendations, user activity (such as a click), and much more on a user's visit to a webpage. Currently, the pageview record has about 1,000 fields. Two billion pageviews generate a huge amount of data. This data is processed and analyzed using Apache Spark clusters, which have ~25K (on-prem) cores. On-prem processing works great for periodic analysis, such as billing. However, our analysts and development teams perform complex interactive queries on long-term data. These ad-hoc queries create peak demand for resources we don’t have. This is where Google’s BigQuery cloud service comes in. BigQuery is a highly scalable and high-performance database service. It lets you run queries over huge amounts of data (PBs) in a short amount of time. For example, BigQuery enables us to run queries on a month or more of pageviews in a matter of minutes. Furthermore, it enables us to store six months of pageviews, over 13PB. Obviously, this is not free of charge. BigQuery’s pricing model is by data storage and per query (the scanned bytes). PV2Google is a Taboola service responsible for building and uploading pageviews to BigQuery, over 2 billion pageviews per day. It builds the pageviews at our own data-center and then uploads them to Google cloud, about 150TB/day. At high level, the PV2Google service can be divided into three parts: - Data path - handles the data processing and the data relay - Control path - schedules the data path operations - Monitoring path - exposes the service status and alerts when things go bad ### The Data Path PV2Google’s data path handles the data processing and the data relay. This process is composed of three steps: - Building the pageviews and uploading them to Google Cloud Storage using Spark - Loading the pageviews from Google Cloud Storage to a stage table in BigQuery - Copying the stage table to a production table Figure 1. PV2Google’s data path The first step, building and uploading the pageviews, is a heavy and complex data-processing step. This step has to be parallelised and executed over hundreds of cores. We chose Apache Spark to do so. Each Spark task is responsible for processing about 1/3000 of the data. Each task reads the relevant events from Apache Cassandra datastore, builds its pageviews in memory, and uploads them to Google Cloud Storage as one (newline delimited) JSON file. The newline-delimited-JSON file format supports nested data structures, a key requirement to represent a pageview. Each page view is a line in that file. Lastly, the pageviews BigQuery’s schema is automatically generated based on the Java classes. At the second step we load the JSON files to a stage table in BigQuery. The stage table enables us to validate the loaded data before using it in production. This is a simple PV2Google step. All it does is submit a few load jobs to BigQuery. BigQuery does all the data heavy lifting. The last step is to append the stage table to the production table. BigQuery’s copy job is used to append the data. Appending the data is an atomic operation, ensuring that the production table’s data is always consistent. The above functionalities are implemented by a Java microservice named PV2Google engine. It is responsible for building the pageviews, uploading them, and controlling BigQuery jobs. The engine features a REST API to trigger and monitor its functions. ### The Control Path The control path schedules the data path operations. Every hour, PV2Google uploads an hour of pageviews data to BigQuery by executing the three data path steps described above. To do so, we better have a good control plane! We chose Apache Airflow for the control plane. Airflow is an open source platform used to schedule and monitor workflows. Each workflow is described by a DAG (Direct Acyclic Graph). Airflow schedules the DAGs to run periodically based on its cron configuration. The DAGs’ nodes represent operators. Each operator is a single task required to complete the workflow. Airflow manages the executions of its operators: ordering them, running them, retrying and monitoring them. Figure 2. Airflow and PV2Google control path PV2Google data path workflow is described by a DAG, which is composed of two operator classes: - Data path operators: execute PV2Google engine jobs. The operator submits an engine job and polls it's status until it completes using the engine’s REST API. - Validation operators: protect the process from invalid data. The operator queries the status and production tables to validate the uploaded data: row count, byte count and average record size. When a severe anomaly is detected, the operator would fail and halt the DAG execution. ### The Monitoring Path PV2Google users expect complete and timely pageviews at BigQuery. Most of the time, the process does that well. But, once in a while, things might go wrong: network issues might fail uploads, databases might be unavailable, Google might have temporary issues, and other unexpected problems might occur. Hence, good monitoring is a key part for the service availability and performance. Figure 3. PV2Google’s monitoring path PV2Google monitoring is composed of two parts: - Alerts: although we have retry mechanisms, some problems might require human intervention. A variety of checks detect such cases and call for human help using Pagerduty. - Visibility: visibility exposes us to a variety of metrics on the current and historical state of the service. In case of an anomaly, comparing the current and historical data is essential for troubleshooting the problem. In addition, the historical data exposes the service performance trends. This enables us to further optimize the service (e.g. add resources due to data growth). Every two minutes we query Airflow and PV2Google’s state databases looking for anomalies like a failed DAG run, or an operator that’s running for too long. There are two types of anomalies: - Alerts: triggered when things go wrong and human intervention is required. An alert creates a Pagerduty incident and sends a Slack notification. For example, a DAG run has failed or table row count is significantly low/high. - Warnings: warnings have lower thresholds than alerts. They warn us when the service is getting close to the edge and an alert might happen. A warning sends only a Slack notification. For example, DAG is taking too long and about to reach timeout or table row count is outside the expected range. For each operation we keep metrics and statuses in MySQL state tables. These metrics are kept for a long period of time. Grafana is a great tool to expose time-series data as graphs. We created various graphs showing: - DAGs and operators duration, retries count and state - BigQuery production tables metrics such as row, byte and average record size - Performance of DAGs, operators and engine jobs. For example, bytes upload rate Figure 4. Top: DAG duration, middle: build and upload files operator duration, bottom: load to stage table operator duration ### 150TB/day is complicated Building and operating a service that uploads over 150TB/day is a complex task. At this scale, the right architecture is important. Decomposing the service into a few paths, as demonstrated above, is a key to the service success. In part 2, we will share the lessons we learned from operating PV2Google service over the years, how to reduce cost, increase reliability, ... --- ### Don’t just run DNS, run it FAST URL: https://www.taboola.com/engineering/dont-just-run-dns-run-fast-2/ Last Modified: 2025-01-14 14:37:23 By Ariel Pisetzky and Tarek Shama ## Taken for Granted Taken for granted. That's the way most users and even techies think of DNS. Or more precisely, they just don't think of it at all. DNS is one of those things that for most users is a solved problem. You have a server with very reliable and stable software that can run for a very long time with little maintenance. The resolvers even have a nice built in failover mechanism for a secondary server. So, what more is there to say about this subject? Performance. Performance with DNS services has been seen as a geographical issue for years now. Yes, there have been paid DNS services that are faster at the DNS search level itself, especially if you are talking about complex records that have logic attached to them. Yet, the popular discussion is mostly around the global DNS providers. In this blog post, we would like to examine the performance gains to be had in HPC and data center environments. ## The New Way Over the past few months, the Taboola SRE team has moved away from stand alone DNS servers (and secondary as backup) in the different data centers and towards server clusters running under Anycast. The reasoning behind this is simple and straightforward, when the primary DNS server has an issue, while the resolver has a secondary backup, the failover time introduces latency. It’s that simple. The resolver failover time to the secondary DNS, while short, is still too long in the world of high performance computing. Moving from one DNS server to another causes the application to wait extra time that can be avoided. Yes, there is the option to set this timeout in the resolver.conf file (here is how), that is one path to take that can reduce the timeout issue. Even with reduced timeouts, you still have all the servers waiting that additional timeout period for an answer. That seems wasteful. On top of that, we wanted more bang for our resolver buck, so to speak. We wanted more servers in each data center. After all, this is a “downtime inducing” service when it goes dark. ## Building the Solution Our goto DNS solution for years was BIND. It works, it’s very well documented, and there is a large body of knowledge easily available for most of the cases one might run into. For us, after a short check of what other options are out there, we chose to continue with BIND and add anycast (and servers). This would allow us to continue with our existing automation tools for DNS and work on the load balancing aspect of the solution only. Choosing the clustering and load balancing mechanism for the DNS was the next step. While we already operate a few load balancing solutions, none were relevant in this case. DNS is a sessionless service, as such, introducing a network based anycast solution made the most sense for us. This choice means we don’t need additional software or hosts running as load balancers, we could use our existing network infrastructure. In our production environment (that is every data center) we have two DNS servers which share the same ip address, and the routing infrastructure does all the rest, it directs any packet to the topologically nearest instance of the service. Each DNS server is connected with two 10G links and each link goes to a different switch, both links are configured with bonding for redundancy. The IP address of the DNS service is the loopback address which is configured on each server. This IP is advertised via BGP from the servers to the switches, which in turn learn the route and allow the connectivity. There are multiple solutions for BGP but we have picked FRR (https://frrouting.org/). FRR is easy to deploy and manage, making it an easy choice (you should notice a theme by now). Adding the support of multiple routing protocols like BGP, IS-IS, OSPF, and the configuration management via puppet, is an added dose of gravy. ## Monitoring Making sure the servers are actually working and the network isn’t sending traffic to a malfunctioning server is an important part of the puzzle.Both health checks and monitoring need to be ready for production before the rollout. Our health checks include the service check itself and additional checks such as resolving, error rate and low successful resolving rate. We also check that the BGP routing is operational. There are a few more, and the idea is that some checks will kick the server out of the cluster and others will alert the team before taking action. Observability for the DNS is solved with the Prometheus exporter sending the metrics out and allowing us to view the DNS activity via the Grafana dashboard. The ability to graphically view different metrics of the DNS operations allows for easy troubleshooting and when introducing DNSSEC or other configurations into the mix, this is something that you want to keep your eye on. ## Into Production The rollout to production was based on our compartmentalized network. We operate 9 global data centers in an N+1 configuration. This allows us to deploy a critical infrastructure change such as this within a limited part of our global infrastructure and observe the behavior. While a change in DNS tech (on in this case stack) is scarry, the change was unobservable to anyone outside the SRE team. Once the first data center was operational under full load for a week, the other data centers came into play in rapid succession. Performance was also a boon when moving to the new configuration. In our production environment the resolving time was cut in half on average and in the 99.9 percentile by 16%. This improved performance is transferred to faster performance in the data center itself, as clustered applications gain from faster response times. Performance and stress testing was done using Flamethrower (by NSONE) that can be found here. ## Conclusion Investing time and engineering effort in any of the data center infrastructure services will always pay back. When that performance gain kicks in and your operations stay steady or when you have a DNS maintenance cycle and no one's the wiser. Keeping DNS running in the data center should be taken for granted. When a host asks for resolving, that A record should come flying back blazing fast. When the primary or secondary DNS fail, no one needs to know. No service should resolve slower.  Treat yourself to an anycast DNS today. --- ### Don’t just run DNS, run it FAST URL: https://www.taboola.com/engineering/dont-just-run-dns-run-fast/ Last Modified: 2025-01-14 14:37:23 By Ariel Pisetzky and Tarek Shama ## Taken for Granted Taken for granted. That's the way most users and even techies think of DNS. Or more precisely, they just don't think of it at all. DNS is one of those things that for most users is a solved problem. You have a server with very reliable and stable software that can run for a very long time with little maintenance. The resolvers even have a nice built in failover mechanism for a secondary server. So, what more is there to say about this subject? Performance. Performance with DNS services has been seen as a geographical issue for years now. Yes, there have been paid DNS services that are faster at the DNS search level itself, especially if you are talking about complex records that have logic attached to them. Yet, the popular discussion is mostly around the global DNS providers. In this blog post, we would like to examine the performance gains to be had in HPC and data center environments. ## The New Way Over the past few months, the Taboola SRE team has moved away from stand alone DNS servers (and secondary as backup) in the different data centers and towards server clusters running under Anycast. The reasoning behind this is simple and straightforward, when the primary DNS server has an issue, while the resolver has a secondary backup, the failover time introduces latency. It’s that simple. The resolver failover time to the secondary DNS, while short, is still too long in the world of high performance computing. Moving from one DNS server to another causes the application to wait extra time that can be avoided. Yes, there is the option to set this timeout in the resolver.conf file (here is how), that is one path to take that can reduce the timeout issue. Even with reduced timeouts, you still have all the servers waiting that additional timeout period for an answer. That seems wasteful. On top of that, we wanted more bang for our resolver buck, so to speak. We wanted more servers in each data center. After all, this is a “downtime inducing” service when it goes dark. ## Building the Solution Our goto DNS solution for years was BIND. It works, it’s very well documented, and there is a large body of knowledge easily available for most of the cases one might run into. For us, after a short check of what other options are out there, we chose to continue with BIND and add anycast (and servers). This would allow us to continue with our existing automation tools for DNS and work on the load balancing aspect of the solution only. Choosing the clustering and load balancing mechanism for the DNS was the next step. While we already operate a few load balancing solutions, none were relevant in this case. DNS is a sessionless service, as such, introducing a network based anycast solution made the most sense for us. This choice means we don’t need additional software or hosts running as load balancers, we could use our existing network infrastructure. In our production environment (that is every data center) we have two DNS servers which share the same ip address, and the routing infrastructure does all the rest, it directs any packet to the topologically nearest instance of the service. Each DNS server is connected with two 10G links and each link goes to a different switch, both links are configured with bonding for redundancy. The IP address of the DNS service is the loopback address which is configured on each server. This IP is advertised via BGP from the servers to the switches, which in turn learn the route and allow the connectivity. There are multiple solutions for BGP but we have picked FRR (https://frrouting.org/). FRR is easy to deploy and manage, making it an easy choice (you should notice a theme by now). Adding the support of multiple routing protocols like BGP, IS-IS, OSPF, and the configurationmanagement via puppet, is an added dose of gravy. ## Monitoring Making sure the servers are actually working and the network isn’t sending traffic to a malfunctioning server is an important part of the puzzle.Both health checks and monitoring need to be ready for production before the rollout. Our health checks include the service check itself and additional checks such as resolving, error rate and low successful resolving rate. We also check that the BGP routing is operational. There are a few more, and the idea is that some checks will kick the server out of the cluster and others will alert the team before taking action. Observability for the DNS is solved with the Prometheus exporter sending the metrics out and allowing us to view the DNS activity via the Grafana dashboard. The ability to graphically view different metrics of the DNS operations allows for easy troubleshooting and when introducing DNSSEC or other configurations into the mix, this is something that you want to keep your eye on. ## Into Production The rollout to production was based on our compartmentalized network. We operate 9 global data centers in an N+1 configuration. This allows us to deploy a critical infrastructure change such as this within a limited part of our global infrastructure and observe the behavior. While a change in DNS tech (on in this case stack) is scarry, the change was unobservable to anyone outside the SRE team. Once the first data center was operational under full load for a week, the other data centers came into play in rapid succession. Performance was also a boon when moving to the new configuration. In our production environment the resolving time was cut in half on average and in the 99.9 percentile by 16%. This improved performance is transferred to faster performance in the data center itself, as clustered applications gain from faster response times. Performance and stress testing was done using Flamethrower (by NSONE) that can be found here. ## Conclusion Investing time and engineering effort in any of the data center infrastructure services will always pay back. When that performance gain kicks in and your operations stay steady or when you have a DNS maintenance cycle and no one's the wiser. Keeping DNS running in the data center should be taken for granted. When a host asks for resolving, that A record should come flying back blazing fast. When the primary or secondary DNS fail, no one needs to know. No service should resolve slower. Treat yourself to an anycast DNS today. --- ### If you fall, fall right - a tale of SRE critical incident management URL: https://www.taboola.com/engineering/if-you-fall-fall-right-a-tale-of-sre-critical-incident-management/ Last Modified: 2025-01-14 14:37:23 # If you fall, fall right - a tale of SRE critical incident management ### By Yehuda Levi, Tal Valani, Ariel Pisetzky & Eli Azulai Imagine this scenario - your data center is down. 1500 servers are down. Each server needs to be handled and monitored and the responsibility for each should be divided between all teammates. Each team is looking after the status of their services. Client facing services are impacted. New information keeps flowing in from different channels and the status of the outage and servers keep changing. How to get the list of the server affected? How to put it all in one place? How to assign responsibility for each? What is the status of each server? How can the internal clients receive ongoing status updates? What happens if a server was intentionally down before the incident? What happens if a more complex issue occurs and the time to handle it takes longer than the outage itself? Jeannette Walls said “Most important thing in life is learning how to fall” and to be honest, during coronatimes we all need to get an extra lesson Handling critical incidents is always a hard thing to do. Working together in the same space (offices, remember that?) made it a bit easier to communicate. The physicality, the fact we are all in the same place and can just see or talk to each other easily, this all made a huge difference. In COVID time we found ourselves struggling to manage such easy communications, this directly inhibits our ability to manage an incident even more. We had to step up our game. Our journey started a few months ago, we had an incident involving multiple servers and the importance of working with a central task management system designed for IT incident management was crystallized. Not a console, not the central alerting system. This needed to be a crisis management tool that augmented our abilities in real time. Our first trial by fire was, well, just that! Firefighters raided one of our data centers. 1500 physical servers went down that incident (you can read about it here). During that day we were able to mobilize faster, respond in a much more coordinated fashion and bring back the services faster than ever. ## PART 1: Tell everybody something really bad happened The first challenge is to make sure that when something bad is happening, everyone who needs to know - will know (and fast). Yes, We have oncall engineers and multiple alerting systems. We have logs, metrics, dashboards, pagers, you name it, we’ve got it. On top of that we created an aggregated rule base service that monitors all kinds of sources. When a rule is matched - a set of actions can be triggered. For example: - paging multiple users. - sending notifications to various channels. - starting a conference call. - Initiating the "critical incident flow” ## PART 2: To the War Room Alarm! Alarm! All hands on deck! The system automatically opens the “War Room”, the incident board, where items can be assigned to engineers, priority can be set, and status can be updated. When we thought about this service, we wanted the following: It needs to connect multiple sources and generate actionable items. The UI has to provide a clear view of the incident. The different parts of the system should incorporate existing tools such as #Slack, PagerDurty, Zoom, Hashicorp Consul (to name a few). We decided to go with Monday.com as our dashboard and lineitem state engine. The system should be easy to build, maintain and ready to use in a short period of time. The War Room service is triggered and generates a list of items, builds a custom Monday dashboard and pushes the items into the board - along with status, priority and other properties. The War Room service updates the board items continuously until the incident is resolved. Simple yet not simplistic. Key features Multiple sources Collecting data from multiple sources - asset management systems, alerting systems etc. and creating a detailed list of items that contain physical and virtual devices, deployments, keepalive data and more. The service determines the priority of each item and sets it accordingly. Pre incident existing issues can be filtered out. Monday board For this task we created a Python library for Monday to assist us. We implemented some methods like: - Boards (create and archive). - Columns ( create, set values). - Groups (create, fetch, archive). - Items (fetch, create, update). - Subitems (create). We generate the right board for the incident Item population Once the items list is generated we push it into Monday and we keep managing the item’s state so we can update it in case of any change. Keep things up-to-date We really like our data up-to-date. So we make sure to keep updating each item every time we detect a change, so our incident manager can get the right calls Notifications Notifications are an important component of our service, we really need to make sure we can communicate when we need. Messages are pushed via various channels automatically to help facilitate streamlined communications with various other internal clients. Notifications are sent when the incident starts or finishes and we invite people to the boards and to a conference call. Status changes trigger notification for owners and subscribers. ## PART 3: Manage the incident The board is created and we can get right to work. Shift priority to where it is needed Assign / reassign the right engineer to the right task Get a clear status of the incident ## PART 4: After the storm A lesson should be learned Most critical incidents require a lesson learning process so we can improve. To assist this phase we can get a clear timeline based on the board log. So this simplifies the timeline discovery process, we know exactly what has been done, when and who did it. The ripple effect - Follow up on items Many times once the main event is over, we all just want to continue with the less urgent problems when the sun is out. This is a natural response when you get yanked into a crisis situation. Facilitating this need, open items will remain open after the incident is resolved. This keeps us on top of the lesser issues and assures the ripple effect ends well for all involved. ## Conclusions and what you take from this Once the system was online, there was no need to convince people to use it, it was just so clear how much value this brings to everyone using it. Within weeks from the launch and after the first inciendet, the “buy-in” from the organization was impressive. The system is now used by all production facing teams, SRE, R&D and more. Automation in SRE is the only path. There is no other way to manage the levels of tasks and the need to perform tasks on vast amounts of servers (or instances) if you do not have automation. This is the same for incident management. Reliability when breached is always a chaotic event. On or off the cloud you always need to find a way to bring order when multiple services or servers fail and fall out of production. You need a way to bring in more SRE resources to fix the problem. The ability to multitask the team members on the different tasks within a high pressure situation is a huge win. The War room automation was our way of pulling it off. Improving communications is always a good practice, when it helps reduce incident timelines that’s even better. --- ### Anomaly detection using LSTM with Autoencoder URL: https://www.taboola.com/engineering/anomaly-detection-using-lstm-autoencoder/ Last Modified: 2025-01-14 14:37:23 Taboola is one of the largest content recommendation companies in the world. We maintain hundreds of servers in multiple data centers around the world, while obligated to strict SLAs. Thus, you might understand why our engineers would appreciate a little heads up when the system gets overloaded. Like most companies today, we use metrics to visualize our services' health, and our challenge is to create an automatic system that will detect issues in multiple metrics as soon as possible, without any performance impact. ## A real life example Wouldn’t it be nice if we could predict the impact on our response time metric when major events are about to happen? For example “Black Friday”, “Cyber Monday” or even the “Kobe Bryant’s tragedy”, on the 26/1/2020? - as can be seen below: Figure 1: Kobe Bryant’s downtime 26/1/20 And yes - the gap with no metrics around the 26/1 is the downtime we had … Our current anomaly detection engine predicts critical metrics behavior by using an additive regression model, combined with non-linear trends defined by daily, weekly and monthly seasonalities, using fbProphet. Today, we get a single metric as an input and predict its behavior for the next 24 hours. Lately, we became interested in predicting the health of our entire data center based on multiple metrics given as an input to the anomaly detection engine. The motivation is to solve the common use case of an anomaly being detected in one metric - but there is no real issue, where multiple anomalies in several different metrics might indicate with higher confidence that something is wrong. #### Can we do better? In this blog, we will describe a way of time series anomaly detection based on more than one metric at a time. Our demonstration uses an unsupervised learning method, specifically LSTM neural network with Autoencoder architecture, that is implemented in Python using Keras. Our goal is to improve the current anomaly detection engine, and we are planning to achieve that by modeling the structure / distribution of the data, in order to learn more about it. In our case, we will use system health metrics and we will try to model the system’s normal behavior using the reconstruction error (more on that below) of our model. If the reconstruction error is higher than usual we will define it as an anomaly. Anomaly detection can be modeled as both classification and regressions problems. In classification we usually need tagged data (at least for some of the samples) and then we train the model to distinguish between samples labeled as normal vs those tagged as anomaly. In our case, we do not have labeled data. So we chose to model the problem as a regression problem where we try to quantify the reconstruction error of the model. Our hypothesis is that repeated events are easy to reconstruct since the system is mostly healthy, and healthy events are abundant while unhealthy events are rare. Thus, the higher the error, the more confident we can be that the current sample is an anomaly. ## Some preprocessing for breakfast Let’s start by examining our dataset, as you can see in Table #1 below there are 6 different important system health metrics. These metrics differ from one another in scale.  Datetime  Requests rate  Requests rate in a 5 minutes window  Failed requests rate Successful requests rate  Request response time p95 (ms)  Request response time p99 (ms)  04.11.2020 0:00  8,783  8,842  142,618   7,983  310.40   398.37  04.11.2020 0:05  9,971  10,307  170,124  7,319 306.07   399.30  04.11.2020 0:10  9,798   9,836  152,490  7,331  304.69  395.92  04.11.2020 0:15   9,879  9,939  153661   7,617  308.57  404.71  04.11.2020 0:20   10,086  10,158  153,972  7,188  306.93   400.64 Table 1: Example of our dataset. The metrics that were used as an input to our LSTM with Autoencoder model, and were calculated over the entire data center, each data center holds >= 100 servers. We chose to normalize the dataset by using min max scaler: # Python # ensure all data is float values = values.astype('float32') # normalize features scaler = MinMaxScaler(feature_range=(0, 1)) scaled = scaler.fit_transform(values) And here is the dataset after normalization:  Datetime  Requests rate  Requests rate in a 5 minutes window  Failed requests rate Successful requests rate  Request response time p95 (ms)  Request response time p99 (ms)  04.11.2020 0:00 0.37637 0.38709 0.34508 0.27990 0.26912 0.0533  04.11.2020 0:05 0.42730 0.45124 0.41163 0.25659 0.26165 0.05412  04.11.2020 0:10 0.41988 0.43063 0.36896 0.25701 0.25927 0.05127  04.11.2020 0:15 0.42334 0.43511 0.37179 0.26704 0.26598 0.05867  04.11.2020 0:20 0.43223 0.44470 0.37255 0.2520 0.26314 0.05525 Table 2: Example of the normalized dataset, after using min max scaler. #### Choosing a model or where the fun begins... We decided to use LSTM (i.e., Long Short Term Memory model), an artificial recurrent neural network (RNN). This network is based on the basic structure of RNNs, which are designed to handle sequential data, where the output from the previous step is fed as input to the current step. LSTM is an improved version of the vanilla RNN, and has three different “memory” gates: forget gate, input gate and output gate. The forget gate controls what information in the cell state to forget, given new information that entered from the input gate. Our data is a time series one, and LSTM is a good fit for it, thus, it was chosen as a basic solution to our problem. Since our goal is not only forecast a single metric, but to find a global anomaly in all metrics combined, the LSTM alone cannot provide us the global perspective that we need, therefore, we decided to add an Autoencoder. Figure 2: The Autoencoder architecture An Autoencoder is a type of artificial neural network used to learn efficient data encodings in an unsupervised manner. The goal of an autoencoder is to learn a latent representation for a set of data using encoding and decoding. By doing that, the neural network learns the most important features in the data. After the decoding, we can compare the input to the output and examine the difference. If there is a big difference (the reconstruction loss is high) then we can assume that the model struggled in reconstructing the data, thus, this data point is suspected as an anomaly. The LSTM Autoencoder is an implementation of an autoencoder for sequential data using an Encoder-Decoder LSTM architecture. By using this model we can have the benefits of both models. ## Lunch time #### The LSTM Model #Python self.model.add(LSTM(units=64, input_shape=(self.train_X.shape,self.train_X.shape))) self.model.add(Dropout(rate=0.2)) self.model.add(RepeatVector(n=self.train_X.shape) self.model.add(LSTM(units=64, return_sequences=True)) self.model.add(Dropout(rate=0.2)) self.model.add(TimeDistributed(Dense(self.train_X.shape))) self.model.compile(optimizer='adam', loss='mae') #### Fit self.history = self.model.fit(self.train_X, self.train_y, epochs=50, batch_size=72, validation_split=0.1, shuffle=False) For training we’ve used features as the health metrics described above, and we aimed to predict what would be the data center health in the near future. Let’s focus on line #9: - train_x is a vector containing the server health metrics at some point of time. - train_y is a vector containing the same metrics at a later point in time. ##### Note: We are using shuffle=False in line #9 because in a time series data the order is important. #### Predict # Python X_test_pred = self.model.predict(self.test_X) test_mae_loss = np.mean(np.abs(X_test_pred - self.test_X), axis=1) test_mae_loss_avg_vector = np.mean(test_mae_loss, axis=1) For our “Reconstruction error” we used Mean Absolute Error (MAE) because it gave us the best results compared to Mean Squared Error (MSE) and Root Mean Squared Error (RMSE). In both MSE and RMSE the errors are squared before they are averaged, this leads to higher weights given to larger errors. This causes the model to be more sensitive to noise which might cause false positives. Since our data is noisy by nature, we defined (a business decision) that an “anomaly” is a spike or a trend that is lasting at least 10 minutes. Therefore, in this model we needed a loss function that is more “forgiving” to small spikes in a feature or two. #### With that in mind, let's start plotting! ### Static threshold The simplest way of deciding what is an anomaly could be: “anything greater than a fixed threshold considered to be an anomaly, otherwise normal”. Figure 3 presents the reconstruction error, which is being measured by the mean absolute error (MAE). In addition, we plotted a static threshold which is calculated as the overall mean + 2 std steps of the MAE on the entire trained data. In figure 4 we can see the points that are detected as anomalies according to the static threshold defined, plotted over the total success request rate actual data. Figure 3: Mean loss over time with a static threshold Figure 4: Total success request rate anomalies, using a static threshold We differentiate between two types of anomalies: A local anomaly (green points) is triggered when a single metric loss crosses a threshold, and a global anomaly (yellow points) is triggered when the mean loss of all metrics cross a threshold (which is lower than the local anomaly threshold). When both local and global anomalies are triggered, we colored it as pink. ### Dynamic threshold Figure 4 is noisy and full of anomalies while we know they are not. The “noise” is seasonality, which made us realize we should use a dynamic threshold which is sensitive to the behavior of data. We can change the static threshold by using rolling mean or exponential mean, as presented in the graph below. Figure 5: Loss over time with static and dynamic thresholds Figure 5 presents the mean loss of the metrics values with static threshold in green, rolling mean threshold in orange, and exponential mean threshold in red. We decided using the EMA (i.e., Exponential Moving Average) threshold for detecting anomalies. The alpha is the smoothing parameter. Higher alpha values will give greater weight to the last data points, and this will make the model more sensitive. Following, are the results using the EMA on the total successful request rate metric. Figure 6: Total successful request rate anomalies Success! The exponential mean threshold is less noisy and here we predicted a major drop in successful action conversion metric 30 minutes before the disaster really happened! #### Better be smart than right While implementing the LSTM model, we tried two different strategies, the Batched model and the Chunked model. - Batched model - This model gets all the data at once, splits it into train and test sets. The bigger the batch - the more accurate the model, but more expensive in resources and prune to drifts and changes in the data. - Chunked model - This model gets the data in small chunks (e.g., 5 minutes chunks) and is being updated online. This model can sometimes be less accurate than the batched model but it is much more dynamic. ### In sum, We recommend using the batched model when conducting a POC in order to estimate how good the model is. In production, the data mostly comes in a stream, thus the chunked model is better. ## The Golden Model for desert ? The following parameters were chosen by us to create the best model using the above architecture. - train_size = 0.7 - epochs = 50 - batch_size = 72 - n_nodes = 64 - time_steps (how many time steps ahead we want to predict, each time step stands for 5 minutes, so 6 time steps = prediction of 30 minutes ahead) = 6 - prediction_size (how many days ahead we want to present in our prediction graph) = 1 We are on Github! - Batched Model - Chunked Model Acknowledgements This blogpost was created by Michal Talmor and Matan Anavi, undergraduate students from the Software and Information System Engineering at Ben-Gurion University, who were mentored for four months in the Starship internship program by Taboola mentors Guy Gonen and Gali Katz. --- ### Running a Post Mortem Following "The Moment" URL: https://www.taboola.com/engineering/running-a-post-mortem-following-the-moment/ Last Modified: 2025-01-14 14:37:24 Failure. I need to talk about failure, and not any failure, my failure. I need to share it with everyone in the production group, everyone in R&D. My team, my peers, my managers. The meeting will start in just a few minutes and I am under fire to explain what went wrong, how I failed the organization and how we need to be better. General George S. Patton Jr. said “The test of success is not what you do when you are on top. Success is how high you bounce when you hit the bottom.” There is a lot to learn from that saying, and not only for people. Successful systems need to bounce back from a failure and do it well. IT systems need to be able to endure a catastrophic event and just dust it off. This is what we expect of our production systems in Taboola, nothing less. ## Crisis manager Back to the meeting, I was the crisis manager during a data center failure. (you can read more about it here - https://www.linkedin.com/pulse/moment-ariel-pisetzky/.) The entire data center went dark and it’s my team that needs to bring it all back. It’s my job to lead the effort through the crisis management process to a successful conclusion. One of the most important parts of that crisis management process comes a few days after the dust settles in the form of a post mortem review. It’s minutes before the post mortem meeting and I’m already connected to Zoom. How do I embrace the failure and bounce back? There are about 20 people in the meeting. Everyone here was impacted; from working into the night with restore procedures to validating reports and making sure all the systems and data are in perfect condition. It’s easy to look for people and external reasons to blame. “It’s the provider that didn’t follow through” or “the priority wasn’t set, so it wasn’t done” and so on. It’s so very easy to blame the system. That is not what we do. We make sure to go over the facts and use no names. ## Post mortem We never use names in a post mortem. A name is associated with blame. The point of the post mortem is to learn, improve operational excellence. One can say, to bounce back and be better. There is no way we can talk about the facts if there is fear. Fear that names and reputations will be dragged into the mud. I start the meeting with the facts and timelines. If I did my job right it’s documented in #Slack and we should have a good account of what we did and when we did it. The alert system and supporting graphs are also reviewed, as are the logs to make sure we checked the relevant engineering angles of the incident. Once we have the facts out and the timeline sorted it’s time to talk about the failure. We love what we do, and we love the solutions we came up with. It’s hard to look at a system you built and say, this is the failure point. This is the design flaw that caused the problem. It’s exactly what needs to be done. It’s time to voice the failure points in the system that caused the incident. It doesn’t matter if this is internal, external, code I wrote or a system someone else manages. The root cause is the raison d'etre of the meeting and needs to be debated. Yes, additional factors come into play: was there monitoring? was there alerting? Did we fix the problem fast enough? All of these need to be debated and improved. But we need to start with the root cause. ## Post Mortem Rules Any good post mortem will assure the participants have the confidence to share everything and anything they know about the failure. So here are the ingredients we mix in our culture and post mortem meetings so you can do the same: - It’s OK to fail. It’s part of what we do and how the innovation process works. You can and will fail, make sure to learn from it. - We don’t blame anyone for a failure. To further help this, we don’t use names. If it’s a code push or a configuration change, we don’t name the person or team that did it. It’s just not relevant to the discussion. - The most important part of the meeting is the focus on improvement. What needs to be done, what monitoring should be set in place and what alerting is missing. - No matter how bad the incident was, we make sure to use the crisis as a learning experience. The SLA budget was already spent, the lost revenue already in the books. You can lose all that budget, or make sure to use it for a learning experience. Sharing failure should be easy. It’s not. Talking about my failure in a data center downtime event in front of all my colleagues is easy. It’s because we all make sure to take the best out of the incident and to fix everything we can before a data center goes dark again. It will fail again as data centers do fail from time to time. It’s how we respond to that failure that defines our success. ## Take Away from the Post Mortem This was our first major production incident in the COVID work from home era, so we took a few lessons from our incident, most of them in regard to war room management when we are all home. The most valuable lesson we learned? Multiple and parallel Zoom meetings with a few people acting as production liaisons. It worked like a charm and helped us immensely to bring back the system from the DR event. Each task force working on a specific realm of issues had a person designated as a liaison that was attentive to the task force Zoom and the main war room Zoom. This allowed fast communications and feedback to the other task forces working on the recovery process. From our past experience we were able to cut 90% of our recovery time from the last DR test in this real time event. Much of the saved time was due to improved communications. So, how was it to talk about my failure to a group of Taboolars? It was easy, it was insightful and it helped me, my team and Taboola improve. Being the person on-point to lead the process was not daunting nor scary, it was a chance to improve our SRE skills, our production procedures and our promise to provide the best platform for our clients. ### P.S If you have not read Radical Candor by Kim Scott, it’s a good example of how feedback should be provided and why it's important. I recommend reading it. --- ### That Moment URL: https://www.taboola.com/engineering/that-moment/ Last Modified: 2025-01-14 14:37:24 So, the firefighters are in your data center, there is no electricity, and the pager is more like a DDoS attack on your phone than anything informative. You look at your watch, multiple thoughts running through your head. Why me? Why now? What was the last DR test result? How do you pull the team out and through this IT catastrophe and survive to write about it? This is my story, my personal fight with the IT “Murphy laws” and how we can all benefit from it. It was a Friday, one you know you need to be extra careful with. It's always the end of the work week or smack in the middle of the night. (No IT catastrophe ever happens when it's convenient to you, now does it? They always cluster and bunch around the most difficult times.) Anyway, it's the end of the day Friday and multiple systems just go dark. You get that specific ringtone you configured for PagerDuty and it just will not stop playing. You get "The feeling" (with a capital T). You know you are connecting into a bad situation. Oh, and it's already in the middle of your BCP (as it's Covid-19 time) and things are hard all over. pictures are all taken from this very real life event by the dedicated Taboola data center logistics staff Connecting to the VPN (sigh of relief, the dual VPN solution in 2 different data centers is working) and the basic dashboards show that the front-end services are operational, yet the war room procedure was triggered. That means that the monitoring set in place for a large-scale downtime event was matched. While you're connecting to the Zoom call, you get the following message in the #Slack channel for production incidents - "There is no power in the backend data center". Now everything starts falling in place, the multiple alerts, the services that are down and all the possible bad things that are yet to come. Wait, one more thing pulls your attention. It’s the data center logistics group direct messaging on the phone. Now, you already know there isn’t power, so this message is now a higher priority than others and you open that thread: there is some text and a picture. Skipping the text you open the picture and you see that firefighters are in the data center. Many things go through your mind, but the one that is high up there - at least they don’t have hoses and you don’t see water anywhere in the picture. ## Crisis Manager First things first, I'm the crisis manager, so getting information and prioritizing actions is my job. I would say something like “Aviate, Navigate, Communicate”, only for IT/SRE, but I haven't found the magic acronym that works for me. So what’s important when managing a large scale IT incident that can develop into a full out business downtime?: - What services are impacted? - What is the reason for the service interruption? - What is the business impact of each service interruption? (for prioritizing the next steps) - Do we have a plan to fix it (each interruption you identify)? - How do we best communicate to the organization (and clients if need be) what we know and what to expect? ## First things first Now our top priority is safety and human life. That might sound pretentious as we are a software company and we do not manage life support systems, but we do run data centers and these are high energy locations with fire hazards and fire suppression systems that are not human friendly. So, once it’s clear that everyone is safe and the data center staff is not at risk, our next priority is to bring the business impacting services back online. Be it via the DR plan or just brute forcing through the challenges, it's clear that there is a lot of work ahead of us. From making sure we have task forces working on the right problems (that’s two issues to manage right there, assembling the task force & prioritizing the tasks) to making sure the situational awareness is as complete as possible. When managing a large-scale crisis with multiple engineers from different reporting structures helping in, it’s easy to get lost. Before diving into the game plan and this specific incident, it’s important to understand the rule of the crisis manager. What does this person do and why do we need this hands-off the keyboard person in the mix? Engineers are more than capable of solving problems, complex problems, especially if they were part of the team writing the code. Over the past years, with the DevOps movement it has become more and more acceptable to have coders on call and not only the SRE team. Something along the lines of “you built it, it’s your responsibility to keep it operational”. The engineers (and specifically the coders among them) are now also part of the team assuring the services are up and running. Having people from different backgrounds, different teams and different approaches to problem solving is a great thing! It’s also a challenge. Solving problems in production is very different than working on a project. The time pressure, the SLA needs, communications to the business and different groups working to resolve the incident are now also factors in getting things done. This is why you want to have a crisis manager. Once you cross some scale threshold (each organization needs to set that threshold) you want to have that position well defined, to help steer the work effort and bring the problem to a speedy resolution. IT problems are nothing new. Even the term “bugs” traces way back to Rear admiral Grace Hopper from the 50s, and to Apollo missions and other places over 50 years ago. With the rise of ecommerce and internet services the realm of service management moved from the work of the few to a business need that is now touching almost every business. Finding frameworks or methodologies of solving IT incidents is well documented under the ITSM, ITIL and other standards. One that I found easy to use for the uninitiated is here (thanks Atlassian for publishing). In our IT organization, the rule of incident managers is not set, but defined in the moment based on the availability of the personnel on call. That said, we do have a short list of people on-call that can act in this capacity. ## Solving the problem Back to our incident, so the firefighters are on site, it’s already clear we don’t have power. The list of impacted services is mostly well defined (I write mostly, as services that failed over to DR might not be keeping up over time). We already have a Zoom war room open and active and my first move as incident manager is to type messages in the relevant #Slack channel. This choice is driven by two factors, the need to communicate as far as possible in the organization and the need to pull in additional people even if they are not on call (aka - all hands on deck). Next we need to define priorities according to business needs. This is very specific to each business, so I will share in broad strokes that the needs come from SLAs, cost, on-line and offline processing. So, while the billing system or backup systems are super important, and might take longer to restore the longer we wait, they don’t get higher priority over client facing systems that directly impact the business SLAs or client operations. What's the next step? Setting the task force teams onto the prioritized tasks and understanding if there are shared resources between them, that is, if any task force needs a resource that is in use by another (this can definitely be people, as network skills are in need or the data center logistics team is called on for physical actions). While working on the the task force resource allocations, I make sure to document it all in the #Slack channels, so it’s clear to anyone in the organization, who is working on what. The main idea here is that each task force needs to have some interaction with the business to make sure the situational awareness is accurate. While the teams get into action (preferably in other Zoom calls), one person from each task force acts as the liaison to the main war room Zoom. This is important as it helps with feedback and again that so important situational awareness. While we are finishing our DR run and bringing services back up, the power is restored in the data center and we now have new avenues of action open to us. Do we leave the DR site as the main site for the weekend or roll back to the main site? The optimal situation would be to just set up the main site as the new DR and make sure the systems are back online and operational. On top of that make sure that the lower priority services are back online. ## Working through a crisis There are a lot of details, a lot of work and some of this is just grit. Making sure that the long hours in Zoom, asking people what can and should be done isn’t overlooked. Making sure it’s all documented and while people are talking you are typing feverishly on your keyboard to let others know what’s going on. Yet it does boil down to grit. The ability to continue on while the path to service availability isn’t always clear. Yes, you should always have a plan, a DR, a culture of chaos engineering, the ability to fail partially and not have all the service come crashing down when something isn’t working. However this isn’t always the case. And then there is Murphy: any good plan or well-documented system will always have some new and unexplored ways of failing. It’s at that point where you need to make sure the language of managing IT incidents is clear. That the vocabulary is in place to talk about business impacts, time to resolve, task forces, communication liaison and meticulous documentation. If it doesn't help you solve the issue faster, it will absolutely help you learn how to do so for the next time. There is always the next time. That next time is something you need to make sure you prepare for. When we have suffered an IT crisis, especially if it was “man made” (as most of them are), it’s important to learn from this mistake. In a quote attributed to Winston Churchill “Never let a good crisis go to waste” you can find the embodiment of the idea behind running a post mortem. That is, if you already persevered, suffered the hardship and fixed something, use that to improve. Otherwise, you are compounding the loss from the crisis. Not only do you suffer the loss of revenue or customer trust, but you also missed a learning opportunity and increased the chances of the same problem happening again. Our data center didn’t burn down. Faulty wiring in our fire suppression system triggered the fire alarm that in turn cut power and blasted the “offending” electrical circuit with a suppression agent. This in turn called the firefighters and defined the data center blackout duration (firefighters on site, making sure air quality is back to normal, understanding that the fire suppression backup system is up to the task for return to normal operations). I can only imagine what my reaction was when I saw the picture of the firefighters, my focus on incident management was so intense, I just remember the world slowing down for me to get all the task force teams lined up and the feeling of crazy rush of adrenaline coursing through us all in an attempt to keep all business critical systems up and running. pictures are all taken from this very real life event by the dedicated Taboola data center logistics staff --- ### High Scale Service Deployment: Taboola’s Recommended Flow URL: https://www.taboola.com/engineering/high-scale-service-deployment/ Last Modified: 2025-01-14 14:37:24 This post is not about K8S - nor is it about AWS. It is not about containers - nor is it about some new, “cool” technology for managing large-scale applications. Rather, this post is about how we deploy a highly sophisticated Java service, a heavy service that is very actively developed on a daily basis, to 1000s of servers across our 7 data centers around the world. So what’s the problem? Isn’t it enough to take a list of servers, get the version to deploy and run it with an automation tool like ansible? Well, it’s not as simple as it might seem. This service serves Taboola’s recommendations and responds to hundreds of thousands requests per second. The service has to be fast - so fast that its p95 should be below 500 milliseconds per request. Which means we can’t have any downtime at all, or even afford slower responses. In addition, it’s critical to prevent the installation of a faulty version. A faulty version could lead to downtime or degraded performance, which can directly result in a loss of revenue. For this reason, we have multiple testing gateways during development - to help prevent a bad version. However, based on our experience, sometimes when the software meets production, unexpected (bad) things happen - and we need to be ready to prevent that. Another important requirement is to deploy during office hours, when most of the engineers will be available to assist should something go wrong. So how do we do that? In the following post, I will explain how we deploy our Recommendation Service at Taboola. The post will detail the deployment flow, step by step. ## Stage 1: Is today a deployment day? The first thing we want to do is to check if today is a deployment day. The requirement is to deploy during working hours, so if it’s a weekend or a holiday, we skip the deployment. ## Stage 2: Is today’s version valid? It’s a deployment day and we want to know that the code is valid and isn’t going to cause any production issues. This is why our first step is canary testing. In this stage we deploy the new version to a single machine in each of our 7 data centers around the world. After the service is ready, we let it run with real traffic for one hour. During that hour we gather various metrics - from the newly deployed server, as well as a server with the old version. These servers must have the same hardware specs; in addition, we also restart the old version to make sure the environments are as equal as possible. When the hour is up, we compare the results. Every metric has a well defined threshold. If the newly gathered metrics are within the boundaries of those thresholds, we can continue to the next stage. If the new version “misbehaves”, we stop, alert the relevant developers and analyze the issues. After the analyses, we can still go either way: abort the deployment (will happen automatically) or approve manually. Example for metrics compare from the canary testing ## Stage 3: Data center verification - part 1 The new version seems to be OK - at least as much as we can verify in 1 hour - and we decide to continue with the deployment. However we still want to ‘play it safe’. Enter stage 3: data center verification. We start deploying slowly on only one of our data centers. The first data center is carefully chosen by production engineers. It has to be outside of peak hours in that specific geographic area, but still have enough traffic for us to verify. The deployment procedure on a single data center goes like this: - Get the list of servers to be deployed - Calculate the size of the server batch (see below) - For each server in the batch Silence all alerts - Stop the old version and remove it - Install the new version - Start the service - Verify that the service started correctly - Unsilence all alerts - Run a batch verification to check various metrics of the domain - Wait for a minute for the next server batch - Repeat until no servers are left When the deployment on the data center finishes, we enter the validation phase - but first, let’s get back to the calculation of the server batch. ## Let’s do some math As mentioned above, when we deploy the new version, we don’t want to affect our ability to serve requests in the appropriate response time (<= 500 milliseconds). Therefore, we developed a formula that will tell us how many servers we can take out of the data center’s pool without causing “noise”. The formula looks like this: Where: - Xis the amount of servers we can currently stop. - Sis the total amount of functioning servers in the data center. - Rtotalis the total amount of requests in the data center. - RTavgis the average response time per request. - RTmaxis the critical limit for the response time. - Threadsare the amount of threads we use per server. When we get X, we normalize it so we won’t stop too many servers in off-peak hours - or too few servers in peak hours. Batch size calculation from June 3rd to June 10th ## Stage 3: Data center verification - part 2 OK - we deployed the first data center, and now we want to validate that it behaves correctly. We pause the deployment for one hour and perform several checks. The important ones are: - Check that the response time did not increase. - Check that the server logs contain no unusual errors. These metrics are monitored periodically, regardless of the deployment flow. In order to catch production issues faster, we integrated these checks into the flow. Any issue that arises at this point will raise an alert and pause the flow, until it is either approved manually or is manually or automatically aborted. ## Stage 4: New version for all When we get the "green light" from the first data center deployment and verification, the flow will automatically continue to deploy the rest of the data centers in parallel.   Deliver value and quality in 6 steps Want to build your own deployment process? We suggest implementing the following steps: - Test your code continuously - although not mentioned yet, code review is mandatory and unit and integration tests are a crucial part of code validation. - Collect metrics - both for the deployment and the production service - all the time. - Build a process that can validate the new code in production with real data/traffic. Compare with the metrics you collect. - Define the proper pace for deploying - you probably want it to be dynamic. Figure out the right pace for your specific needs. - Continuously validate that the new code doesn’t break anything. - Continuously improve. If you have questions, suggestions or comments, feel free contacting me at @tizkiko Good Luck!   ## Technology used A note about the tools we used to build the flow. The service is written in Java and packed as rpm. The logic of the deployment flow is written with Jenkins pipeline, mostly Groovy with some calls to shell commands (yum install, for instance). We use Grafana and Prometheus for monitoring and metrics. I would like to take this opportunity to thank Taboola’s Release Engineering and Production Engineering teams, who helped build this great process. --- ### Using Spark Dynamic Allocation URL: https://www.taboola.com/engineering/using-spark-dynamic-allocation/ Last Modified: 2025-01-14 14:37:24 The story starts with metrics. Every mature software company needs to have a metric system to monitor resource utilisation. At some point, we noticed under-utilization of spark executors and thier CPUs. Usually, dynamic allocation is used instead of static resource allocation in order to improve CPU utilisation through sharing. In this blog post, we'll define the problem, share the goals we worked towards and highlight many technical peculiarities regarding dynamic allocation usage along the way. At Taboola, we use Grafana, Prometheus with a Kafka-based pipeline to collect metrics from several data-centers around the world. Metrics at scale is a very interesting topic and involves multiple problems in itself and we have previously covered these in our blog and meetup presentations. Our data platform comprises several services that compute data projections and, importantly, those are long-running processes with long-living spark context. Periodically, when triggered, these services process new chunks of data, however, they block until the following occasion leaving the resources unused while no other framework can use them due to static resource allocation of cores. Here is a Grafana dashboard that shows the problem: The total number of cores taken from the Mesos cluster is invariably 500 while actual usage peaks at 400 occasionally leaving the cores idle a lot of the time. We can define our goals as: - Make better use of available resources - Improve end-to-end processing time One way to release unused resources in the static cluster (we are running on-premise, with a static number of Mesos-worker nodes) is to start using a dynamic allocation feature. ## What is dynamic allocation? - Spark provides a mechanism to dynamically adjust the resources your application occupies based on the workload - Your application may give resources back to the cluster if they are no longer used and request them again later when there is demand - It is particularly useful if multiple applications share your Spark cluster resources So what is happening under the hood? Spark driver monitors the number of pending tasks. When there is no such task or there are enough executors, a timeout timer is installed. If it expires, the driver turns off executors of the application on Mesos-worker nodes. Other executors might still want to access some of the data on the executor that completed all its tasks so we need an external shuffle service to provide them with a way to continue accessing shuffle data. ## How to start - Enable External shuffle service: spark.shuffle.service.enabled = true and, optionally, configure spark.shuffle.service.port - Enable dynamic allocation feature flag: spark.dynamicAllocation.enabled = true - Provision external shuffle service on every node in your cluster that will listen to spark.shuffle.service.port ## How to make sure external shuffle service is running on every mesos-worker node The natural approach is to use Marathon. You can think about it as "init.d" for Mesos cluster frameworks. By default Marathon can decide to distribute the service instances across the cluster so that some machines will host more than one shuffle service instance or none at all. This is undesirable as we would like to have strictly one instance per host machine. To ensure at most one service is instantiated we will use service constraints that allow us to specify max instances per machine explicitly. For example, we can add an entry to our config: "constraints": ] Equally, to ensure at least one service can be placed on a machine we will simply reserve resources designated to this service. ## Static resource reservation for the “shuffle” role - The needs of the external shuffle service will be fulfilled by the "shuffle" role in Marathon terminology. - We configure the mesos-agents on each node to report resources to the cluster taking the shuffle needs into account. The following launch params are used for the agent: --resources=cpus:10;mem:16000;ports:;cpus(shuffle):2;mem(shuffle):2048;ports(shuffle): Let’s break down these parameters one by one: - The default port range for the mesos agents is 31000-32000. - We are allocating 2Gb of RAM for the external shuffle service. - We are allocating 3 ports (7337 to 7339) for external shuffle services (for green-blue deployments, different spark versions etc) - The resources might be over-provisioned (the 2 cpus dedicated to the shuffle role are not included in the total count of 10 even though there are actually only 10 cpus in total on the machine). To set-up Marathon masters to use resources correctly, we need to add the same role (shuffle) to the --mesos_role parameter when launching. Now we made sure that the external shuffle service will get its own resources to run exactly once on each node regardless of the resource utilisation. During testing in the staging environment we discovered that after 20 minutes, the tasks started to fail due to missing shuffle files. It seemed that spark management of shuffle files has its corner cases. ## External Shuffle Service and Shuffle files management As mentioned before, external shuffle service registers all shuffle files produced by executors on the same node and is responsible to serve as a proxy to the already dead executors. It is responsible for cleaning those files at some point. However, a spark job can fail or try to recompute files that were cleaned prematurely. - There are some traces of the problem out there, e.g. SPARK-12583 - solves the problem of removing shuffles files too early by sending heartbeats to every external shuffle service from application. Driver must register to all external shuffle services running on mesos-worker nodes it have executors at - Despite the complete refactoring of this mechanism, it still doesn’t always work. We opened SPARK-23286 - At the end (even if fixed) it's not good for our use-case of long running spark services, since our application “never” ends, so it's not clear when to remove shuffle files - We have disabled cleanup by external shuffle service by -Dspark.shuffle.cleaner.interval=31557600 - We installed a simple cron job on every spark worker that cleans shuffle files that weren't touched more than X hours. This requires pretty big disks in order to work to have a buffer. So, we adjusted our external shuffle service parameters. Here are details on how to install this service on marathon. ## Defining External shuffle service to run as marathon service - Marathon supports REST API, so you can deploy service by posting service descriptor as follows curl -v localhost:8080/v2/apps -XPOST -H "Content-Type: application/json" -d'{...}’ - We commit json descriptors to source control repository to maintain history - The Marathon leader in quorum runs periodic task to update if necessary the service descriptor through REST-API - Following is Marathon service json descriptor for shuffle service that runs on port 7337: instances are dynamically configured - Using Mesos REST-API to find out active workers - Using Marathon REST-API to find out number of running tasks (instances) of the given service { "id": "/shuffle-service-7337", "cmd": "spark-2.2.0-bin-hadoop2.7/sbin/start-mesos-shuffle-service.sh", "cpus": 0.5, "mem": 1024, "instances": 20, "constraints": ], "acceptedResourceRoles": , "uris": , "env": { "SPARK_NO_DAEMONIZE":"true", "SPARK_SHUFFLE_OPTS" : "-Dspark.shuffle.cleaner.interval=31557600 -Dspark.shuffle.service.port=7337 -Dspark.shuffle.service.enabled=true -Dspark.shuffle.io.connectionTimeout=300s", "SPARK_DAEMON_MEMORY": "1g", "SPARK_IDENT_STRING": "7337", "SPARK_PID_DIR": "/var/run", "SPARK_LOG_DIR": "/var/log/taboola", "PATH": "/usr/bin:/bin" }, "portDefinitions": , "requirePorts": true } As mentioned before, we need to configure spark application appropriately: ## Spark application settings: - spark.shuffle.service.enabled = true - spark.dynamicAllocation.enabled = true - spark.dynamicAllocation.executorIdleTimeout = 120s - spark.dynamicAllocation.cachedExecutorIdleTimeout = 120s infinite by default and may prevent scaling down - it seems that broadcasted data falls into "cached" category so if you have broadcasts it might also prevent you from releasing resources - spark.shuffle.service.port = 7337 - spark.dynamicAllocation.minExecutors = 1 - the default is 0 - spark.scheduler.listenerbus.eventqueue.size = 500000 - for details see SPARK-21460 By now, we are running services with dynamic allocation enabled in production. For the first half of the day everything was great. After a while, however, we started to notice degradation in those services. Despite the fact that Mesos master was reporting available resources, the frameworks started to get less and less cpus from Mesos master. We enabled spark debug logs, investigated and found that frameworks that were using dynamic allocation, rejected resource "offers" from Mesos master. There were two reasons for this: - We were running spark executors that were binding to jmx port so while using dynamic allocation, the same framework in some cases got an additional offer from the same mesos-worker and tried to start the executor on it and failed (due to port collision) - Driver started to blacklist mesos-workers after only 2 such failures without any timeout of blacklisting. Since in dynamic allocation mode the executors are constantly started and turned off, those failures were more frequent and after 6 hours of the service running, approximately 1/3 of mesos-workers became blacklisted for the service. ## Blacklisting mesos-workers nodes - Spark has a blacklisting mechanism that is turned off by default. - Spark-Mesos integration has a custom blacklisting mechanism which is always on with max number of failures == 2. - We have implemented a custom patch, so that this blacklisting will expire after a configured timeout and so Mesos-worker node will return to the pool of valid nodes. - We've removed jmx configuration and all other port bindings from executors' configuration to reduce the number of failures. ## We still have to discover external shuffle service tuning Some params are only available with spark 2.3 or above : SPARK-20640 - spark.shuffle.io.serverThreads - spark.shuffle.io.backLog - spark.shuffle.service.index.cache.entries ## What we achieved - We are using dynamic allocation in production where it makes sense (e.g. services with some idle times) - We have better resources utilisation: instead of four services we are able to run five services on the same cluster - We were able to provide more cores to every service (800 vs 500) which reduced end-to-end running times. Notice how total_cpus_sum (the allocation from the cluster) follows real_cpus_sum (the actual usage of all workers for the framework) Overall we can say that: - Dynamic allocation is useful for better resource utilisation. - There are still some corner cases, especially on Mesos clusters. --- ### Stop waking up at night over MySQL replication URL: https://www.taboola.com/engineering/stop-waking-night-mysql-replication/ Last Modified: 2025-01-14 14:37:25 ## MySQL Slave Replication Optimization Written by Yossi Kalif & Ariel Pisetzky ## MySQL in Taboola So you love MySQL - what do you know, so do we here at Taboola. We spend a lot of our time with MySQL building our infrastructure to provide over 30 billion recommendations a day on over 3 billion web pages. In this blog post we would like to share how we optimized our MySQL to replicate faster over WAN connections so that we would not need to wake up at night and fix things. Oh, and it also helped us speed things up, so when we do have issues, they resolve faster. So, if your infrastructure has MySQL and you have replication, this blog post is for you.  TL;DR - at the bottom of the post Taboola operates in multiple data centers around the world, and at the time of the post, we have 9 data centers. We have our MySQLs in each of them and the Taboola network depends on the data in these MySQLs reaching all the data centers in a timely manner. Each data center (of the 9) has at least 10 MySQLs that are acting as slaves to our "Master" database (for the sake of simplicity, let's call it "Master-MySQL"). The daily replication volume to each of the globally distributed nodes is about 200Gb, while the local storage of one of our MySQLs is about 7TB. The use case for the MySQLs has evolved over the years. For the last 6 years the front end MySQLs are used to store configurations for the applications, publisher settings and campaign data, all of which change frequently. ## Problem statement deepdive So, what was the problem we tried to find a solution for within the world of replication? The Taboola services are sessionless, so each time a user connects to our service, the connection can (and will) go to a different app server. The app servers in turn will connect to different MySQLs slave for information on the transactions and this information needs to be replicated quickly to the data centers. When the Master-MySQL sends out updates, not all data centers have the same quality of WAN connections and the physical distance plays a role in the latency. When we set up the replication years ago, the changes were lighter and the size of the infrastructure was much smaller. The larger they both got, the larger the problem of replication interferences became. Worst of all, when something causes the replication to break (and you know this will happen at night) we need to fix it and wait for the replication to catch up. So we set out to (a) improve the replication resilience. (b) reduce the time it takes to replicate something and (c) improve the reliability of the replication, so we don't need to wake up at night and fix it. One important thing to note - this was all based on MySQL 5.5. ## First stab at it - 2015 Our first attempt at fixing the replication problem revolved around the slave replications settings, of retry and timeouts. Most of our issues were caused by packet loss or disconnections on the WAN lines (think VPNs, ISPs, routing and other networking issues Data Engineers don’t want to care about). With any issue on the long connections our replication suffered. We would see the replication slow down and in severe cases break. Once it broke, a manual restart of the replication slave was needed. Even a small degradation in the network service, such as a 2% packet loss caused the replication on multiple MySQLs to slow down and eventually break. (see graphs below). The first thing we changed was the default retry value, this would seem like a small thing, almost trivial, but network conditions do tend to change and that actually helped with the broken replication == manual restart condition. Now, the MySQL replication slave would retry more times and when the network recovered, so would the replication. There are actually two settings here “slave_net_timeout” and “MASTER_CONNECT_RETRY”. We wanted to reduce the amount of time the servers wait to declare the connection broken, then significantly increase the retries to establish a new connection  Network quality degradation causing increased replication delays: Almost full disconnection of network (high packet loss) causing a break in the replication: ## Second improvement - the network level Basking in our initial success, we let things be for some time and enjoyed the improved stability. During that well-earned quiet time, the size of our replication continued growing and our need to move data faster came to light. This made us look at the networking side of things and check how we could improve the network performance.  There are multiple ways to accelerate the networking output an application sees, most of them are hard to implement (move to UDP, compression, dedup and so on). When we came across BBR, it was clear it would be the simplest network tweak to implement and deploy in our network and servers. For further reading about Google BBR. We have implemented BBR on our binlog-servers, since the slaves of these machines are located on remote DC’s ( and affected by the WAN lines) , afterwards the MySQL replication was less sensitive to the events of packet loss. We have upgraded our linux kernel to 4.19.12 , and set the following kernel tunables: - net.core.default_qdisc = fq - net.ipv4.tcp_congestion_control = bbr ## Taking it to the next level The next phase in our quest to improve MySQL slave replication was aimed at improving the replication time. This new goal came out of necessity. The size of our data has been growing constantly so the following needed to be addressed: - Every break in the replication took longer to recover from. - Creating a new replication slave became a long process. Each frontend mysql has 5TB of data that needs to be restored from backup. - Maintenance of the replication slaves (patches, minor and major upgrades, hardware issues) became more expensive as the time to run them was compounded by the replication catchup time. - DML changes at the MySQL master cause mass changes at the slave and send a large chunk of data downstream from the master. Now at this point in our quest we were already using MySQL 5.7 and the replication slave clients were accessing the servers via ProxySQL. We use ProxySQL to manage read connections in such a way that the replication delay was a factor in the server selection process (the ProxySQL would send the query to the available servers taking into account replication delay). It was time to push 5.7 into the world of parallel replication. In order to run the test we had to build an identical environment for the test. We chose two nodes which have the same hardware & software, both worked with the same master MySQL. To complete the test environment and to remove external noise, we removed all the client connections from the two servers, so that the load on the servers would be identical. Once all traffic was removed from the servers, the replication was stopped for one hour, this way both servers have one hour of delay. When we turned the replication back on, we compared the two servers to see how this setting impacted the replication. The changes and test results are on the following table. ## Spec for test environment  Spec   Server 1 Server 2  cpu  Intel(R) Xeon(R) CPU E5-2630 v4 @ 2.20GHz  Intel(R) Xeon(R) CPU E5-2630 v4 @ 2.20GHz  cores  40 40  memory  128GB  128GB  Disk   Samsung NVMe   Samsung NVMe  masters  Master-4  Master-4  sync-binlog  2  2  innodb_flush_log_at_trx_commit  2  2  slave_parallel_type   database   database  slave_parallel_workers  0  0 ## Test scenarios # Test Results 1 Checked that our test servers are the same. Stopped the replication for one hour , and then let the replication lag finish. (running with slave_parallel_workers=0) Both servers finished the lag at the same time 2 slave_parallel_workers=4 slave_parallel_type=DATABASE(most of our data is in one database, so we didn’t expect any improvements here) Both servers finished the lag at the same time 3 In MySQL 5.7 a new replication parameter was announced slave_parallel_type slave_parallel_type=logical slave_parallel_workers=4 We observed little improvement in the parallel replication (but it was only about 10%) Not what we had hoped for 4 In MySQL 5.7.22 a new parameters was announced Binlog-transaction-dependency-tracking transaction_write_set_extraction Those parameters were applied on binlog-server/slave Parameters of Master-4 Slave_parallel_workers=8 slave_parallel_type=LOGICAL_CLOCK binlog_format='ROW' transaction_write_set_extraction='XXHASH64' binlog_transaction_dependency_tracking = WRITESET On the slave we used  Slave_parallel_workers=8 slave_parallel_type=LOGICAL_CLOCK  binlog_format='ROW' transaction_write_set_extraction='XXHASH64' binlog_transaction_dependency_tracking = WRITESET Replication delay catchup on Server2 finished in 17min compared to 50min on server1. 5 Since we saw an improvement when working with  Slave_parallel_workers=8, we tested it with 16 parallel workers. Going to 32 parallel workers didn’t give us additional improvement. Replication delay catchup on Server2 finished in 11 min, compared to 43min on Server1. ## Results Test 4 graph (you can see the improved time it takes to close the replication gap) The replication improvement might have cost on the CPU, so we also checked the CPU metrics to make sure that the impact on the server will not be too high and we could not run this configuration under production load. The CPU graph showed that the CPU increase for the replication work is about 7%. Test 5 replication catchup time showed the best result and CPU impact was within acceptable levels to go into production. The CPU consumption ## Bottom line Now with the improved configuration we can manage our 100 slaves with far more ease, as the cost of production operations is lower. The ability to perform patches, run hardware maintenance and overcome WAN issues is an important part of our MySQL routine.  TL;DR Here is a short version of all the steps we took in order to sleep better (not in the order of implementation) - Implement google BBR on binlog servers (blackhole engines) Kernel version 4.19.12 - net.core.default_qdisc = fq - net.ipv4.tcp_congestion_control = bbr - For parallel slave replication Upgrade master and slaves to 5.7.22 or higher - Change the following parameters on the master/slave slave_parallel_type=LOGICAL_CLOCK - binlog_format='ROW' - transaction_write_set_extraction='XXHASH64' - Slave_parallel_workers=8/16 - binlog_transaction_dependency_tracking = WRITESET - Configure network recover parameters: slave_net_timeout=60 - MASTER_CONNECT_RETRY=60 --- ### Fear of breaking production? Use Grafana! URL: https://www.taboola.com/engineering/fear-breaking-production-use-grafana/ Last Modified: 2025-01-14 14:37:25 In Taboola, we deal with scale, huge scale. A small issue might turn into a disaster in a matter of hours. Re-writing and replacing an existing service with a new one is a real challenge, moreover doing it without causing downtime is SCARY. Reading logs is not an option. Logs are gigantic, unwieldy and span over many machines. It would take hours to combine and analyze them. In this post I will share with you three graphs in Grafana that I think are a must for observing new code. Let’s start… ### Did I break production?  You write your shiny code, you (even) test it, but, how would you verify that you didn’t break the production environment? Luckily, we use Grafana, and this actually makes a big difference. My plan was to compare old code vs. new in Grafana, but, where to start? ### You have Grafana... let's use it! Frankly, I didn’t know anything about Grafana. Walking through the office, I saw on my colleagues screens , fancy, colorful graphs. Yet, I couldn’t understand a thing about them. When I looked for existing graphs to copy, I found way too many. So copy & paste was out of the question. Frustrating... I decided to learn Grafana from the ground up. It took me almost two days - I used Grafana Beginner's Guide on YouTube as a starting point. This really made a big difference. ### Which graphs are absolutely a must? Assuming you are using a JVM, you want to watch the garbage collector and heap usage, because the JVM’s memory is critical. If it is too high it will break everything. Must-Have Graph No. 1: Memory (Heap) Tip: Use a moving average function see the saw-tooth. Your graph will be less accurate; yet, it will give you a clearer view of the memory pattern. Even more importantly, it will give you better insights at a glance. Next, monitor the OS itself, track the CPU. As a rule of thumb, avoid reaching 80%-90% utilization. Remember that there are peaks, and your service needs to cope with them smoothly. Must-Have Graph No. 2: CPU Sidestory - this kind of graph helps me to convince the production IT guys to give me more computing power, as I could show them that I was hogging the CPU. So far these are the technical KPIs, now, what about applicative KPIs? ### Compare Apples to Apples - Keep the Same Metrics If you want to be sure that things still work, use the same KPIs in the new and old code. If your graph lines continue in the same trend - you can be sure it still works. Put the old and new metrics together in the same graph, because you want it to be easy to see any mis-alignment. In the following graph, the new metrics are positive and the old implementation is represented by negative numbers. This “mirror view” is easy on the eyes and helps spot variations. Must-Have Graph No. 3: Old & New Together  Tip: You can get this effect by multiplying the old metrics by -1. ### Things to remember - Take the time to learn Grafana - it’s absolutely critical for large scale applications - Monitor both OS and JVM, in addition to applicative KPIs - Stick to the old metrics - for the sake of feature parity and trend breaking - You can use moving average and scaling to -1 for easy comparison Although replacing a service without downtime is frightening, Grafana can really make a big difference. Grafana gives you the visibility needed to make sure your new code is running and the application is stable. --- ### Growing by Learning - DIY URL: https://www.taboola.com/engineering/growing-learning-diy/ Last Modified: 2025-01-14 14:37:25 To facilitate flexibility and technological hype, you want to work with people who know how to learn. This is much better than having someone who knows a specific programing language, because a person ‘ who knows how to learn’ can learn any new language! This agility is crucial, because technology is always changing and learning is endless: My story begins two and a half years ago in Taboola Engineering, where I arrived with dozens of new employees. In fact, 50 percent of the developers were new (less than one year)! Taboola was growing, and with great growth comes a great need to learn. My goal was to create learning programs, but along the way I realized that it was far beyond this - learning brings personal development, curiosity, doubt, and insights into the organization's working methods. The past years has been an exciting journey of many collaborations, trial and error, failures and successes. We learned a lot on the way, and today our training rooms are filled to capacity. You will even be able to find our VP R&D with an open notebook. In this blogpost I will share insights we have learned that can help you grow your environment (and make people crazy about learning). ## #1 - Expand the supply of knowledge Let's think about the knowledge market - the demand for knowledge is massive. We are all thirsty to expand our depth and areas of knowledge. To meet this demand, the trick is to create a wide pool of internal “suppliers” that can provide the knowledge. Meet your partners - engineers with one year’s experience Engineers with one year in the company are looking for self development and growth. They can be a great boost for the internal “suppliers”. This is a win win situation - let me explain why: This great demand does not fall on the shoulders of the veterans only. At the same time, it allows your engineers with one year’s experience, to deepen their knowledge, develop themselves and give additional value to the organization. ## #2 - Water the flowers you want to grow If you want to keep your pool of presenters motivated and passionate -  it is very important to recognize and support this group. The first step is to share those people’s wonderful contribution with the organization . As a result, they will be able to receive recognition not only from you but from their manager and colleagues. In the opening session of the learning program, I always show this slide: This is also an opportunity to create a circle of giving and receiving - invite the newcomers to join the contributors group and be on this slide as well. Secondly, support your content owners - - Assign a mentor who will help them professionally with the material - Create workshops about how to convey a clear message and upscale presentation skills This will help you improve the training and raise the motivation to belong to the group of presenters. ## #3 - Data Matters Data collection? Why should you invest in it? Analytics give you valuable insights. They help you to improve your learning programs. But even more importantly, they pinpoint weaknesses. So learning can be valuable for the organization and contribute to its growth. Let me give you an example - It popped up that people were not passionate to learn about certain technology, and we had to research for the reason why. It may be an old technology, there may not be many developers who code it, and anyway, this is an interesting insight for the organization. Most important - communicate your findings to others so you can turn insights into action. ## #4 - Build the learning brand Make your audience realize that they can earn something beyond their daily role - they have the opportunity to develop themselves. Let your audience feel that there is someone who is concerned about the accessibility of the right knowledge at the right time. These are the emotions that your learning brand should convey. One of the most exciting and fun outcomes will be that people and initiatives will reach out to you! You will feel like an excelerator for growth, entrepreneurship and innovation. But how can you create a brand? - Think about a logo related to learning and add it to all of your communications - Take advantage of opportunities to present learning activities - toast, happy hour, slack channel... - Show results - people talk and spread the word, so, one success leads to many more - Make the learning programs prestigious - if you want to both learn or teach: you have to invest. For example, set prerequisites for workshops ## Checklist - Engineers with one year’s experience are looking for places to grow, they can expand the supply to meet the demand for knowledge. - Recognizing and supporting the group of content owners is VERY important. This will drive motivation and energy for learning activities. - Track your data and leverage organizational projects through your insights. - Build the learning brand to create awareness of learning activities and to attract collaboration. ## Up next Today we are facing with challenges of an R&D that has grown greatly. In my upcoming blog post I will share from our experience - what you should keep in mind while you are growing.   --- ### 'Tis the Season: Fun with (Decision) Trees URL: https://www.taboola.com/engineering/tis-season-fun-decision-trees/ Last Modified: 2025-01-14 14:37:25 At Taboola, we work daily on improving our Deep-Learning-based content-recommendation model. We use it to suggest personalized news articles and ads to hundreds of millions users a day, so naturally we must stick to state-of-the-art deep learning modeling methods. But our job doesn’t end there - analyzing our results is a must too, and then we sometimes return to our data science roots and apply some very basic techniques.  Let’s lay such a  problem out. We are investigating a deep model that behaves rather strangely: it wins over our default model for what looks like a random group of advertisers, and loses for another group. This behavior is stable in the day to day, so it looks like there might be some inherent advertisers qualities (what we’ll call - campaign features) to blame for this.  You can see a typical model behavior for 4 campaigns below.  Daily results for selected campaigns, compared to a baseline of 100%. We can see the new model is very stable - good for some campaigns, and bad for others. How can we predict which campaign is going to succeeded? So we hypothesize that something about the campaign makes this model either work or not. Could it be the type of publishers the campaign uses? The size of their audience? Their cost per click? Maybe all of these things together? Santa is not here yet to help, so we should investigate this! If we thought there might be a single numerical feature responsible for this, we could write a simple script in python to look for correlation between the different campaign features and our label. But we think there’s more than one campaign feature to blame, plus we have many categorical features - so we can’t simply correlate. What to do? Let’s harness the power of classical machine learning! So, we’ll use a machine learning algorithm (specifically, a decision tree) to help explaining our deep-learning model results:  Some questions you might have before we go forward:  - Why use a tree, and not just use deep learning explainability tools like SHAP? We are trying to insulate the campaign features here, and the general model uses a lot of other features that are irrelevant (user features, context features) now. Also, we might have more meta-data to use than what goes into the deep model - say, the language of the campaign, how long it’s been running etc.  - Why use a tree, and not just another deep model to figure this out? We are looking for maximum explainability, even if our model isn’t 100% accurate - as long as it points to the main suspects that make us win/lose. - Why use a simple decision tree, and not random forests or xgboost? Just like the previous question. A decision tree is powerful enough to catch the main bad guys, it’s only a few lines of code and you can plot it!  - How do we choose the depth of the tree? Try to keep it simple, if you think you might have 1-3 leading features to find, a tree of depth 2 will suffice. You can look at depth 3 to see that you didn’t miss anything. Anything more is probably too much. So, to recap, we generate a table with all our campaign features and add a label column with 0/1, depending whether we do better or worse than our default model (This is what we are trying to predict). Use this table to fit a tree classifier. Then plot the tree and see what it found or just look at the features sorted by their importance. Results come after only ~15 lines of code and in less time it took you to read this post. This technique ended up being pretty useful: If there are indeed features that make a model work we can spot them almost instantly. Even more important, if the tree finds just noisy unexplainable columns - this means our hypothesis was wrong.  Some highlights before we paste the actual code below.  - Trees can work with categorical features, but you have to 1-hot encode them first. Don’t work hard and implement it yourself, but use the pandas built-in function below.  - Be careful with weird columns correlating with your label column! Remember explainability is key here - so if you suddenly find some random categorical value being very high on your decision tree, it's probably noise! remove it and start from the top.  - No need to split the data between train and test, using a low depth should take care of overfitting for us. That’s it. Sometimes simple solutions are sufficient even for baffling questions.  Trees are fast, explainable and their plots look really nice in presentations :) Merry Christmas and Happy Holidays everyone! As promised, sample python code here:  # Imports from sklearn.tree import DecisionTreeClassifier from sklearn import tree import pandas as pd  # Read the files data = pd.read_csv("your_file_here.csv") label = “your label column” # Split data from label y = data X = data.drop(label, axis=1) # Data preprocessing:  # Fill in NaN  X = X.fillna(0) # Turn categorical features to 1 hot encoded columns X = pd.get_dummies(X) # Fit tree  max_depth=2 classifier = DecisionTreeClassifier(random_state=0, max_depth=max_depth) classifier.fit(X, y) # Print top 5 features by importance print("max depth:", max_depth) print(sorted(zip(X.columns, classifier.feature_importances_), key=lambda x: x)) # Plot tree from sklearn.tree import export_graphviz import graphviz export_graphviz(classifier, out_file="mytree.dot",feature_names=X.columns) with open("mytree.dot") as f:     dot_graph = f.read() graphviz.Source(dot_graph) --- ### Monitoring and Metering at Scale URL: https://www.taboola.com/engineering/monitoring-and-metering-scale/ Last Modified: 2025-01-14 14:37:25 In this blogpost I will describe how we, at Taboola, changed our metrics infrastructure twice as a result of continuous scaling in metrics volume. In the past two years, we moved from supporting 20 million metrics/min with Graphite, to 80 million metrics/min using Metrictank, and finally to a framework that will enable us to grow to over 100 million metrics/min, with Prometheus and Thanos. ### The journey to scale begins Taboola is constantly growing. Our publishers and advertisers increase exponentially, thus our data increases, leading to a constant growth in metrics volume. We started with a basic metrics configuration of Graphite servers. We used a Graphite Reporter component to get a snapshot of metrics from MetricRegistry (a 3rd party collection of metrics belonging to dropwizard that we used) every minute, and sent them in batches to RabbitMq for the carbon-relays to consume. The carbons are part of Graphite’s backend, and are mainly responsible for writing the metrics quickly to the disk (in a hierarchical file system notation) using Whisper. Then, the Graphite web app read the metrics from Whisper, and made them available for Grafana to read and display. This is how our primary architecture looked: Figure 1: Graphite architecture When we wanted to scale horizontally - we added additional Carbon, Graphite-web and Whisper DB nodes. The cluster of carbons became big and complex. In addition, we used different kinds of carbons for caching and aggregations. We started to suffer from metrics data loss, when the carbon-cache nodes crashed due to out of memory, or when the disk space (used by Whisper) ran out. Moreover, the RabbitMq cluster wasn’t well monitored, and it was complicated to maintain. At this point, we supported 20 million metrics per minute, and we wanted more. The new requirement was: 100 million metrics per minute (!), and we knew the current architecture would not be able to scale enough to support this. ### First infrastructure change: Metrictank We chose Metrictank, because of the following reasons: - It is C* backend - which is known to be very stable and we were experienced with it, from other data paths. - It’s Multi-tenant - thus, we could share our content with it. - It receives various inputs (carbon/kafka, etc.). - It’s 100% open source. But mainly we chose Metrictank because it’s compatible with the hierarchical Graphite syntax - thus we didn’t have to refactor our current dashboards to work with the new key-value notation. The architecture at this time was composed of Taboola services, who reported their metrics using a new MetrictankReporter component. This reporter took a snapshot from MetricRegistry every minute, and pushed them, after batching, into a local kafka cluster. The metrics moved to the backend aggregative kafka cluster using a KFC-Mirror component (KFC = KaFka Consumer). There, they were consumed by the KFC-Metrics consumer, unbatched and compressed to a msgpack format, and finally pushed again to the same kafka with a different topic, for Metrictank to consume. In addition, we had external services (e.g., kafka, cassandra), who reported their metrics to carbon-relays that were used as kafka producers. They pushed their metrics into the local data center’s kafka as well. A KFC for external services consumed their metrics, batched them, and pushed them back to the local kafka, for the KFC-Mirror to transfer them to the backend cluster as well. There, they went through the same process of un-batching and compressing into msgpack. Last, one essential service for us (Sensu, used as an alerting system) worked only with RabbitMq, thus dedicated carbon-relays consumed its metrics from Rabbit and pushed them to the local kafka. This is how our Metrictank architecture looked: Figure 2: Metrictank architecture We started with an architecture of 9 Metrictank VMs - 24 cores each, and 10 nodes of Cassandra. Each node saved 1-2.5 terabytes of data to the disk. After a year, our disk usage incrementation looked like this: Figure 3: The increase in disk space usage in percentages of Taboola’s cassandra nodes (upper plot) and the disk usage in terabytes (lower plot) Metrictank introduced new challenges. First, it handled the downsampling and the query module in the same service. It had to flush the metrics twice a day to perform the necessary aggregations and write them to Cassandra. This caused a performance overhead, and daily downtimes in the query side. In addition, we suffered from kafka disk space issues, so we had to reduce their retention. Last but not least, our kafka consumers tried to overcome consuming lags, by pushing to Metrictank unproportional loads of metrics, which in return made it crash. In order to overcome the daily downtimes, we moved our Metrictank nodes to Kubernetes, and created 25 primary nodes who were responsible for writing the metrics to Cassandra, and 25 replica nodes who were available for queries from Grafana. We grew in the last year from 20 million metrics per minute to 80 million metrics per minute. Figure 4: Taboola's metrics amount in the Metrictank path This new reality made us constantly request for additional storage such as Cassandra nodes and larger disks for kafka. We still had to deal with Metrictank pods crashing from time to time after consuming huge unbalanced amounts of metrics. And finally, we received a new requirement: defining metrics per publisher. Taboola has over 10,000 publishers, which meant a lot of new metrics being added to the metrics flow. The current infrastructure could not be scaled enough to support the above, thus we decided to change it once again. ### Second infrastructure change: Prometheus Why choose Prometheus? - It is simple to install & scale. - It has a federation ability - every Prometheus service can turn itself to an endpoint for another Prometheus service to consume the metrics from. - It has a built-in alert manager. - It is widely used and maintained. - It is 100% open source. The major difference between our Metrictank architecture and Prometheus was that Prometheus scraped metrics from dedicated endpoints, while Metrictank consumed the metrics from kafka. Thus, our implementation had to be changed. In addition, there was one major drawback: we had to refactor all our existing dashboards to key-value notation. So, we wrote an automatic tool for converting complete Grafana dashboards from hierarchical graphite syntax to a key-value notation. Similarly to the MetrictankReporter, we wrote a new TaboolaExporter. The exporter took a snapshot of metrics every minute and exposed them in the service’s endpoint. The Prometheus service in each data center scraped the metrics and stored them internally. For external services such as Kafka and Cassandra we used additional exporters (mainly JMX exporters) to expose the metrics to dedicated endpoints as well. The Prometheus service used Consul’s discovery abilities in order to know which servers to scrape the metrics from. Finally, we used Prometheus’ built-in Alert Manager for reporting local data center alerts. This was our first vanilla deployment of prometheus in each of our data centers: Figure 4: Taboola's metrics amount in the Metrictank path The above deployment worked like a charm, but we still had to answer some hard questions: - How can we store terabytes of data in a reliable and cost-efficient way, without sacrificing our queries response times? - Can we execute an aggregative query over all data centers, from different Prometheus services? For this purpose, Thanos came into play. Thanos is a collection of services created specifically for using Prometheus at high scale and across data centers. We used Thanos as follows: - Thanos query - handled our cross data center queries, using client-side join of the data from each Prometheus db. - Thanos ruler - recorded our heavy / frequently used queries ahead, and the results were available for the Thanos query service as a time series data to scrape. - Thanos sidecar - used as a mediator for our long term storage that was defined for each prometheus server (e.g. Google storage). Adding Thanos to the former deployment looked as follows: Figure 6: Adding Thanos for cross data center queries After adding Thanos, we met the first requirement we had - to support +100 Million metrics per minute. ### Finally, a success story After deploying Prometheus and Thanos framework, we reduced the amount of metrics sent through the Metrictank path by almost half, and this flow is still functioning and stable on Kubernetes, with close to zero downtimes. Currently, we have 48 million metrics/min (800,000 metrics/sec) in the Metrictank path and 73.2 million metrics/min in the Prometheus & Thanos Path (1.2 million metrics/sec, and keeps growing). We changed our metrics infrastructure twice in two years as a result of continuous growth in our metrics volume. We faced complex implementations and gradual deployments along the way. Remember - major changes do happen frequently in dynamic companies, and with a strong team they can lead your company to great success. Regarding our metrics per publisher requirements - it was simple to implement after the last infrastructure change. If you are interested in how we did it, you are welcome to check out the Taboola Engineering blog, and search for our next blogpost on “Publisher Metrics and much more @Taboola”, soon to be published. I want to thank the best Infrastructure Engineering group in Taboola who made all of this possible. Special thanks to Moty Lavi and Tidhar Klein Orbach for helping along the way. --- ### A QA Party is the BEST Party in town! URL: https://www.taboola.com/engineering/qa-party-best-party-town/ Last Modified: 2025-01-14 14:37:26 Sometimes we need to test urgent features fast. It has to be within a very short timeframe, when there is not enough time to run a full test plan for that feature. This might occur on different occasions. When not having enough manpower in QA to cover a full test plan for a feature. New special demands from an important client right before the release deadline. Product management needs new adjustments before the developer deploying a new product version. It can also happen when a client, team lead or PM wants a new feature and it should have been done YESTERDAY! It can also happen actively. Running every once in a while a wide post-production test, or dedicating limited time for a bug hunt. We at the Taboola Video Solution department call it “Search for a Bug Thursday”. This unplanned development might end up launching a “half baked” product. It will produce bugs in production that could be very expensive to fix so late. QA Parties are epic! When we need to cover more tests in less time, we hold a QA party. In this type of parties, members from different teams join in on the QA testing effort. This enables new untrained eyes to test QA scenarios. With a fresh look, they can find overlooked bugs which “tired” users might take for granted and ignore. Some places use what is called a Beer Driven Testing session. Having people drinking alcohol is both fun and productive. It is a time-saving session covering a wide range of tests. But always remember - don’t drink and debug on production! Some drunk scenarios might seem illogical at first. Some real users might not think so. Especially when believing in the practice that the user is always right! Better face it in-house first than paying for it later. When organizing a QA party, sending an email is not enough. Go over in person and ask the people you want for their help. Talk to them. Explain why and how you need them. Creating a buzz and a common interest to deliver the product tested and ready fresh out of the oven. QA Party themes There are different kinds of QA parties: You can host a full blown out QA team party. People from all different teams, who don’t usually work together, join the entire QA team. When the lone QA team is not enough, you can organize a wide QA-product party. Everybody who works on the same product joins the fun - both QA and developers. If more effort is required in a short amount of time, plan a full QA-R&D party. Gather all R&D teams and QA for a short time to test the one product. If the product’s scope is very wide, everybody is welcome to a full QA-company party. Send all the troops from the entire company to the war zone. Even reserves, such as finance and HR, are invited. Taboola’s QA party - where the fun never ends There were several occasions where we at the Taboola Video solution had to throw a QA party. Without enough time to come up with a proper test plan, we enlisted the help of several R&D teams. We asked them to verify features together to speed up the process and release on time. We found several critical showstopper bugs this way. Without the party, QA probably won’t have the resources to find these showstoppers. The party was a huge success, and we decided to make a monthly event out of it. Each month all the video teams dedicate a few hours to debug their code. We call it “Search for a Bug Thursday”. We managed to explore and catch some very interesting bugs that way. Either in production, or on features currently in development. If someone in R&D had a lead, or a suspicion for a potential buggy behaviour, they used this time to look into it. While having our periodic QA parties we managed to discover several issues this way. We caught a frozen video from a 3rd party script. We found out that we log the same messages twice in the debugging console. In one of our QA parties we tested a feature which adds text info to our video players. The feature worked well in dev environment. At the party, we tested the feature in the staging environment. We caught that Unicode characters are being displayed as gibberish. We prevented this bug from being deployed to production. Without the party, we probably would have missed this bug. All such issues were reported and addressed appropriately. Without QA parties, we might have missed these issues. Even worse, we might have found out about them only when severity rises due to users' complaints. QA party upgrades Using a QA Party as a method to increase testing wideness can be handled wisely. Divide people into groups and assign them different tasks. Assign a group for each environment, platform, OS and web browser you need to test. There is no “phone basket” in a QA party. If needed, ask people to use their personal phones. Non QA personnel should be asked to look into happy path scenarios. This will free QA's time to focus on more in-depth specific scenarios. In this type of party, QA works hard and plays the hardest! Invite everybody - entrance is allowed for developers, administration, finance, business and more. This will get more eyes looking at the product, and each one will look from his/her perspective. This practice may lead users seeing more edge case bugs, that otherwise might be missed. This also lets the team go through a wider range of unique and different user styles of the product. Maximizing testing scope while minimizing time spent actually leads to bulk testing. Take a big group of testers from different backgrounds. Have them dedicate time to test a single product. They will cover more scenarios than the lone tester can. Each of the party people can contribute valuable feedback on the product from their perspective. This can get product managers to tweak and change the interface accordingly. Another effect of a QA Party is exposing features to other members of the company. They might not have known or experienced these features otherwise. It will make them more familiar with new aspects of the company. It will help them know their own business better! Saving time, saving costs The QA Party is an elegant solution for quick and effective testing. It saves time. It encourages a wide-spread teamwork. It helps knowledge sharing between teams. Most of all - it is the best party in town! Cheers and QA on! --- ### Performing Exploration, Robin-Hood Style URL: https://www.taboola.com/engineering/performing-exploration-robin-hood-style/ Last Modified: 2025-01-14 14:37:26 Our core business at Taboola is to provide the surfers-of-the-web with personalized content recommendations wherever they might surf. We do so using state of the art Deep Learning methods, which learn what to display to each user from our growing pool of articles and advertisements. But as we challenge ourselves manifesting better models and better predictions, we also find ourselves constantly facing another issue - how do we not listen to our models. Or in other words: how do we explore better? As I’ve just mentioned, our pool of articles is growing, meaning more and more items are added each minute – and from an AI perspective, this is a major issue we must tackle, because by the time we finish training a new model and push it to production, it will already have to deal with items that never existed in its training data. In a previous post, I’ve discussed how we use weighted sampling to allow more exploration of items with low CTR (Click-Through Rate) while attempting not to harm the traffic of high CTR items. In this post, I’ll extend this dilemma even further, and discuss how we can allow meaningful exploration of items which our models have never seen before. ## Understanding Embeddings and OOV Understanding the use of embeddings – and what they represent – is the key to understanding the discussed exploration method, so let’s briefly review how they work. An embedding is a set of weights (or more simply, a vector) learned by the model to represent the characteristics of an input feature. Let’s say that we have two items in our pool, where item A is super-successful among gamers of age 20 to 30 living in the US and Canada, and item B is very popular among cooking-lovers of ages 30 to 40 living in German-speaking countries in Europe. Where should the model store this learned information? In each item’s embeddings, of course! So embeddings are value-specific vectors the model has learned, and it allows it to distinguish one from another, and estimate their performance more accurately. For each of our items, the model learns a new vector. If our model is training correctly, another item which is successful with item A’s crowd will have an embedding vector which will be close to him in our vector space. But let’s say that our training data has another item, item C, which is so new it appeared on the training data only once. Is this single appearance sufficient for the model to learn anything about this item? Of course not. So there’s no point in creating an embedding for it – as it will obviously just be noise. Yet, we still need an embedding for it, so what should we do? Whenever we build embeddings to a feature, we decide on a threshold of minimum appearances each value must have in order to have its own embedding. All the values which have less appearances than the required threshold are then aggregated and are used to construct a single embedding which will be used for all of them. This embedding is named the Out Of Vocabulary embedding (OOV), as it is used for all the values that didn’t make it into the list (or vocabulary) of values which have their own embedding. Even more, the OOV embedding is also used for all the completely new items which the model never seen before. It’s easy to see why having an OOV embedding is really bad for any value. It is constructed of many many completely different values, which rarely have anything to do with one another. This means that OOVs are usually noisy and have very low CTR estimates – even though there’s nothing wrong with these items, they simply weren’t displayed enough when the model was trained. But there’s a vicious cycle here – if a new item received an OOV embedding due to lack of appearances, it will receive very low CTR estimates by the model – which again won’t allow it to receive many appearances. This can make the model stick to the same items it always displays, and this isn’t good for us. So... it’s time for us to head over to Sherwood forest! ## The Robin Hood Exploration One of our most important goals at Taboola is to predict whether our users will click on an item we’ve shown them or not. Therefore, our training data is constructed of all the items we’ve shown (also known as impressions), and the user’s response (clicked or not). As there’s an obvious positive correlation between the number of times we show a certain item to the number of clicks it receives, it’s easy to see that our best-selling items are also the ones with the most appearances in our training data. This means these are also the items with the most powerful and well-trained embeddings. So the trick we did here is quite simple – just like there’s a minimal threshold of appearances to the embeddings, which below it a value becomes an OOV, we’ve added a maximal threshold to the appearances, which above it a value also becomes an OOV. We tuned it so only the very top of items will fall above this threshold and will be added to the OOV bucket. This provides us with an OOV made by a combination of the most profitable items we have – so now all new items receive a super-powerful embedding, which makes the model give them higher CTR estimates. We named this method Robin Hood Exploration, as we “steal” data from our best items and move it to our OOVs. And, yes – it works: Impressions (yellow) and revenue (green) by different embedding groups of the Robin Hood method compared to our default model. The Robin Hood method is performing as expected – our OOVs and Weak embedding groups receive more traffic and also yield more revenue, but on the expense of our top-sellers. Medium-strength embeddings are left with almost no change. Reducing traffic of top sellers is something we can deal with when affecting a limited amount of traffic – our purpose here is to perform exploration, a.k.a: give traffic to new items, and there’s no need to perform exploration over our best sellers - we already know they’re our best-sellers. And there is even more good news: when it comes to our overall revenue - this method showed absolutely no negative impact over it. To conclude, our Robin-Hood approach, which transfers data from top items to OOVs, has proved itself to be a valuable method for performing meaningful exploration with minimal technical effort. We hope it will prove itself worthy for our readers too. Safe ride! --- ### Make React Components Consumable Again URL: https://www.taboola.com/engineering/make-react-components-consumable/ Last Modified: 2025-01-14 14:37:26 The integration between a newly developed code into an existing code is always a challenge. Recently I was assigned a task to develop a new app inside Taboola’s Backstage. As a developer who is always looking to learn new technologies, I decided to develop it using React, Redux and Middleware. I have never developed using Redux and Middleware, only old-school React. So the first thing I had to do was to learn how to use these funky libraries. With the help of my wonderful teammates, who had some experience with these libraries, I had an easy start. I was able to create my project and install Redux. Then I became acquainted with the single source of truth concept and how to create actions. Challenges ahead Backstage was not developed using React, which raised two concerns. First, I did not want to break the existing code. Second, how to integrate between my React app and the parent site. Thanks to the React build command, the first concern was solved pretty easily. React build command creates a bundle code, based on defaults configuration. I injected the bundle code into an iframe and I was able to isolate my app’s code. Errors were encapsulated into the iframe scope, and the parent site didn’t break if an error was thrown. But now how can I export data from my app to the parent site? The parent site - iframe communication dilemma My first attempt to interact with Backstage was by using an unfriendly iframe. The way to communicate between parent site and an unfriendly iframe is by using events and postMessages. The disadvantage of using this approach is that the communication becomes asynchronous. Why is that bad? Unnecessary async communication in JS causing CPU and time overhead. Obviously, I didn’t want to lag Backstage. It is also harder to debug and read. I wanted the next developer who works on this project to have a smooth sailing into the code. Make friends with your iframe I still wanted to encapsulate my code inside an iframe, but I had to “make peace” with my iframe before. By replacing the iframe with a friendly one, I still got all the benefits of error handling I mentioned. It also created a new opportunity for me - I was able to expose an API to the main page. Why was that an advantage? It enabled me to synchronize the communication between the main page and the React app. As mentioned before, it makes the work much easier, for both the browser and future developers. Wrap it up I decided on writing the API as part of the main React component. In this case, I had to expose the main component to the main page. In create-react-app’s default code, the entry point to the code is not returning any value. The running of the entry point is independant. Also, ReactDom.render is not returning it’s result. It only renders the basic React code into an existing element. The first thing I had to do is to change this behavior - I wanted to be able to get the result of ReactDOM.render. Sounds simple, doesn't it? I found a helpful GitHub project to start my vision. With the help of this project, I was able to understand how to wrap my React code and expose it to the main page. Make your React code consumable Let’s take this structure for example: First of all, we want the iframe to consume our React app and store it as a variable. When using create-react-app, React provides a native build command - “react-scripts build”. It creates a directory with a bundled version of our code, ready for production. We need to change it a little bit, by creating our own Webpack. Our new webpack would have to return our entry point and make it assignable into a variable. So, we had to change only the output config for our build. All other configs can stay as default. I do suggest adding the entry point, for readability reasons. The option that configures how the library will be exposed is libraryTarget. The safest and most recommended value for libraryTarget is “var”. By using this option, the value returned from Index.js will be assigned into a variable in the iframe. Then, we will be able to use its returned value in the HTML code: Expose your API If I develop the API in Main component, I can return the component from Index.js and call the API from the main page. For example, I will be able call Main’s “changeText” method after 5 seconds from the main page: Easy as a pie. But a short reading in ReactDOM documentation revealed the next note: The solution that is offered in the note is callback ref. In this case, callback ref could not help me. It will allow me to create a reference of Main component and use it as a member of App component. However, I will not be able to simply return the referenced component to the main page. Overcoming the async barrier Besides React’s documentation advice, I faced the same issue when I added Redux into the mix. Redux’s Provider component does not return any value, because it loads asynchronously. How can we overcome this barrier? With Promise.  We can pass the resolve callback to Main component. When the component is mounted, we can resolve the promise. We can pass the component as the resolve value, and store it in the iframe. Then we will be able to use it from the main page: Let’s Wrap it up (yes, I used the same pun again) Now we can add to our API, consume it from the main page and use it as we please. We made our code simpler for the next developer and debugging became easier. We removed redundant CPU usage and run-time lags by avoiding unnecessary asynchronous code. This method is very useful when writing a React app. You can export data to any other app, regardless of which library is being used. By doing some small changes, we can do great things. Using React app, we can integrate a new React app with any Angular, Vue, jQuery and plain javascript apps. Sounds freakin’ awesome! A little anecdote for the end: I was told about this solution from a friend, who used it for a Redux-less React app. Also, the example from GitHub is not directed for Redux projects. I didn’t think it would require any special adaptation. I was wrong. At the beginning, a null object was returned from my Index.js file to the main page. I learned the hard way that Provider is created asynchronously. For an entire day I banged my head against my laptop. But once I understood how Provider works, I was able to synchronize my code using promises. Only afterwards I read the note about ReactDOM.render. Hopefully, this will save other developers precious time. --- ### To Predict Article Performance, Use Machine Learning Models URL: https://www.taboola.com/engineering/understanding-articles-using-machine-learning-models/ Last Modified: 2025-01-14 14:37:27 ### Introduction Newsrooms are under constant pressure to deliver the most up to date, relevant, and engaging information possible. At Taboola, we are building tools to make this faster, easier, and now--predictable. As soon as an article is published the team has a critical eye on engagement data. Garnering insight on article performance as soon as possible is critical for guiding content strategy. Some articles receive wide attention immediately, drawing hundreds of thousands of page views within minutes, others may only see the first page view after a few hours. Taboola aims to narrow this gap even further by leveraging Machine Learning Models to predict article performance the moment after it becomes available to the reader. Read on for details on our latest research and fascinating discoveries around predicting article performance! ### ### Article Data Taboola Newsroom is a real-time optimization technology that empowers editorial teams with actionable data around what stories, headlines, thumbnails and placements generate the maximum engagement across desktop and mobile. When enabled, it constantly processes all the traffic and engagement data from the publisher. For example, all article page view traffic is grouped by source, location, and platform. For the models below, a one-minute time interval is used to understand the article data. Because different publishers may have distinctively different user engagement patterns, our preliminary research models focus on one US main-stream publisher in 2017. Using articles published in a one-week span, traffic from the first two hours post-publish is shown in the following figure. The majority of articles are published during working hours and popular articles usually see substantial traffic immediately. All articles exhibit different performance trends, but among those differences there is no clear pattern that can be determined. Though other article-related information may have an impact on performance (such as author, article category, time of publish), that data is not used in this chart. ### Models We used multiple machine learning algorithms to analyze the news articles, three of which have the results explained below. To maximize the applied value of the models, traffic data is limited to the first 30 minutes post-publish. a. PageView Model Total number of page views from individual articles is an important measure for publishers; by knowing the projection of pageviews, it could help them to promote potential articles as early as possible and replace less-engaged articles with better ones. The target was defined as total page view traffic of each article in the first 24 hours post-publish. Several feature engineering efforts were applied: - Because the minute-level traffic is sparse for low-traffic articles, time-windows of different length were used. - Categorical features such as author, news section, and location were encoded and converted to numerical features. - Low-traffic geographic regions were grouped together. A gradient-boosting-tree based model was built to predict targeted pageviews. The pageview error ratio, pageview error divided by total pageviews, is shown as a function of pageview on the left. The prediction can vary from actual pageviews due to changes made to articles after the first 30 minutes, such as sharing on Facebook. These actions will change the total traffic drastically but are not considered in this modeling. In all, this preliminary model has about 35% error rate. b. Popularity Model In this model, rather than directly predicting specific page views, a classification model was considered. Some articles demonstrate great potential in drawing attention early; being able to identify and promote them in the early stage will give the publisher an advantage over their competitors. We set a pageview threshold to label articles with potential for gaining mass popularity. A model was built to identify these potential articles. The red-dash line in the histogram below represents the threshold. After building the model, an arbitrary threshold was used to separate articles. This threshold provides a trade-off between precision and recall as shown in the right-side figure above. In general, this model can capture more than half of the popular articles at 85% precision level. ### Deep Learning Model Because the traffic data is connected in sequence and organized in the same structure recurrent neural network (RNN) is a good candidate for modeling structure. A hybrid model was built in TensorFlow to predict the total traffic in a 24-hour window; the corresponding graph is shown. This hybrid model contains two parts: an RNN and a fully-connected-layers (FC-layers). The output from the first RNN was concatenated with categorical features and fed into the second FC-layers. An Adam optimizer with decaying learning rate was used to train this regression model. Without any further hyper-parameter tuning, this model has a similar prediction accuracy (root-mean-square-error) to the PageView model, but it has the highest potential. ### Summary The application of machine learning models has shown great potential in understanding the traffic trend of news articles. With as little as traffic within a few minutes, models can identify the popular articles with a good balance between precision and recall. The preliminary deep learning model such as RNN also demonstrates similar performance to the traditional models (with careful tuning). The deep learning model should be expected to exceed with fine tuning in the long term. Taboola will use this experiment to continue building tools that support Editors and Authors in Newsrooms as they refine and adjust their content strategy. --- ### Exploiting Multi-Categorical Features Using Deep Interest URL: https://www.taboola.com/engineering/exploiting-multi-categorical-features-using-deep-interest/ Last Modified: 2025-01-14 14:37:27 At Taboola, our goal is to predict whether users will click on the ads we present to them. Our models use all kinds of features, yet the most interesting ones tend to be related to the users’ history. Understanding how to use these features well can have a huge impact on the model’s personalization capabilities, due to the user-specific knowledge they hold. User history features vary strongly between different users; for example, one popular feature is user categories - the topics a user had previously read. An example for such a list might look like this - {“sports”, “business”, “news”}. Each value in these lists is categorical and they have multiple entries, so we name them Multi-Categorical features. Multi-Categorical lists can have any number of values per user - which means our model must handle both very long lists and completely empty lists (for new users). Supplying inputs of unknown length to machine learning models is an issue which needs to be addressed wisely. This post will walk you through how to integrate these varied length features in your neural network, from the most basic approach all the way to "Deep Interest", the current state of the art. ## Being average Let’s continue with user categories. How can we turn a list of unknown length to a fixed-length feature which we can use? The most naive, yet very popular approach, is to simply average. Let’s assume we have a separate embedding of length d for each category - so that each value of the list is mapped to an Rd vector. We can use average pooling - average all of the user’s categories and receive a vector of length d every time. More formally, if emb(wj) is the embedding of the jth word of overall N words, then a user’s average pooling can be written as: But this is far from perfect. Averaging might work well, but when the list is very long the 1/N factor creates a small final result; summing different pointing vectors may cause them to cancel each other out, and division by N makes this substantially worse (If you come from a physics/EE background this should ring a bell, since this is much like what happens in non-coherent integration). This is bad, since it hurts users with longer history vectors, which are the ones we have the most information about and can give better predictions for. Because of this, we (and the folks from Alibaba research) remove the 1/N factor and use sum-pooling on the user-histories, to ensure predictable customers don’t have negligible impact on the network. ## Weight-lifting, naively Averaging automatically assigns the same weight to all elements. Is this what we want? Let’s consider a user with this list: {‘football’, ‘tennis’, ‘basketball’, ‘fashion’}. Averaging them naively will put an emphasis on the sporting categories, and neglect “fashion”. For some, this would be accurate - people who are more interested in sports than in fashion; for others, it might not. What could we do to give the different categories different weights to better depict users? A better solution will be to use some other information, for example: the number of past views (impressions) the user had for each category. Since the number of times a user read an article correlates with his personal taste, we can use these counts as weights and do weighted average pooling. This will help us differentiate, for example, between the following two very different users:  User 1 = {‘football’: 2, ‘tennis’:2, ‘basketball’:2, ‘fashion’:200} User 2 = {‘football’: 2, ‘tennis’:2, ‘basketball’:2, ‘fashion’:2} ## Are you paying attention? Can we stack even more information? Say we not only have the number of each user’s views per category, but also things like the last time the user read an article of this category, or this category CTR (click-through-rate) per user (meaning, how many times the user read an article from this category out of all the times this category was presented). What would be the best way to combine these features without endless algorithm A/B testing? The answer, of course, is to harness the power of machine learning, and apply an additional layer whose job is to output a weight for each value. This is sometimes called an “attention layer”. This layer has all the relevant data available as input, and learns the ideal weights to average the values. Let’s look at an attention subnet example for a single category value - “tennis”. Assuming the embedding for “tennis” is {5,0.3,5,1,4.2}, and the user was recommended 10 tennis articles but only clicked once, the flow can be illustrated as: The attention layer calculates the weight for each category embedding. In the above example the calculated weight is 0.2. Say this user was also interested in “fashion” and the attention layer output the 0.5 for this category, the final calculation for this user’s history embedding would be 0.2*emb(tennis) + 0.5*emb(fashion). These attention layers learn how to utilize all the information we have about this user’s history for calculating weights for each category.  Great! But can we make this even better? ## Introducing: Deep Interest architecture Remember, we are trying to predict whether a certain user might click on a certain ad, so why not plugging-in some information about the current suggested ad? Say we have a user whose history is {“tennis-equipment”, “kids-clothes”}, and our current ad is for a tennis racket. Surely we would want more weight on the tennis-equipment and neglect the kids part for this specific case. Our previous attention layer is only designed for modeling the user’s interests, and does not consider interactions between the user’s personal taste with the ad suggested to him during the pooling stage. We would like to intertwine this user-ad interaction before the pooling occurs in order to calculate a weight which is more specific to the current ad presented to the user. For this purpose we can use the Deep Interest architecture, a work that was presented at the KDD2018 conference by a group from the Alibaba research. If we go back to our previous illustration, we can now add an embedding for the specific ad suggested to the user: With Deep Interest, the ad features are used as inputs for the neural network twice: once as an input for the model itself (contributing information for whether or not presenting this ad to the user will lead to a click), and a second time as an input for the user’s history embedding layer (tuning the user’s history attention to the parts relevant for this specific ad). We can also have different ways to create embeddings for the ad-features: one is to use the same embedding vector for both cases mentioned above, and another option is to generate two different embeddings for the two cases. The first option gives the embeddings twice as much gradients while training, and might converge better. In the latter, we can have two different embeddings (say embedding of size 12 for the full network, and embedding of size 3 for the user history weight layer). A shorter embedding could save run-time serving but adds more parameters to the learning process, and might lengthen training time. Choosing between the two options is of course dependent on the specific use case. ## Summary Multi-categorical features are crucial for predictions in various common problems, and particularly when trying to predict future user behavior (like clicks, conversions) from past interactions. There are simple ways to do so - but for better predictions we recommend the Deep Interest architecture, as it allows dynamic weighting of the feature vector according to the relevant context. This architecture is quite simple to implement and gives great results for many recommender systems. Go give it a shot! --- ### Taboola Hackathon 2019 URL: https://www.taboola.com/engineering/taboola-hackathon-2019/ Last Modified: 2025-01-14 14:37:27 ### My First Time Running a Hackathon I’ve planned events before - but never anything like this 48-hour marathon spanning 200 participants across 3 continents and 2 time zones! When I was given the opportunity I was extremely excited, but at the same time somewhat anxious. Would I succeed in matching everyone’s energy and meeting their expectations? The task was daunting, but I took it slow and steady, step by step. First, I designed (with help from our talented graphic designer) a cool and eye-catching theme that would decorate all of our hackathon materials. Decorating slide decks, headers, banners, and t-shirts, our hackathon branding quickly became an R&D favorite. With the event only a month away, we held the official Taboola R&D Hackathon kickoff. Immediately, all of our participants - from engineers to designers to managers - began to gather in groups and cultivate project ideas. To encourage creativity and tech innovation, we even held an intro workshop, “The World of Making”, where our hackers learned about smart homes, automation, robots, and much more. One of the most important values here at Taboola is collaboration and in company spirit, three of our developers from Tel Aviv found themselves flying to Los Angeles only a few days before the hackathon. Jet lag didn’t seem to be a problem, who ends up sleeping at a hackathon anyway? ### Ready, Steady, Go! The stage was set for the event of the year and Adam Singolda, CEO and Founder of Taboola, rang the bell. 43 projects showed up at the starting line and with only 48 hours, the teams had no time to spare if they were going to realize their ideas and get them working and ready for the semi-finals demo presentation. Meanwhile, every few hours, we made sure that the participants had plenty of sugar to keep them going :) For those looking to take a break and rejuvenate, a challenging 30-minute Escape Room (made and hosted by enthusiastic Taboolars) awaited them. By 5:00 AM, our participants had demonstrated a clear ability to find new and interesting places to sleep all around the office. ### The Semi-Finals Our Semi-finals took place in 3 countries - one in LA, and 6 in Israel and The Ukraine. We carefully selected judges from within Taboola, chosen because of their broad vision, passion and love of innovation. They spent a lot of time listening to the teams, deliberating amongst themselves until finally deciding who would qualify to make it to the Finals. ### The Winners’ Podium The 2019 Hackathon winners was a group with developers from both our Tel Aviv and LA offices! Their solution offered SMB advertisers a solution to help optimize their campaign based on similarity to popular campaigns amongst various groups of users. The 2nd place was won by the Taboola Beacon team. Taboola Beacon created customized recommendations for users based on very specific location using Beacon-based targeting in a Taboola mobile app. Last but not least, our runners up in 3rd place - the Taboola Police team! Taboola Police decided to help our Taboola kitchen staff and added a facial recognition component on the kitchen sink, that makes me frightened every time I put dirty dishes in the sink instead of stacking the dishwasher ? ### The Hackathon Magic Two weeks of sleep and sugar-balancing later, I am finally starting to understand what we like to call the “Hackathon Magic.” As builders, the absolute best way to grow our skills is to innovate new ideas and create new things and when we do this alongside others, we learn from them and they learn from us. Most importantly, hacking helps us make lasting connections with others, and that is the Hackathon Magic! Now it’s time to plan for Taboola Hackathon 2020! Check out our 2019 Hackathon Highlights in this video: https://www.youtube.com/watch?v=CnvEFEuBYLY&t=5s --- ### Mentorship made easy URL: https://www.taboola.com/engineering/mentorship-made-easy/ Last Modified: 2025-01-14 14:37:27 I have not always worked at large scale companies such as Taboola. I have started my career in a small startup, where I was a full stack developer in a student role. As a first time developer, it was very important that I will be teamed-up with someone more experienced to learn from. Lucky for me, I was the only employee, working under the two founders - the CEO and CTO. Having a lot of one-on-one time with the CTO, working closely on bugs and features and discussing ideas, helped me improve my skills as a developer. Today, when I have more experience, I know how important it is to influence others who work with you. I try to spread the knowledge I have earned in the last few years, and help new and old employees as much as I can. In the day-to-day job, you mostly grow your skills as a developer. Since the beginning of this year (2019), I have tried to also grow my managerial skills. I have taken a mentorship role, where I try to help our QA automation engineer improve his abilities. Is mentorship necessary? Yes, it is. I believe that the best way to influence people is by setting an example. Of course you can improve your mentees or employees in a non-personal way. You can direct them by sending them materials to read or discuss with them only in group meetings. I think that a personal touch is more influential. Having one-on-one personal time or doing some pair programming is much more efficient. People skills 101 The first thing a mentor should have is a good relationship with the mentee. You don’t have to be best friends, but you have to be able to discuss ideas and respect each other. Show your mentee that their opinion matters. Have an open discussion about future goals, project design and code skills. When I worked on our automation infrastructure, I would ask my mentee to review my code. I found it is a useful way to improve his code skills, by reading the work of a more experienced developer. It also improves his confidence, by letting him know that I appreciate his opinion. Communication breakdown is always the same In every relationship, good communication is the key. It takes time to build good communication. In order to work on it, we have scheduled weekly one-on-one meetings. Why is it important? Weekly one-on-ones not only meant for tasks update. When you have one-on-one time, without the entire team around, it is easier to speak freely. Provide a safe environment and help your mentee understand that he can say whatever he wants, and it will stay in the room. It may take several meetings, but over time you will gain each other’s trust. A tip I found to be useful is to just shut up. You don’t always have to talk, sometimes it is even better to stay quiet. When you ask a question, someone will have to answer it eventually. Let it be the mentee. It’s ok to shut up and let the mentee answer on their own time. Don’t be afraid of the silence. The feedback loop It is not easy to give feedback for the first time, but it is the best way to improve your skills. It’s very important to give an honest review. Sometimes we are afraid to criticize the work of our colleagues, because we don’t want to hurt their feelings. However, we have to provide full feedback if we want our mentee to improve. Criticize when needed, but remember to justify your comments. The main point of feedback is learning and expanding the knowledge. Every comment should be fully explained. If you ask for a code rewrite, you have to convince both your mentee and yourself that the changes are required. On the other hand, a positive feedback is also required. Acknowledge a good work, it is important for learning. Don’t hesitate to say a good word if your mentee has written a beautiful piece of code or taught you about a new library. You can praise them in person or in a team meeting, it is important for self-esteem. From beginner to a developer Actually, writing code is usually the easiest part of the day-to-day job. Reviewing ideas and solutions with your teammates or PM and meeting deadlines is the real obstacle. In the first few times, it’s even a bit of a shocking experience. These are skills you learn overtime, and they are not easy to achieve. Most of the time, your mentee has never dealt with such tasks before. Let’s take time estimations for example. In my opinion, both of you have to miss deadlines at least once in order to understand how to provide more accurate time estimations. Missing the deadline will help you, as a mentor, to understand how much time a task should usually take. Before assigning a task to your mentee, evaluate how much time it should take, based on your previous experience. The more tasks you do together, you will have more data to base your time estimations on. Write to yourself how much time each subtask takes (DR, code writing, testing, etc.), and see overtime which area is changing. This way you can review how much your mentee has improved. In addition, you will know where they struggle and you will be able to focus on how you can help them improve. Walk in their shoes The most important thing to remember is that once upon a time you were a first time developer. Hopefully, you remember mistakes you did and mistakes that were done in the process of shaping you. You already know what influenced you to become the developer you are today. You also know how to improve methods that you think are not very good. Keep in mind that for some it takes more time and for others it takes less. Treat each mentee as an individual. Find their weaknesses and strengths. Provide your personal touch and help them improve, but don’t ruin their better skills. Use real life based experiences to influence your mentee. Help them grow into the best developers they can be. Your mentee and the company you both work for will gain a lot of benefits from this process. And most importantly, you will benefit a lot, both professionally and personally, from this kind of experience. --- ### Analog Clocks with NTP URL: https://www.taboola.com/engineering/analog-clocks-with-ntp/ Last Modified: 2025-01-14 14:37:28 Synchronized Clocks: Someone has to be blamed... I blame Ariel. He took one look at the five analog wall clocks set to timezones for various offices and said “I hate that”. I looked to see why. Yes, all the second has were wildly different and the minute hands were all slightly off too. I’d been thinking about what my next electronics project would be. “How hard could this be?” ## Goals Thinking about what I’d want this to do I came up with a few goals: - Correct and accurate time viaNTP(Network Time Protocol) - No two clocks should differ more than 100ms, they should all appear to tick at the same instant. - Battery operation. About one year between battery changes. - Automatic Daylight Savings Time adjustments - Remember clock position/state on power loss - Inexpensive - less than $20 in parts, not including the clock & batteries. ## Initial Research Started with checking if anyone had done this type of clock sync or something similar. Internet searches turned up quite a few clock projects. Many were digital and while interesting did not help much. Quite a few were complete custom clocks. I did find a few Analog Clock hacks. I found two that were similar to what I wanted: ( Networked Analog Clock and ESP Clock). The problem for me was I wanted battery operation and the ESP8266 cpu/wifi module uses 70 milliamps on average - much too much for the battery powered device I wanted. Another interesting one I found was “Crazy Clock”. This remained lower power but had no network connectivity to keep the clock synced with an external time source. It did however have a really nice description on how standard quartz clock movements work. And another, "Lunchtime Clock". It uses a real time clock module to keep accurate track of the correct time. This one also uses too much power and does not have network connectivity. ## Ticking the Clock I went shopping with the wife and bought two fairly cheap ($10) analog clocks and survived "the look" when I immediately disassembled them upon arriving home. Most quartz clock movements use a Lavet type stepping motor to physically drive the clock hands. Driving one of these requires us to alternate polarity each second. This can be done with a half H-Bridge type of circuit. I ran a number of experiments both attempting to directly drive the clock from microcontroller I/O pins and using an H-bridge type circuits. Testing driving directly meant using 3.3v or 5v depending on what micro-controller I was using at the time and I was concerned with using too much power since this was designed to be used with a 1.5v battery. I found a small low power motor controller (DRV8838) that supported motor voltages including 1.5v and decided to use that. (photo of testing operation of the DRV8838) While testing I found that the DRV8838 does active mitigation for inductive spike coming from its load. The downside is that the DRV8838 draws more than a milliamp when enabled and when it’s disabled the active mitigation vanishes leaving you with spikes as seen here: These spikes can cause tick failures and I was forced to delay disabling the DRV8838 for up to 50ms which increases the overall power usage. I actually built a few clocks using this until I realized that I could use pulse-width modulation (PWM) to control power when directly driving the motor with two of the microcontroller pins. To drive the clock I chose an ATTiny85V. Its an 8 pin microcontroller with 8k bytes of flash, 512 bytes of RAM, 512 bytes of EEPROM and 5 I/O pins, 6 when you configure the reset pin as I/O. I used all 6 of the I/O pins: 2 to drive the clock motor, 2 for i2c communication, one for a 1Hz signal to time clock ticks, and 1 to signal a power outage so that the clock position and state can be saved. I selected a DS3231 as a real time clock (RTC) as it is low power, has integrated temperature compensated crystal oscillator, supports a 1Hz square wave output, and has an i2c interface. I chose an ESP8266 as the main controller to manage configuration, NTP synchronization as it’s a really cheap (less than $3) wifi module with a flash-able microcontroller and multiple I/O pins. The way this works is that the ESP8266 wakes up at a configurable interval and makes an NTP request and computes the time offset. If the time offset is more than a threshold, and I’m using 20ms currently, the RTC is updated using a method that aligns its second boundary to where NTP says it should be. The RTC is configured to output a square wave at 1Hz with the falling edge being the second boundary. This is connected to one of the ATTiny85’s interrupt pins and configured to interrupt on the falling edge waking the the ATTiny85 from sleep to tick the clock each second. I assembled the version using the DRV8838 on a solderless breadboard for testing and initial software development. In this early version I was using an Arduino Nano in place of the ATTiny85 for convenience. ## Software Design The three main components, an ESP8266, a DS3231, and an ATTiny85 are interconnected with a two wire protocol called i2c. The ESP8266 is the bus master and both the DS3231 and ATTiny85 are configured as slave devices. When the ESP8266 starts, and this could be from a cold start like a power fail or a wakeup from its deep sleep mode, it attempts to connect to wifi. If it fails, or was never configured, it creates a captive WiFi portal that allows configuration of WiFi SSID and password, current clock position, NTP server, time change information, and some advanced settings that tune the parameters used to drive the clock motor. After connecting to WiFI it makes an NTP request and computes both the time offset and network delay as described in section 8 of the NTP RFC. The offset is used to adjust the RTC only if the offset is greater than a threshold and if the network delay is within 1 standard deviation from the last 7 or 8 requests. The last 8 offsets that were used to adjust the RTC are saved and used to compute the drift in parts per million of the RTC. This is used between NTP requests to help keep the RTC and the clock, ticking at the correct time. Next the current time is read from the RTC and converted to a clock position. Since most analog clocks only have a 12 hour face there are only 43200 possible unique hand positions so we represent the position of the analog clock as an unsigned integer between 0, midnight/noon, and 43199, 11:59:59/23:59:59. The current clock position is also read from the ATTiny85 and an adjustment is computed and sent back to the ATTiny85. The ESP8266 then enters its lowest power sleep mode, deep sleep, where only a wakeup timer and the 512 bytes of memory that we use to store the NTP samples are powered. Setting the RTC within milliseconds is a challenge since it keeps time only at a one second granularity. The key is that when you write the seconds value to the RTC it synchronizes its 1Hz square wave output to that moment. The specific verbiage in the DS3231 data sheet is “The 1Hz square-wave output, if enabled, transitions high 500ms after the seconds data transfer, provided the oscillator is already running. This makes the falling edge the start of each second. So by using the offset from NTP and coordinating when to set the time to be where NTP says the second should start syncs the RTC within a few milliseconds. The other RTC trick is that when we read the time for the origin timestamp of an NTP request we wait for the falling edge of the square wave first. Since this is a second boundary the NTP timestamp will be with zero fraction value. One area that’s still a work in progress is deciding how long to wait between NTP polls. The longer this is the less overall power will be used and the longer the batteries will last. Currently, and this is a work in progress, I’m using samples obtained since the previous adjustment to compute an estimated drift. This is then used to compute how long at the estimated drift rate it would take the clock to drift by the offset threshold. (Actually testing shows that less than 20% of the power usage is caused by NTP polls, so I’ll probably disable this feature and opt for a regular poll schedule based on the DS3231’s ±2ppm accuracy specification.) The ATTiny85 is interrupt driven and when not servicing an interrupt it’s kept in one of two low power states. Most of the time it’s in the “power down” state where it uses the least amount of power. It it is woken with either an interrupt from the falling edge of the 1Hz signal from the RTC, by an i2c “start condition”, or lastly by a power fail interrupt. On each falling edge of the 1Hz signal the ATTiny85 will execute an interrupt service routine that will start a tick on the clock. This is performed by configuring one of the timers to send a 4khz signal for a short duration to one of the two pins used to drive the clock. The exact pin alternates each second so that we are properly alternating the polarity to drive the clock motor. The duty cycle and duration of this is configurable allowing adjustments for different clock motors, different clock hand sizes, etc. While the PWM timing is active the ATTiny85 uses ‘idle’ sleeps between PWM interrupts to reduce power usage. A start condition indicates that the ESP8266 is attempting to communicate and the ATTiny85 will use “idle” sleeps until the command has been received then executed in an interrupt service routine. When it's complete the ATTiny85 returns to power-down. Typical commands are reading the current clock position and sending an adjustment, or possibly sending new tick duration or duty cycle values. When any value has been updated it it saved to EEPROM so settings will survive power outages. Power failure is signaled by the voltage regulator when it can no longer maintain its output at the correct voltage. This signal will interrupt the ATTiny85 causing it to finish any in-process tick and then save the current clock position in EEPROM. When power is restored it can immediately resume ticking without waiting for commands from the ESP8266. To maintain power long enough to accomplish this I used a large capacitor, isolated by a diode on the power (VCC) pin of the ATTiny85. The diode insures that the ATTiny85 is the only device powered from the capacitor. ## Power Management All of the main components have some form of low power mode that I’m using to reduce power usage. The ESP8266 has “Deep Sleep that allows it to use approximately 20 micro-amps. Deep sleep timer only lets you sleep for just over an hour at a time so I keep track of how much mode sleeping is needed and set deep sleep to wake without the radio turned on, saving power, if mode sleep is needed. During these intermediate wake-ups I apply drift compensation if its been computed. - Deep sleep: ~20 uA - Radio off: ~15 mA - Radio on: ~ 80mA The ATTiny85 has a “power down” mode where interrupts will wake it back up. Its woken on the falling edge of the 1Hz signal from the RTC and powers back down after the tick has been completed. An “idle” mode is also used during the “tick” processing where timer interrupts are used to wake from idle. - Power down: 0.2mA - Idle: 0.8 mA - Awake: 3mA I measured the overall power usage and found the following: (these were measured before I added hourly drift compensation) - NTP Request: 70 mA for 10 seconds - Hourly Wakeup between NTP requests 21 mA for 1 second - normal ticking: 0.28 mA Measurements were taken with an INA219 current sensor module attached to a Raspberry PI and graphed using gluplot: ## Clock Modification In order to drive the clock we have to solder a pair of wires to the clock’s motor. I usually use a sharp knife and cut any traces on the clocks original PCP that connect to the motor as well. ## Configuration When the clock is initially powered on it creates a wifi captive portal. It will show up in the list of available wifi networks as SynchroClockXXXXXXX (where the X’s are some number). When you connect to this you are given a menu that lets you set many configuration options: - Wifi Network (SSID). - Wifi Network password. - Clock Position - enter the current time shown on the clock as HH:MM:SS (note that the clock is always stopped when the config portal is up and it should display the correct hand position.) - NTP Server to use for time synchronization. - 1st time change as 5 fields (US/Pacific would be: 2 0 0 3 2 -25200 meaning the second Sunday in March at 2am we change to UTC-7 hours). occurrence - ‘2’ would be the second occurrence of the day of the week specified, ‘-1’ would be the last one. - day of week - where ‘0’ = Sunday. - day offset - needed for “the friday before the last sunday of the month”, this would be occurrence = -1, day of week=0, day offset = -2 - month - where ‘1’ = January. - hour - where ‘0’ = midnight. - time offset - this is the offset in seconds from UTC. - 2nd time change as 5 fields as described above (US/Pacific: 1 0 0 11 2 -28800 meaning the first Sunday in November at 2am we change to UTC-8 hours). Advanced options: - Stay Awake - when set true the ESP8266 will not use deep sleep and will run a small web servers allowing various operations to be performed with an http interface. (I use this to test new clock timings) - Tick Pulse - this is the duration in milliseconds of the “tick”. - Adjust Pulse - this is the duration in milliseconds of the “tick” used to advance the clock rapidly. - Adjust Delay - this is the delay in milliseconds between “ticks” when advancing the clock rapidly. - Network Logger Host - (optional) hostname to send log lines to. - Network Logger Port - (optional) tcp port to send log lines to. - Clear NTP Persist - when set true clears any saved adjustments and drift calculations.   ## PCB Design I designed a simple PCB available from OSH Park. ## Source Code and Design Files All source code and design files are located in a GitHub repository: https://github.com/liebman/AnalogClock. ## Future Ideas - Li-ion battery & charging - Solar or other energy scavenging technique - Support 16 HZ "non-ticking" or “silent” clocks --- ### How teaching in high school helped me become a better team lead URL: https://www.taboola.com/engineering/teaching-high-school-helped-become-better-team-lead/ Last Modified: 2025-01-14 14:37:28 A few years ago, one of my friends suggested me to become a cybersecurity teacher in high school once a week as part of a program called Gvahim. I have not planned that it will contribute to my professional career, but I find a lot of analogies to my day to day role. I hope you will enjoy a different angle of management 101 guidelines. ## Program overview The program’s goals were to increase the knowledge of high school students in cybersecurity and increase the number of girls who study computer science. For three years in the program, students studied about Assembly, networks and operating systems, with an emphasis on security. Unlike traditional materials learned in high school, the lessons in the program put an emphasis on self-learning. The first two semesters were dedicated to learning the theoretical background using self-reading and small coding exercises. The last semester of the year was used to code a large-scale project, which the kids presented at the end of the year. ## My Role The format of the program was that a teaching assistant (that’s me!) from the tech industry joined the computer science teacher, for a few hours during the week. While the teacher was the pedagogical authority in the class, my role was to give the kids the tech industry’s perspective on the material. Another goal I had was to find and challenge top performers. The most common techniques I've used were raising the standards of those kids’ code, and asking them to help other kids. The busiest time was at the end of the year, in which the kids were working on a different project of their choice. For example DOS game like chicken runners in assembly and a chat application in python. In that time, different kids dealing with different problems, which was more of a challenge for me. Last but not least, I had to give special attention to girls, whose ratio in the industry is pretty low. During this time, my role was changed and I became a team lead. Though it was a gradual change, I feel there are a lot of similarities between those roles. Actually, the time in the class was much more intense. Mentoring 30 kids is more challenging than mentoring 5 experienced developers. ## My Insights  ### 1. The “sweet spot” - give the answer or let the mentee learn on his own Learning from mistakes is a powerful learning tool. Thus, I’ve let the kids do similar mistakes and learn from them “the hard way". At the end of every exercise, I explained the solution to the entire class to make sure everyone is on the same page. Most kids were seeking immediate answers. Some of them were unfamiliar with the concept of self-learning. Only in Cyber class the kids were required to learn on their own. The teacher and I had to insist on self-learning, which is a fundamental quality in this ever-evolving tech industry. Also, I had to learn how to answer questions with other questions. This was important to develop their self-learning skills. ### 2. Get the most from your 5 minutes per student Managing your 3-4 hours per week during the class is challenging. I had to choose how to spend my time between 30 kids: Challenge the high performers, motivate the kids who don’t want to learn, or motivate the strugglers that do have motivation. Also, context switches happened all the time, which was a good practice for me. Average of 5 minutes per kid is not a lot of time. Giving the "right" feedback will save you time. But what is the "right feedback”? A feedback which can help the mentee to progress towards a solution, and is adjusted to the mentee skill level. For high performers, I’ve tried to raise the bar by insisting on higher standards. For example, code separation to functions, meaningful names, and other clean code standards. For strugglers, I focused on software engineering basic skills: First, how to translate written requirements to an algorithm - whether they are clear or not. Second, how to debug your code by teaching kids how to use the debugger and where to put breakpoints. It’s also important to adjust your feedback according to different personalities. For insecure kids, I’ve tried to create a feeling of success. Giving encouraging words during code reviews such as “great”, “good job” or “you are in a good direction” can help. Asking questions can help to boost confidence - when you answer someone else’s questions on your own, it will increase your feeling of achievement. ### 3. Women in high tech - should we approach differently? Professional skills and personalities are not the only attributes which we should refer to. Another important and controversial topic that I was exposed to was that physical traits, such as gender and general background, are also affecting the way of thinking.  One of the goals of the program was to increase teachers' awareness to gender difference. A special supervisor was responsible for this area across all schools in the program.  Only 20% of tech jobs are held by women - Why does this happen? According to studies, the misperception of scientific professions as a ‘men profession’ starts from an early age. There are different social perceptions between women and men roles. Poll results from 2009 showed that only 10% of the girls say their parents have encouraged them to think about an engineering career ( Harris Interactive on behalf of the American Society for Quality). Second reason is related to the insecurity of female students. A survey among female students in the program showed that 54% of the girls feared from learning cyber (119 respondents). 90% thought that the material will be hard, they would not understand it, and the studies will be too intense. Only 10% fear from a bad teacher, boring material and negative effect on their social life. According to studies, it’s not related to biology or lack of skill-set. Some studies have shown that girls are over performing over boys in math and science all over the world. What can you do Be aware! Sometimes we can have unconscious tendency towards specific gender (checkout this test). According to Prof. Juliette Walma van der Molen, “It is important to organise a training course on gender for teachers, because, more than they think, teachers subconsciously interact with students based on gender stereotypes. By raising awareness of these situations through training, we can try to counter such stereotypes in education.” Sometimes we don't give enough significance to the various backgrounds and cultures we come from - gender is only one example.  How can you change that? First, create meaningful relationship with the mentee. You can do it by providing mostly positive criticism and sometimes constructive criticism. Adjust your feedback for every mentee, and show genuine interest in mentees inside and outside of the workplace. Second, pay attention to structure and organization. When transferring knowledge, write the structure of the lecture, organize the board and make sure knowledge transfer sessions have similar and well known structure.  Last but not least, set clear expectations from the mentee: Define clear answers to questions like what is required when approaching a problem, how does readable code look like, and what is required to succeed in a mission. ### 4. Learn from your peers I wasn’t the only authority in class. The teachers I worked with were experts in creating a positive learning environment. They were making sure kids were listening. We’ve used each other’s strengths to get the most from each lesson. Sometimes, we used the top performers to mentor the low performers. Most of the time, you won't be the only leader. Advise with your colleagues, within the team or outside, to achieve your team goals. ## Get the most from your mentees As a leader and a mentor, your main goal is to make the mentees fulfill their potential while executing on time. Ask questions, and don't give straight answers. Adjust your mentoring methods according to the mentee skills and personality. You should also refer to the background, gender and mental mood when giving feedback. Unfortunately, your time is limited - you need to learn how to manage it effectively. Decide how much time you want to spend with each mentee, and how to use it effectively. Juggle between improving your employee skills and letting them make mistakes, and push them to deliver missions on time. ##### References ##### TWIST - Enhancing gender awareness in teaching ##### Girls can learn physics (video, hebrew) ##### Girls Make Higher Grades than Boys in All School Subjects, Analysis Finds ##### Gvahim --- ### Video campaigns budget pacer URL: https://www.taboola.com/engineering/video-campaigns-budget-pacer/ Last Modified: 2025-01-14 14:37:28 As a content discovery product, we need to be able to pace the campaign through its life on real time - spending the budget entirely without overspending. The team I am leading is responsible for serving Taboola’s video content. Our main goal is to enable growth of our business. Owning the entire serving process of the video content can be crucial to this end. Depending on 3rd parties serving systems with the core process would leave us vulnerable to rising prices, compromised features, serving latency, reduced performance and so on. In order to serve the video on our own we had to come up with a way to pace the amount of times we want to display the video and prevent instances in which the entire budget of the video would drain in a few seconds - common scenario in Taboola’s scale when the video is not targeted aggressively. We also needed to support partners that want to display the video for a very short time (targeting sport events, for example), so we needed to develop real-time mechanism as well. Thankfully, our solution was ready just as we had to stop working with our 3rd party serving system, which was responsible for serving and budget pacing, due to business needs. Pacing video campaigns We started by meeting with a few talented developers and began throwing ideas and metaphors, from banking mechanism to toilet niagara. The straightforward solution would be to calculate the amount of impressions out of the total budget and then distribute it per server across the campaign life time. Due to our scale, we have about 100 servers (only within the video group) and growing, all deployed in ci-cd process, which makes the actual number of serving machines to change all the time. Understanding how many servers are requesting budget would be hard to handle. Moreover, this solution would behave as serving bursts, so we decided the pacing would be managed by throttling each campaign serving. Understanding the terminology Throttle is a value between 0-1, which represent the probability of a campaign to be displayed on each video request. The pace is the time percentage that had passed from the total life-time of the campaign divided by the budget percentage that was spent. When the pace is equal to 1, the spending is evenly paced. To reach pace = 1, we update the throttle by multiplying each iteration by the inverse pace. As a result, we achieved the wanted behaviour of paces < 1 to increase throttle and vice versa. *Evenly paced campaign - linear spent graph over time *Pace = %Budget / %Time *Evenly paced spending should always be on pace = 1 Data pipeline Using this pacing technique requires us to have real time capabilities. Our data pipeline uses Kafka, so having real time properties would simply require having Kafka consumer in the process. In order to have stateless mechanism we chose Aerospike for our nosql, because it excels in low latency batch reads of small records, just as we needed in the software. It also enables sharding out of the box to support scaling. The throttle values for all the paced campaigns are being requested from the video servers every few seconds. The request is made by a separate thread and stored in a local cache using GRPC. Every video campaign request generates a random number - if the number is lower than the throttle in the cache, the campaign is being served. If the number is higher, the campaign is being filtered. Architecture Overall, we wanted to build a robust, scalable and reusable program which will be decoupled from the rest of the server. We expected it to run in real time, and be agnostic to the amount of servers uses the program. Sound plan, but will it work? We wanted to reduce time wasted on developing, so we started by exploring and simulating the entire flow - thread pool of server objects requested the throttle and reported back the amount of budget they used. The throttle was recalculated to adjust the pace and the server objects used the new value and so on, during the configured life time and with the exact wanted budget. Time for the real world We wired Backstage,Taboola’s campaign management dashboard, for video campaign creation through an API. This way each campaign will have budget entities value - each one represents an entity we need to pace. RX Java based consumer was extended from an infrastructure we have in video server, listed to a topic containing revenue data and aggregates the campaigns with the relevant dates. We then write the aggregation into Aerospike. Throttle calculation Each campaign starts with a small throttle value of 10-6 and every calculate iteration (10 seconds) the value is increased by 20% until spent appears in the Aerospike record. Then we apply the formula T = T*1/P. During this process we constantly check the delta of the pace values. When the delta is smaller than a given threshold, we save this value. When the pace crosses value of 1, we apply it and keeping it on the right pace. All the calculated values are being stored in Aerospike and local map, which is being requested by the video servers using GRPC. We reduce the load of consuming by having two Budget Pacers. They also balance the load of the video servers requesting throttles, but only one Budget Pacer is allowed to calculate and write throttles into Aerospike - Zookeeper decides which Budget Pacer is the master, the slave allowed only to read the throttles and expose them through GRPC. All our data is stored in two data centers for full data recovery. Enabling this capability in Budget Pacer, we have instance on both data centers, each having its own separate Kafka consumer and Aerospike. Console directs the server calls to the master DC and balance the load between the two Budget Pacers within the DC. Happy budget pacing During the last quarter the activity has more than doubled the amount of campaigns and budget it serves. Campaign can reach full serving capabilities within minutes from creation. The infrastructure proved itself so much, that we decided on using it for two new features - Throttling RTB calls based on their performance and calculating pricing of one of our business models. --- ### Being an enabling QA Engineer - You should try it! URL: https://www.taboola.com/engineering/enabling-qa-engineer-try/ Last Modified: 2025-01-14 14:37:28 On our day to day lives, professional relationships matter. Theoretically, how QA should handle themselves with developers is very obvious. Or is it? Well, it’s not. Reporting to developers about an issue and leave it like that is not a good enough approach. It’s way too basic and distant. Professional relationship should include talking to the developer about the issue. Sometimes it requires further explanations. Other times, it will require helping them reproduce it. It will also have to involve a good set of interpersonal communication skills. “Us” vs. “Them” From the early days of my career, I never stopped hearing about the “Us” (QA) vs “Them” (developers) perception. I never joined those calls. Not because I feared to speak up my voice, but rather because I could never relate to it, even to this day. I think this perception is useless and has nothing to do with teamwork. The QA are a part of a bigger team. QA are not lone wolves that wait for the attack and then retrieve. If you have any wish for your product to be successful, you have to work in a collaborative way. Only great teamwork can take your product to great places. Expect to find bugs, but keep a positive attitude New features to test will always have bugs. That’s ok. It keeps us working and ever challenged to find new “holes in the plot”. Wherever there’s code, there will be bugs to keep that code warm at night. One of the main characteristics of QA engineers is being skeptic about the code. Yet, they must stay positive. That means they should not start testing while being judgmental about the code, nor should they hold assumptions about the developer who wrote it. Also, they should never be cynical about the upcoming test. The above traits are harmful in the day to day relationship of the QA engineer and the developer. Human communication instead of tools Yes, Jira (or any other bug reporting tool) is a nice tool to communicate through it. Using it can be efficient and clear. Yet, you need to keep in mind that the best communication form is the verbal one. Get up from your seat and go talk to the developer if you are unsure about your findings. Talk to them about any issues you caught, especially the critical ones. Worst case, you gained some exercise. A professional QA engineer should have excellent communication skills. These skills come into display in 2 main forms: First, they should keep in mind that the developer who wrote the code cares about the code. When the developer says that the code is bugless, they actually mean that. It is not just a saying to make the QA think he is waving them off. When returning to the developer with bugs, they will obviously dislike it. You should break it to them gently - don’t be arrogant. Second, before you start your test, you need to have a good handover talk with the developer. Ask about what exactly changed and where, what are the important cases to look at and so on. On the one hand, the QA engineer sees the “bigger picture”. On the other hand, the developers know their code well and can guide you through it. They can help you find stuff you might not have thought of before. Becoming the enabling QA During my career, I handled a positive relationship with any developer I’ve ever worked with. Developers aren’t all alike. They each have a different approach, belief and coding strategy. The QA engineer should know his way around with any developer he works with. Respect them and you will get respect back. Talk to them in a cool and easy tone and you will always get answers in the same manner. Walk them through what you’re going to test. Seek their advice and show them you appreciate it. The enabling QA is the one that doesn’t talk in the form of “I am..”, but rather in the form of “We are”. Remember that you are part of a team and act like it. Help the developers create the best possible version. This type of version has as little as possible amount of bugs (and zero critical ones!) within the preset timetable. If they fail, then you have failed as well. If they succeed, then so have you. Let it go, let it go Know when to stop testing. Let go of too minor or too extreme cases. You can get advice from your colleagues - peers, developers, PM. Communication, remember? Don’t hold back on giving the green light on a version just to spite. It will not make you look more professional. It will often be perceived as the opposite. Remember to have fun “Breaking” stuff is fun and so is software testing. We get to play with innovative products and new technologies. We also get to find creative new ways to make them work less than expected or not work at all. You get to work with smart and fun people that you can learn from. Yes, our job is important and can prevent any product from being a complete disaster as it rolled out. Caring about our work is super important. You should respect it, but you need to remember to have fun while you do it. Enjoy the ride. If you appreciate what you do, You’ll do it better! Love what you do. I know I do. --- ### Deep Learning - from Prototype to Production URL: https://www.taboola.com/engineering/deep-learning-from-prototype-to-production/ Last Modified: 2025-01-14 14:37:29 About 8 months ago my team and I were facing the challenge of building our first Deep Learning infrastructure. One of my team members (a brilliant data scientist) was working on a prototype for our first deep model. The time arrived to move forward to production. I was honored to lead this effort. Our achievements: we built an infrastructure that ranks over 600K items/sec, our deep models have beaten the previous models by a large margin. This pioneer project has led the way for the subsequent Deep Learning projects at Taboola. So the prototype was ready, and I was wondering: how to go from a messy script to a production ready framework? In other words, if you are into establishing a deep model pipeline this post is for you. This blog post is focused on the training infrastructure, without the inference infrastructure. ### Prerequisites Assume you have basic knowledge in: - Python - Train Machine Learning model - TensorFlow - Docker Let's take one step back and think…Ideally such a training infrastructure should support: - Comfortable development environment - Portability among environments - Run on CPU/GPU - Simple & Flexible configuration - Support Multiple Network Architectures - Packaging the model - Hyper Parameter Tuning One mantra to keep in mind: ‘Speedy Gonzales Time to Market’ == ‘no over engineering!’ == ‘Keep it simple’ Enough with the baloney, let's get down to business: ### Technology stack BigQuery, Google Storage (GS), Python, Pandas, TensorFlow, Docker, and some other Python nice libraries: fabric, docker-py, pyformance & spacy ### Comfortable development environment This makes the whole team move faster, plus putting a new team member into action is super easy. To achieve this we did 2 things: The first was using a small dataset for sanity training, a JSON file with 2K rows. Used both for testing on a developer machine and for sanity automation tests. This is very helpful for quick testing your code changes. The second is a project management script (just one big Python script). We implemented a simple command line tool using ‘fabric’ (Python library). Here are few example commands: $ fab --list Available commands: init_env initializes Python virtual environment & installs 3rd party test runs unit tests train runs sanity training train_all runs sanity training on all model variants docker_build builds snapshot Docker image docker_test same as test, only from inside the Docker image docker_train same as train, only from inside the Docker image Using docker-py library was very useful for the Docker stuff. CI tasks were also added to this script (used by Jenkins build agent). ### High level flow Dataset is generated from raw data tables in BigQuery and stored as a JSON file in GS. Then the training process kicks in: download the dataset, train the model, and export the model back to GS (to some other path). ### Fit dataset into memory Not always possible, but it was for us. Our dataset contains about 5M rows (~ 4GB JSON file). The JSON file is loaded into pandas’ DataFrame for pre-processing, then split into train & test. This approach is simple to implement, gaining fast training run time. If the dataset does not fit into the memory, a quick (maybe temporary) solution could be subsampling it. ### Portability among environments No more “It works on my machine”! By using Docker, training runs the same on the developer machine, automation and cloud environments. Training input & output is stored in GS, no matter where the training process ran. ### Running on CPU/GPU - Install both TensorFlow GPU and CPU versions (in the project management script: use the GPU for production training and the CPU for development and testing) - Use NVIDIA Docker base image - note that the CUDA version should match the TensorFlow version (see more instruction on TensorFlow’s website) FROM nvidia/cuda:8.0-cudnn5-devel RUN pip install virtualenv==15.0.3 RUN virtualenv --system-site-packages /venv-cpu RUN virtualenv --system-site-packages /venv-gpu RUN chmod +x /venv-cpu/bin/activate RUN chmod +x /venv-gpu/bin/activate COPY ./requirements.txt /tmp/requirements.txt RUN /venv-cpu/bin/pip install -r /tmp/requirements.txt RUN /venv-gpu/bin/pip install -r /tmp/requirements.txt ENV MY_TF_VER 1.1.0 RUN /venv-gpu/bin/pip install https://storage.googleapis.com/tensorflow/linux/gpu/tensorflow_gpu-$MY_TF_VER-cp27-none-linux_x86_64.whl RUN /venv-cpu/bin/pip install https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-$MY_TF_VER-cp27-none-linux_x86_64.whl - Use nvidia-docker command line tool - a thin wrapper for Docker command line. It enables the Docker container to use the GPU ### Simple & flexible configuration Flexible for research, live models & hyper parameter tuning. Configuration should be simple to reason - I have seen too many configuration mechanisms get out of hand. I think we found a fair trade off between flexibility and simplicity: First, no hierarchies! (keep it as simple as possible). For example: Hierarchies are Bad: -- pre_prossesing: -- subsample_ration = 0.3 -- training_loop -- batch_size = 124 -- epochs = 100 -- hyper_param -- learining_rate = 0.1 …. Flat is Good: -- subsample_ration = 0.3 -- batch_size = 124 -- epochs = 100 -- learning_rate = 1.1 Config values are set in 3 levels: Level 1 - NetConfig class: class NetConfig: def __init__(self, graph='deep_v1', batch_size=256, learning_rate=0.001 ): …. As you can see default values are hard coded as part of the configuration class. Level 2 - variant_config.json: "variant_1":{ "learning_rate" : 0.001 }, "variant_2"{ "graph": "deep_v2", "batch_size": 124 } Here we define our model variants. Only overriding non default configurations. The configuration file is held with the code, and baked into the Docker image. A new Docker image is built for every configuration change. Level 3 - environment variable: nvidia-docker run -e batch_size=64 …. Using environment variable enables overriding a specific configuration for a single training run. We use it for research and hyper parameter tuning. ### Support Multiple Network Architectures Supporting multiple network architectures is a key concept for us. It enables many team members to work in parallel on different architectures (and set A/B tests on live traffic). So how do we support multi network architectures? Let's take a closer look into ‘build network graph’ step: def build_model_graph(net_config, dataset_shapes, tf_session): if config.graph == ‘deep_v1’: return build_graph_deep_v1(net_config, dataset_shapes, tf_session) elif config.graph == ‘rnn_v1’: return build_graph_rnn_v1(net_config, dataset_shapes, tf_session) ... ‘build_graph_deep_v1’ & ‘build_graph_rnn_v1’ are just static methods that contain the TensorFlow code for defining the graph. ‘build_model_graph’ returns an OpsWrapper instance, holding the TensorFlow ops for the training loop: class NetOps: def __init__(self, batch_size_placeholder, dropout_placeholder, input1_placeholder, input2_placeholder, loss_op, train_op …. ) Note that the ‘build_model_graph’ has 3 inputs: net_config (discussed earlier), dataset_shapes and tf_session. The dataset_shapes object is created during the preprocessing step. It contains features vector information that is used to build the network graph. For instance, the cardinality of a categorical feature is used to define the size of the embedding lookup table. class DatasetShapes: def __init__(self, input1_cardinality, Input2_cardinality, num_of_unique_words, sentence_max_size … ) The tf_session in just an instance of tf.Session class. ### Packaging the Model Packaging the model TensorFlow files inside a zip with other extra files, can help debug and monitor the training process. What extra files did we store? - TensorBoard files - Graph image of train & test offline metrics - Offline metrics CSV: - > row for each epoch - > add all kinds of offline metrics, such as train & test error & loss - DeFacto configuration - used to train the model - Dataset metadata - for example dataset size - Other things you may find useful for troubleshooting & debugging ### Hyper parameter tuning Hyper parameter tuning is critical. A big improvement on model performance can be gained by tuning parameters such as, batch size, learning rate, embedding dimension, hidden layer depth, width, and so on…. We started by implementing a simple Python script that runs a random search on the hyper parameter space. It runs sequentially using a single GPU. We just ran it for days. To get faster and better results we moved to the cloud, running in parallel on multiple machines & GPUs. All-in-all we had lots of fun working on this project (at least I did). We ended up with a relatively simple and straightforward infrastructure, being used by the whole team, raising somewhere between 10-15 live A/B tests in parallel. That is it! I hope you find it useful. Special thanks to my team: Yedid, Aviv, Dan, Stavros, Efrat and Yoel. --- ### Leveraging Vertica Performance by Reducing CPU System Calls URL: https://www.taboola.com/engineering/leveraging-vertica-performance-by-reducing-cpu-system-calls/ Last Modified: 2025-01-14 14:37:29 What is the connection between kernel system calls and database performance, and how can we improve performance by reducing the number of system calls? Performance of any database system depends on four main system resources: - CPU - Memory - Disk I/O - Network Performance will increase while tuning or scaling each resource - this blog will cover the CPU resource.It’s important to note that whenever we release a bottleneck in the system, we might just encounter another one. For example, when improving CPU performance the database load shifts to IO, so unless our storage is capable of delivering more IOPS, we might not actually see the improvement we hoped for. But don’t be discouraged, performance tuning is sometimes a game of whack-a-mole... We all know that the more processing power available for your server, the better the overall system is likely to perform. Especially when the CPU spends the majority of its time in user state rather in kernel state. At Taboola we struggled with two cases which caused Kernel state to consume a lot of CPU; your system might have the same problems. ### How do we collect CPU metrics at Taboola? Metrics are collected using Sensu and Graphite, dashboards are built with Grafana. The Sensu CPU plugin cpu-metrics.rb collects the following CPU metrics: user, nice, system, idle, iowait, irq, softirq, steal, guest. Since the Vertica process runs in a CPU user space that has been niced, we will mostly focus on the nice and system metrics. ### System calls CPU impact by calling Vertica new_time function What are system calls? System calls are how a program enters the kernel to perform a task. Programs use system calls to perform a variety of operations such as: creating processes, performing network and file IOs, and much more. If a system call is such a basic internal interface between an application and the Linux kernel, why should we care about it? - Sometimes this is the case, However if the system calls’ CPU usage remains high for long periods of time, then it could be an indication that something isn't right. A possible cause of system calls’ spikes could be a problem with a driver/kernel module or sometimes due to high concurrency - To run a system call the CPU needs to switch from the user to kernel mode, that’s called context switch, context switching itself has performance costs.  At Taboola, all timestamp data is saved in UTC timezone. Aggregated reports for raw data tables produce reports converted to the publisher (customer) timezone using Vertica’s “new_time” function  to convert between time zones. For example, aggregating actions by hours: SELECT date_trunc('hour', new_time(action_event_time,p.time_zone_name, 'UTC') ), count(action_name) FROM actions a JOIN publishers p ON a.action_publisher_id = p.id WHERE a.action_event_time BETWEEN new_time('2017-04-20 00:00:00.0',p.time_zone_name, 'UTC') AND new_time('2017-04-24 23:59:59.999', p.time_zone_name,'UTC') GROUP BY 1; The following image represents a server’s overall CPU usage over time, stacking all types of CPU metrics: System CPU Usage: We can see regular spikes in usage reaching up to 74%. We need to profile our CPU calls to find the root cause. ### How do we profile CPU calls? Linux perf is a powerful tool to diagnose this, and a FlameGraph can be used to virtualize the perf report by drawing a graph. Useful perf commands: - perf top - realtime events - perf record  - run a command and record its profile into perf.data - perf report - read perf.data (created by perf record) and display the profile   Running the following commands will record and report CPU calls for a duration of 5 minutes: perf record -F 99 -a -g -- sleep 300 -- Take 99 samples/second for five minutes perf report - -sort cpu > /tmp/report.out -- Generate calls reports order by cpu time The /tmp/report.out file output shows the following: 67.72% vertica __ticket_spin_lock | --- __ticket_spin_lock | |--99.38%-- _raw_spin_lock | | | |--53.12%-- futex_wake | | do_futex | | sys_futex | | | | | |--100.00%--system_call_fastpath | | | | | | | |--99.98%--__lll_unlock_wake | | | | | | | | | |--99.95%-- getTZoffset | | | | | | | | | | | |--35.86%-- _ZN2EE5VEval16ZoneTS_TStz_SkipEPciiPi | | | | | | _ZN2EE5VEval16ZoneTStz_TS_SkipEPciiPi If flame graph is installed, we can run: perf script > /tmp/script.out FlameGraph-master/stackcollapse-perf.pl /tmp/script.out > /tmp/flame_out.folded FlameGraph-master/flamegraph.pl /tmp/flame_out.folded > /tmp/kernel.svg Flame Graph (kernel.svg) example: Each call to new_time function triggers the OS call getTZoffset, while many simultaneous calls are causing kernel contention expressed by spin locks. ### Solution We removed the use of the new_time function, and started joining our data to  “conversion tables” instead.   We created two timezone tables that contain a local time for each timezone and the corresponding time in UTC: 1. time_zone_daily_conversion 2. time_zone_hourly_conversion Daily Table Definition Column | Type | Size | Default ----------------------------+------------------------+---------- time_zone_name | varchar(400) | 400 | day | date | 8 | NULL::date day_starts_at_utc | timestamp | 8 | NULL::timestamp day_ends_at_utc | timestamp | 8 | NULL::timestamp Sample Content time_zone_name | day | day_starts_at_utc | day_ends_at_utc ---------------------+------------+---------------------+--------------------- Africa/Nairobi | 2017-07-28 | 2017-07-27 21:00:00 | 2017-07-28 20:59:59 America/Bogota | 2017-07-28 | 2017-07-28 05:00:00 | 2017-07-29 04:59:59 America/Mexico_City | 2017-07-28 | 2017-07-28 05:00:00 | 2017-07-29 04:59:59 America/Sao_Paulo | 2017-07-28 | 2017-07-28 03:00:00 | 2017-07-29 02:59:59 Asia/Bangkok | 2017-07-28 | 2017-07-27 17:00:00 | 2017-07-28 16:59:59 Hourly Table Definition Column | Type | Size | Default ---------------------+--------------+------+-----------------+--- hour_at_utc | timestamp | 8 | NULL::timestamp time_zone_name | varchar(400) | 400 | hour_at_time_zone | timestamp | 8 | NULL::timestamp day | date | 8 | NULL::date Sample Content hour_at_utc | time_zone_name | hour_at_time_zone | day ---------------------+------------------+---------------------+------------ 2017-07-28 00:00:00 | Pacific/Auckland | 2017-07-28 12:00:00 | 2017-07-28 2017-07-28 00:00:00 | US/Central | 2017-07-27 19:00:00 | 2017-07-27 2017-07-28 00:00:00 | US/Eastern | 2017-07-27 20:00:00 | 2017-07-27 2017-07-28 00:00:00 | US/Mountain | 2017-07-27 18:00:00 | 2017-07-27 2017-07-28 00:00:00 | US/Pacific | 2017-07-27 17:00:00 | 2017-07-27 Example of aggregating actions by hours using the conversion tables: SELECT hour_at_utc, count(action_name) FROM actions a JOIN publishers p ON a.action_publisher_id = p.id JOIN time_zone_hourly_conversion tz ON p.time_zone_name = tz.time_zone_name WHERE tz.day BETWEEN '2017-04-21' AND '2017-04-23' AND action_event_time BETWEEN '2017-04-20 00:00:00.0 ' AND '2017-04-24 23:59:59.999' GROUP BY 1; ### Result Query time reduced by 50-75%, system calls CPU metric graph shows: ### Conclusion - It is essential to monitor server metrics (CPU, memory, load, and others) to identify performance bottlenecks. - Perf is a great profiling tool to drill down into the major system calls. - In our case we found the issue was caused by spin locks, but other customers’ clusters or use cases might suffer from other kernel calls. --- ### All code is guilty until proven otherwise URL: https://www.taboola.com/engineering/code-guilty-proven-otherwise/ Last Modified: 2025-01-14 14:37:29 Delivering good product to live environment requires big effort from R&D. Under the software development life cycle, we can find 6 basic phases: Understanding the requirements, design, coding, testing, deployment (incl. A/B test, if necessary) and maintenance. But how can we measure product quality? By its stability? Scalability? Easy to maintain? Bug free code? There are probably many definitions for what is a good product, but in my opinion, the two foundation stones are product behavior & functionality as defined (be aligned with the product manager's requirements), and zero critical bugs. The product can serve many goals, but if it doesn’t achieve the main one, it might not have a reason to exist. Naturally, customers are always expecting high quality from the product, so before releasing it to production QA should make sure that indeed critical bugs don’t exist. In order to respect these two, both R&D and QA should be fully committed here. From QA perspective, things can get complicated while running end to end tests which involve UI/UX. Why is that? Cross browser testing - 9 different environments Everyone who's involved with the product, should ask themselves what's the product's target audience. In other words, different geographical locations dictate different browser usage. For example, if I would like to approach a U.S. desktop audience, I would probably start my test coverage with Chrome, Edge, Firefox & Internet Explorer. Below is a glimpse of data taken from Statcounter: But of course, the product should run under additional browsers, that should also be supported. unless the product meant to run only on one browser. In this case, for desktop environment, QA should cover Chrome (PC & Mac), Edge, Internet Explorer, Safari & Firefox. As for mobile, we should cover Chrome in Android and Safari & Chrome on iOS. Configure your tests Every new product, or even a feature, holds different configurations. So, in addition to a sanity check, we should also do regression tests which should be focused on new functionality under the different configurations. A simple feature can be composed of many parameters that can get different values representing different states of the product. When examining the test coverage, we encounter a 3-dimensional matrix which consist of Browser X Configurations X test cases. Below is an example of this matrix. Looking at the matrix for a feature that we have tested a while ago, we can find for instance, 9 Browsers X 3 Configurations X 53 test cases, which adds up to ~ 1500 tests. Besides running the matrix, there are many action items that should be taken from the QA side before actually starting to run the tests. Learning from past failures & successes during our QA cycles, we’ve started to adapt a new approach for managing our tests. In a perfect world, QA should run all scenarios under all configurations & environments in order to give a green light to go to production. Obviously, this isn’t the case. Being the last wheel before production, QA should find the optimal cross section between quality and time consumption, which involves taking risks. QA Kickoff In order to save time and be as ready as we possibly can, before the feature is ready for testing, QA can start creating the test matrix according to the specs containing the relevant test cases. A good tip here will be doing a test review with another QA engineer in order to get feedback and have an open discussion about the matrix - changing, adding or removing test cases. Actually, it is very similar to the code review which is done by dev. After doing so, it’s recommended to do it also with the relevant PM & dev in order to set expectations - what is tested and where, while managing the risk & giving time estimations. Naturally, when dev finishes their part, the next step is to start running the test matrix. From my experience, there are some necessary steps which should be done beforehand. It’s recommended that QA will do a handoff with dev in order to understand what was developed, which configurations / parameters should be used, how should it look like, and most importantly - ask questions regarding the behavior, environments, list of known issues, etc… Before moving the feature to QA, dev are usually running some tests from their side, to make sure that the feature is not broken. Many times, dev creating its own environment (mockups, tools, setup, test pages) for testing, and afterwards QA also use it. This can be harmful in 2 ways: QA misunderstands the process of building this setup, and so, you are staying in your comfort zone, not thinking about other setups. Doing that, you are missing potential bugs. It’s always preferable that QA will build its own environment and use it regardless of dev’s environment. Last but not least, it’s always useful to go over old test docs & old bugs that are related to the same area you are testing, in order to verify different issues that weren’t in the scope. Hands on Are we ready to start with the matrix already? It depends. You can rush into covering the matrix, but the risk here is catching bugs in a late stage which can delay the deployment to production. A couple of things that can help here: Create 1 basic “happy flow” that tests the feature straight forward (without any special cases) and run it across all environments (OSs). This is the very basic sanity to check if the feature is working and not breaking anything else. In addition, I would run with the feature only on Chrome (most commonly used browser today) and do some freestyle testing, changing some flags and verify that the behavior is expected. In a couple of hours, performing these 2 steps, gives QA a basic look & feel of the feature, some confidence about its stability & could also brings new ideas for additional test cases. If no blocker bug has been found during the last session, QA can proceed with the matrix. According to deadlines, it’s a common decision whether to check the whole matrix or to cover it partially. On very important projects, we tend to cover it all, without exceeding the reasonable timelines. While most of the projects are not considered as super important, our plan is to verify all test cases on Chrome. As for the other browsers, it’s your decision which boxes should be covered in the test matrix and which tests you are going to run - resulting in a combination of randomness with some intentional decision. Learning from past mistakes, encouraged us to build new procedures within our QA process and gave us more confidence in releasing stable & working products. --- ### Fun with (Feature) Flags URL: https://www.taboola.com/engineering/fun-with-feature-flags/ Last Modified: 2025-01-14 14:37:30 Writing features as added chunks into an ever growing one bulk of code is unorganized and messy. Overtime, the tasks of testing new behaviors becomes harder and harder. Why is that? ## Chunky Code is hard to Navigate When working with developers on a new feature, we have to identify what we need to test. As the code grows, the challenge of finding the parts that were modified and should be tested is increasing. If a feature becomes problematic or irrelevant, reverting it becomes more difficult, since we need to go back to every line of code we changed. This way we endanger production environment and are affecting our end-users. Last but not least, it is irritating to impossible to manage when dealing with “legacy” code, that needs to be reverse engineered to find how to control. Coming to compare a buggy behavior with its intended fix or testing a new feature, we need to have two abilities - reproducing the issue and verifying the fix. It becomes even more challenging when the change in the code is based on an existing feature that was overridden or overwritten. It may require to deploy two versions of the product, an old and a new one, to have a clean comparison of the tests. ## Recalculating Route - Feature Flags to the Rescue To avoid building a lot of product versions for QA environment, we need a way to switch features on and off. Why is that so important? In such a way, the change in behavior, can be attributed to the feature being switched on. This way we can be sure the fix was correctly implemented. A great way to deal with this issue is feature flags. Let me explain why... It lets you breakdown your product into a set of controlled features you can simply switch on and off. You can set parameters between possible values in a simple input fashion, giving the user the ability to enable or disable new features. It gives us a way to be more flexible and easily tweak our behaviors. With feature flags, we can provide a sense of isolation to each product feature, and an ability to “divide and conquer” each feature. In a sense, it segments the code into seemingly structured individual components. A good rule of thumb is to set up a new flag for each feature, which will allow us to turn it on or off. This helps the QA (or any other user) to turn the feature on and off appropriately when coming to perform the necessary tests in order to verify the feature before publishing it. It also provides us a safe way to AB test the feature as an individual change. After testing if the feature is compatible with the required behavior, the flag can now be cancelled and the input value we tested can become the default setting. Another great advantage is that now we can easily remove the feature without risking production environment, by simply turning the feature off, if we found it does not work as expected or harm our users. One of the greatest profits from using feature flags is developing features based on customer requests. Building the feature with a configurable flag allows us to set the feature with one behavior for this customer, and will provide us with a safe way to set it up differently for other customers. We can also easily turn it off for the customer, if they were not satisfied with the final result. A great example is our flag for auto-playing video ads. As Taboola's video content platform, we want the publisher to have the ability to decide if they want the video to start playing as soon as the video ad unit comes into view by the user, or if playing is delayed, waiting until the user clicks it. Enabling autoplay can benefit the publisher by getting higher viewability rate, but we want the publisher to have the ability to decide on the user’s video experience. ## “Fun With Flags” saved us from ourselves New features are being developed all the time, each feature with its own unique control flag. Its makes our tests easier, but how you can control so many flags? Overtime, we have developed over 1,000 flags. We can’t expect our developers or QA testers to remember all of them by heart. We spend a lot of time naming our flags. The name may be a long phrase, describing exactly what the feature does, or an abbreviation of the description. Most importantly, flags should be well documented. In order to make our lives even easier, we needed a way to be able to search for a specific flag without constantly asking the developers or looking for old tickets. Using a flags bank, or “Fun with Flags” as we call it, can come in handy. We developed a web page listing all our flags by product version. Our system is an automated data bank tool that scans the code and maps and aggregates flags together for our video platform. For each flag, we provide its full name, description, value type and default value. Now you don’t have to remember any flag, you can simply type any keyword and find the relevant flag you need to use. This flag bank lets us search for flags for certain features. We can easily find a flag we want to retest, or a flag we want to play with in order to see different aspects of the product. We can see how it behaves on different platforms and if the feature “plays” well with other features. We can monitor different combinations of flags and see how they behave together on QA and production environments, and choose the winning ones. We have a lot of benefits from our “Fun with Flags” system. We can check the frequency of a feature use, from live data, to monitor when a flag was last changed or used. If a flag wasn’t changed in a long time, maybe it should be deprecated, since it is no longer relevant, and the value can be set as a standard hard coded value. Another benefit is when mentoring new QA teammates. Enabling them swim alone in the deep water with the sharks, like the big boys, is quicker. Letting them use the “Fun with Flags” system on their own reduces the time they waist, by asking questions or waiting for help from more experienced teammates. It is much simpler and faster if they learn the ropes of video configurations on their own using “Fun with Flags”. ## The Grass is Now Greener with our Fun with Flags The struggle of controlling features is the never ending battle of QA. We always want to be able to reproduce different product behaviors on various locations. Being able to control behaviors was not easy in the past, when all the settings were hard coded.We started using feature flags to set features’ enablement and separation with a simple flick of a switch. Before our “Fun with Flags” system was developed, QA testers had to chase and wait for each feature developer to modify the code each time to compare a bug to its fix. That could take from hours, when the developer was busy, to days, when the developer is on vacation. After we started using feature flags, times were shortened considerably, but could still take a long time For each feature, we should approach the developer, or someone else from the same team, and ask them which flag we should configure. When “Fun with Flags” system was developed, we QA testers were able to lookup the needed flags with ease, within seconds. It allowed us to work independently and individually, without bothering developers or wasting precious time. “Fun with Flags” saved us from chasing our tails, going in circles, and we now run smoothly in a straight line ahead, always looking forward. --- ### Android: Autonomous Library Initialisation Using ContentProvider URL: https://www.taboola.com/engineering/android-autonomous-library-initialisation-using-content-provider/ Last Modified: 2025-01-14 14:37:30 This post describes how to use Android ContentProvider to allow automatic system initialisation for your library, therefore help make your library easier to integrate and control its flow. While this article mostly demonstrates one use case, you can use the idea in other cases as well. ## What is it all about? Always Strive To Simplify Integration Let’s assume you are writing code for a software library that would be used by an Android application. Most common flows require the app using your library to manually call the initialisation of your library and usually, provide it with their own Context. This will require your client to write a code along these lines: This article suggests using a Content Provider to allow: - Completely autonomous initialisation, liberating you from having to ask the client for init at all. - Avoiding the necessity of asking your client for their Application Context. Thus the above code line will transform into this: * Note how your initialisation is now completely independent of publisher. The flow starts in your class and is under your control. So, What's Next? ### Technical Step by Step: 1. Create a ContentProvider extending class: 2. Add an empty implementation for all required abstract methods of ContentProvider: 3. Override the ContentProvider’s onCreate() to retrieve an Application Context instance: 4. In your AndroidManifest.xml add a “provider” declaration, this provider tag will be merged with the host application manifest: * Note: If you use ${applicationId} as the value for authorities key, the system will replace it with the package name of the application hosting your library. Attributes used: - Authorities: Authority helps the ContentResolver system to find your ContentProvider instance correctly before passing data to it. - Exported: If set to true then the ContentProvider is available for other applications to use. - Enabled: If set to true then the operating system can instantiate the ContentProvider. ## A Common Pitfall (and how to avoid it) Registering multiple providers under the same authority can certainly be a serious issue. The Problem: Let’s say multiple applications installed on the same device, all embed the DemoSDK in this example. If more than one of those have not properly set their applicationId, then the value declared in the AndroidManifest.xml will fallback to using DemoSDK’s package name. If this happens then multiple ContentProviders will be registered under the same authority, therefore causing issues with the operating system. Solution: If you want to avoid this, in your ContentProvider extending class, then pay special care to the method ‘attachInfo(...)’: Notice: - You can get system information on your ContentProvider (ProviderInfo) from the AttachInfo callback. - To catch the case where the value returned for ${applicationId} in your AndroidManifest.xml request the providerInfo.authority String and compare it to your own ContentProvider’s simple name. ## Conclusion Keep in mind: If you opt for using the idea of ContentProvider for your projects initialisation, then keep in mind that your code flow will be started by the operating system and not by direct user code. Therefore, this means you might have to adjust how you think of / where you position your code initialisation with respect to the rest of your or your client’s code. How Firebase does this: If you’re interested in comparing information against a source you know well, then feel free to hop on to Google Firebase’s blog: https://firebase.googleblog.com/2016/12/how-does-firebase-initialize-on-android.html Official relevant documentation: - First of all read here: https://developer.android.com/reference/android/content/ContentProvider - Another useful page: https://developer.android.com/guide/topics/manifest/provider-element#auth To read a sample project in Kotlin: If you prefer working with Kotlin, then here’s a sample project that shows this concept along some other code: https://github.com/florent37/ApplicationProvider  --- ### Going Old-School: Designing Algorithms for Fast Weighted Sampling in Production URL: https://www.taboola.com/engineering/going-old-school-designing-algorithms-fast-weighted-sampling-production/ Last Modified: 2025-01-14 14:37:30 If you happen to write code for a living, there's a pretty good chance you've found yourself explaining another interviewer again how to reverse a linked list or how to tell if a string contains only digits. Usually, the necessity of this B.Sc. material ends once a contract is signed, as most of these low-level questions are dealt with for us under-the-hood of modern coding languages and external libraries. Still, not long ago we found ourselves facing one such question in real-life: find an efficient algorithm for real-time weighted sampling. As naive as it might seem at first sight, we'd like to show you why it's actually not - and then walk you through how we solved it, just in case you'll run into something similar. So buckle up, we've got some statistics and integrals coming up next! ## Why We Need Weighted Sampling in Production? At Taboola, our core business is to personalize the online advertising experience of millions of users worldwide. Thousands of websites across the globe trust us to display each visiting user the ads that he or she will most likely relate to, and most likely to click and engage with. We do that by training several deep-learning-based models which predict the CTR (click-through rate) of each ad for each user. The process of predicting CTR and displaying the highest rated items is known as Exploitation, as we exploit the model's predictions. But exploitation is not sufficient for a longterm successful model - we need to allow it to do some Exploration of new possibilities too, in order to find better ads. One of our ideas for such exploration was as following: ask the model to predict the CTR of a list of ads we would like to display, and then instead of displaying the highest rated items, randomly sample items for that list using weighted sampling. This is sometimes known as Soft-Exploration: the highest rated items are still the most probable ones, but every item has some non-zero probability of being shown. So, we need to do weighted sampling. The most naive approach to do so will be something like this: results = empty list max_r = 1 for i from 0 to amount of given weights: ..r = Random number in the range (0,max_r] ..s = 0 ..for each weight in the list of weights: ....s += weight ....if s => r: ......append weight to results ......remove weight from weights ......max_r = max_r - weight ......break return results This naive algorithm has a complexity of . Considering the fact that the lists are long and all this is happening in real-time, this algorithm is a no-go. But there has to be a better way to do this, right? Well, yes, but we had to design it ourselves. ## The Theoretical Solution Looking hard enough for an algorithm yielded a paper named Weighted Random Sampling by Efraimidis & Spirakis. A single line in this paper gave a simple algorithm to what we should do (page 2, A-Res algorithm, line 2): 1) map each number in the list:  ..(r is a random number, chosen uniformly and independently for each number) 2) reorder the numbers according to the mapped values ..(first number will be the one which has the largest mapped-value, and so on) This algorithm involves mapping and sorting, making it , way better than , but there's still one issue - the authors never proved it. And since we had no proof this is actually working, we had to prove it ourselves. Brace yourselves, integrals are coming.   Before we begin, as the following lines can be easily confusing, let me briefly describe what I'm about to do: - I will first describe how a weighted-sampling probability-distribution should behave. As this is what we're eventually looking for, formalizing it mathematically is probably a good idea. - Once we formalized the distribution we want, we will find a specific distribution we can use for weighted sampling. - Lastly, after finding a specific distribution, I'll link it to the Uniform Distribution, (just like the algorithm above). We'll be amazed by the fact that the suggested mapping  really does imitate random sampling. Ready? Let's go. ### Part I: Formalizing How does weighted sampling behave? Let's say we have two numbers,  and , which we perform weighted sampling over. We'd expect to get the sequence (2,1) two-thirds of the time, and the sequence (1,2) a third of the time. So we expect  to be the first number 66.6% of the times and the second 33.3% of the times. Another way to look at this, is that since we're sorting the numbers in a list, we'd expect the priority (how close a number is to the head of the list) of  to be the highest two-thirds of the times, and the lowest one-third of the times. You can easily see that priority, which we'll denote as m, behaves in a way like an inverse-index, meaning the highest m is the first one on the list. We'll prefer it over the index for two reasons: first, the priority increases as w increases, and it's more intuitive than the index, which decreases as w increases. Second, the absolute values of the priorities are not relevant; it doesn't matter if () equal to (4.5, 3) or (-1, -5) or (1024, 5). All that matters is the order between them - the highest will be first, then the second-highest and so on. These two characteristics will allow us to generalize better later on. So to wrap this example up, in the case of and , we would like to find a probability distribution which will yield  which obey: Let's generalize this and formalize it mathematically: for every two numbers , we would like to have two random variables which originate from a probability distribution (meaning: ), where is a probability distribution defined by all w values provided (in this simple example there are only two, and , but generally there could be more). We expect with probability . ### Part II: Finding a Specific Distribution I claim that the probability distribution defined by the Cumulative Distribution Function (CDF)  obeys the requirement above - and I'll prove it. For this, remember that the Probability Density Function (PDF)  obeys  , and therefore in our case: . I'll also denote the Indicator Function as  (which means is 1 when and 0 otherwise). Finally, we'll work only on the range : So we've proved that the distribution with CDF  indeed imitates weighted sampling. ### Part III: Linking to Uniform Distribution As programmers, the Uniform Distribution is usually the most accessible one we have, regardless of language or libraries. It will only make sense to link the custom-made distribution we just found to the Uniform Distribution, which will then allow us to use the latter for weighted sampling. Say some X is yielded from (that is, ), what is the probability X is smaller than some number ? This is given by the CDF: Let’s examine another variable, Y, which we’ll define as , when R originates from the Uniform Distribution . What is the probability that Y is smaller than ? Let’s calculate, remembering that the CDF of  for any  is : This is the same result we got for X which was sampled from , and this means we can sample a number from , take its wth root, and it would be just as if we used all along. This means that the priority m of a number w is given by . Neat. ## Really Making It Work There's a saying I like which states that the difference between theory and practice is that theory only works in theory. So we found a fast-enough algorithm, proved it mathematically, and of course it doesn't work. Why? Because computers. Let's take a look at our m values again: . As mentioned before, we use our models to predict CTR, and so w = CTR, which is always a number in the range of , and usually very small. As r is also sampled from the same range, becomes very small, as and . It actually becomes so small and so often, that the computer doesn't handle the precision very well, and we get zeros for all values. Let's see an example using Python: >>> from random import random >>> m = lambda w: random() ** (1.0/w) >>> m(0.002), m(0.001) (0.0, 0.0) Not good.     So what can we do? Yet again, math to the rescue! As we've previously said, we don't really care about the absolute values of m, but only about the order between them. This means we can apply any monotonic function on the mapped values and control how fast the numbers decrease. For us, a simple logarithm did the trick: . Let's see this in action: >>> from random import random >>> from math import log >>> m = lambda w: log(random())/w >>> m(0.002), m(0.001) (-402.5850708710584, -758.9101125141252) Much better. Still, this doesn't come without a price tag - the logarithm we apply decreases the accuracy of the algorithm. This means that in our example of  and , we won't get  with probability 2/3, but something close. For us though, this deviation is something we're fine with. So, to wrap this up, our random-weighted sampling algorithm for our real-time production services is: 1) map each number in the list: ..(r is a random number, chosen uniformly and independently for each number) 2) reorder the numbers according to the mapped values ..(first number will be the one which has the largest mapped-value, and so on) Success. ## Final Words Summing this process up, we've started with a naive algorithm which wasn't efficient enough, moved on to the exact opposite - an efficient algorithm which doesn't work, and then modified it to an almost-exact version which works great and is also efficient. If I need to conclude, I can only say this - there's something super exciting about stepping down from our daily routine of developing state-of-the-art AI models and return to our roots as algorithm developers; going back to the basics, develop mathematical proofs, sleeping by the river under starry skies and cooking dinner by the fire - we don't get to this every day, and I think we're all glad we did it this time. So wherever you may surf online, know that we just made your experience a little better using plain ol' math.     --- ### Deep Multi-Task Learning - 3 Lessons Learned URL: https://www.taboola.com/engineering/deep-multi-task-learning-3-lessons-learned/ Last Modified: 2025-01-14 14:37:30 For the past year, my team and I have been working on a personalized user experience in the Taboola feed. We used Multi-Task Learning (MTL) to predict multiple Key Performance Indicators (KPIs) on the same set of input features, and implemented a Deep Learning (DL) model in TensorFlow to do so. Back when we started, MTL seemed way more complicated to us than it does now, so I wanted to share some of the lessons learned. There are already quite a few posts about implementing MTL in a DL model (1, 2, 3). In this post I will share some specific points to consider when implementing MTL in a Neural Network (NN). I will also present simple TensorFlow solutions to overcome the discussed issues. ## Sharing is caring We wanted to start with the basic approach of hard parameter sharing. Hard sharing means we have a shared subnet, followed by task-specific subnets. An easy way to start playing with such a model in TensorFlow is using Estimators with multiple heads. Since it doesn’t look that different than other NN architectures, you may ask yourself what could go wrong? ## Lesson 1 - Combining losses The first challenge we encountered with our MTL model, was defining a single loss function for multiple tasks. While a single task has a well defined loss function, with multiple tasks come multiple losses. The first thing we tried was simply to sum the different losses. Soon enough we could see that while one task converges to good results, the others look pretty bad. When taking a closer look, we could easily see why. The losses’ scales were so different, that one task dominated the overall loss, while the rest of the tasks didn’t have a chance to affect the learning process of the shared layers. A quick fix was replacing the sum of losses with a weighted sum, that brought all losses to approximately the same scale. However, this solution involves another hyperparameter that might need to be tuned every once in a while. Luckily, we found a great paper proposing to use uncertainty to weigh losses in MTL. The way it is done, is by learning another noise parameter that is integrated in the loss function for each task. This allows having multiple tasks, possibly regression and classification, and bringing all losses to the same scale. Now we could go back to simply summing our losses. Not only did we get better results than with a weighted sum, we could forget about the additional weights hyperparameters. Here is a Keras implementation provided by the authors of the paper. ## Lesson 2 - Tuning learning rates It’s a common convention that learning rate is one of the most important hyperparameters for tuning neural networks. So we tried tuning, and found a learning rate that looked really good for task A, and another one that was really good for task B. Choosing the higher rate caused dying Relu’s on one of the tasks, while using the lower one brought a slow convergence on the other task. Then what could we do? We could tune a separate learning rate for each of the “heads” (task-specific subnets), and another rate for the shared subnet. Though it may sound complicated, it’s actually pretty simple. Usually when training a NN in TensorFlow you use something like: optimizer = tf.train.AdamOptimizer(learning_rate).minimize(loss) AdamOptimizer defines how gradients should be applied, and minimize computes and applies them. We can replace minimize with our own implementation that would use the appropriate learning rate for each variable in our computational graph when applying the gradients: all_variables = shared_vars + a_vars + b_vars all_gradients = tf.gradients(loss, all_variables) shared_subnet_gradients = all_gradients a_gradients = all_gradients b_gradients = all_gradients shared_subnet_optimizer = tf.train.AdamOptimizer(shared_learning_rate) a_optimizer = tf.train.AdamOptimizer(a_learning_rate) b_optimizer = tf.train.AdamOptimizer(b_learning_rate) train_shared_op = shared_subnet_optimizer.apply_gradients(zip(shared_subnet_gradients, shared_vars)) train_a_op = a_optimizer.apply_gradients(zip(a_gradients, a_vars)) train_b_op = b_optimizer.apply_gradients(zip(b_gradients, b_vars)) train_op = tf.group(train_shared_op, train_a_op, train_b_op) By the way, this trick can actually also be useful for single-task networks. ## Lesson 3 - Using estimates as features Once we’re past the first phase of creating a NN that predicts multiple tasks, we might want to use our estimate for one task as a feature to another. In the forward-pass that’s really easy. The estimate is a Tensor, so we can wire it just like any other layer’s output. But what happens in backprop? Say the estimate for task A is passed as a feature to task B. We probably wouldn’t want to propagate the gradients from task B back to task A, as we already have a label for A. Don’t worry, TensorFlow’s API has tf.stop_gradient just for that reason. When computing the gradients, it lets you pass a list of Tensors you wish to treat as constants, which is exactly what we need. all_gradients = tf.gradients(loss, all_variables, stop_gradients=stop_tensors) Again, this is useful in MTL networks, but not only. This technique can be used whenever you want to compute a value with TensorFlow, and need to pretend that the value was a constant. For example, when training Generative Adversarial Networks (GANs), you don’t want to backprop through the generation process of the adversarial example. ## So, what’s next? Our models are up and running and Taboola feed is being personalized. However, there is still a lot of room for improvement, and lots of interesting architectures to explore. In our use case, predicting multiple tasks also means we make a decision based on multiple KPIs. That can be a bit more tricky than using a single KPI... but that’s already a whole new topic. Thanks for reading, I hope you found this post useful! --- ### My first date with my company URL: https://www.taboola.com/engineering/my-first-date-with-my-company/ Last Modified: 2025-01-14 14:37:30 # My first date with my company – or – how onboarding looks from a freshman’s eye According to LinkedIn, one in three employees decide to quit their job within the first 6 months(!) I've been managing people for over 20 years and I've spent a long time trying to crack the code of successful onboarding. It was only recently, when I started working for a new company, that my eyes were opened - I actually felt what it's like to be a new employee. The lessons I've learned surprised me so much. So, I took it upon myself to build an onboarding plan addressing exactly what a new employee needs. We started to run this program in Taboola and I’m happy to say it gets great feedback. In this blog post, I'll shed some light on the psychology of a new employee, give practical ways to deal with it and share stories from my experience. If you have new members in your team this blog post is for you. See through a new employee’s eyes and understand what the ingredients are that can turn onboarding to your company into a success. ## My first day tale Though I tried it many times, I found it very hard to tailor an outstanding process for receiving a new employee. You know why? Because you cannot really know it until you feel it. You cannot really feel it until you experience it, and you cannot experience it twice… Yep. Like on a first date, there is only one shot… as they say “there is no second chance for a first impression”. Two years ago I was searching for a new job. I joined a successful company. I was very excited to join and I arrived with a lot of expectations and hopes. And… it happened to me! I experienced that notorious first day in its full strength. I learned that things I once considered as important are in fact marginal, and things I wasn’t even aware of turned to shape my entire experience. ## “Be Comfortable” Here’s the first tip: Nothing will be remembered from the first day excepting the emotional experience. That’s it! That’s the objective of day one: feeling comfortable. There is one huge problem though: the first day at work is anything but comfortable. It is a scary unpleasant experience. Think of yourself as a new kid joining a class. Everyone else knows each other, there are social codes in the class you do not know, all faces are strangers, and even the building is unfamiliar. It is counter intuitive. Most places people aim to impress new arrivals with fancy administration, but none of it matters to the new employee on their first day. To create the best first impression, aim to make the new arrival feel comfortable. I suggest three factors that can help creating that emotional comfort: The people This is undoubtedly number one: friendly people is the key. As a new employee, I know no one. That’s a very unpleasant feeling. Like newborn babies that smile at their parents I yearn for familiar faces. Surprisingly, creating familiar faces among strangers is not a complex mission at all... Here are the my best activities from my first day: someone from IT helped me 3 or 4 times (I found myself waiting for her in the afternoon, just because I “knew” her from the morning), a coffee break with a colleague I already knew, a discussion with the friend who referred me to the company, and talking to the people who interviewed me. Simple, isn’t it? The company A big question runs through a new arrival’s mind: “Where have I arrived?” Don’t count on the new people to search for information about the company in the internet. Spend time talking about the company. But while the talk is important remember that most chances nothing will stick. Prepare follow up reading material which can be viewed again later. Small Successes Even during the first day, it is possible to generate a sense of success. There is no need for big achievements. So a list of many small wins will help building self-confidence, satisfaction and comfort. Creating a checklist of small simple activities to strikethrough (e.g. login to laptop, configure your email, connect the phone to the company’s contact list, submit a certain form, intro talk to manager, etc.) can help turn the anxiety of the first day into a joy of success. ## Turn the administrative process into an emotional experience Emotional administration?! Is it a real thing? Normal people do not react emotionally when they encounter processes, do they? But the first day is all about positive emotional experience... The good news is that there is an effective way to turn the tedious inevitable tasks into a satisfying experience. Here's the key: Look at all these activities as opportunities to interact with people, learn about the company and achieve small wins. Filling out forms as a social activity There are many formalities to complete on the first day. If someone fills out the forms together with the new arrival, it becomes a personal touch point. If there are several employees that are starting on the same day, it is an excellent opportunity to gather them in the same room, create a small “community” around them, and let them help each other. Having everything ready in advance is impressive. But it is much more impressive to know that there is someone that will not leave the room until I am all set. Handout a checklist of daily activities. Thus, every form that the new employee fills out is another line to strikethrough, a little success, and another haven of positive experience in this stressful day. The welcoming laptop Like the forms, configuring the laptop is a task that can be done with friends. When I got into my inbox I was happy to find few welcoming letters waiting for me: from my manager and some explanations about the company. Party at my desk It is very nice to have everything ready in advance. It means that people got ready for my arrival. Now imagine that my gear is waiting on my desk in the original boxes, plus some chocolates and a personal congratulation note. This turns arriving at my desk into a little party. Unboxing all that gear feels like I got a present. ## Main activities Meet the team: quality over quantity Meeting the team is the heart of the day. Don’t drag new arrivals through the office for an introductory tour with many strangers. It is much better to have a personal talk with the manager, and to remember the names of your direct team members. Food Eating is much more than consuming food – it is a social activity. Do not underestimate the importance of eating together with the new employee. Two highlights of my first day were the lunch and the coffee break, because they made me feel related. Use these opportunities to interact with familiar faces: the team members, the recruiter, or people known from the past. Never leave a new employee to eat alone! Getting ready for day two There’s a huge difference between day one and day two. As a new employee, on the second day I can come early and start reading. On my first day I do not even know where my seat is. On day one, what I needed is someone to escort me. On day two, I wanted to explore by myself. Celebrate the day’s closure Finishing such a stressful day deserves a celebration. Meeting the people from the morning (familiar faces...) for a short summary session feels great. This can be celebrated with a drink or a swag. Once this summary session finishes, send the new people home early. No matter what you do, the first day is not fun - so staying late can wait for other days. ## Epilogue The first day is an excellent opportunity to create a good first impression - but remember that nothing ends there. In my next blog post I will explain the concepts of a successful onboarding which goes beyond the first day. A period which I call “the real recruitment”. ## Recipe for the perfect first day I hope you find this post useful. It can be great if you share your own experiences and we can help the next new employees in their first day. For your convenience I brought here a checklist of my recommendations: - Remember it is an anxious day for the new employee and be empathic to it. - Create as many personal interactions with familiar faces as possible. - Prepare a checklist of quick wins. - Help as much as you can - don’t leave the new employee alone. - Prepare their desk and leave things for the new employee to unbox. - Personal talks with direct manager and closest circle of colleagues. - Go to lunch with your team. - Have a coffee break with someone (preferably someone you know). - Prepare a list of activities for the second day and beyond. - Gather again at the end of the day to summarize it (and possibly give a swag). - Send home early. --- ### ICLR 19 highlights (and all that Jazz) URL: https://www.taboola.com/engineering/iclr-19-highlights-jazz/ Last Modified: 2025-01-14 14:37:31 A joint post with Ofri Mann We went to ICLR to present our work on debugging ML models using uncertainty and attention. Between cocktail parties and jazz shows in the wonderful New Orleans (can we do all conferences in NOLA please?) we also saw a lot of interesting talks and posters. Below are our main takeaways from the conference. ## Main themes A good summary of the themes was in Ian Goodfellow’s talk, in which he said that until around 2013 the ML community was focused on making ML work. Now that it’s working on many different applications given enough data, the focus has shifted towards adding more capabilities to our models: we want them to comply to some fairness, accountability and transparency constraints, to be robust, use labels efficiently, adapt to different domains and so on. ##### A slide on ML topics, from Ian Goodfellow’s talk We noticed a few topics that got a lot of attention: Reinforcement Learning, GANs and adversarial examples, and Fairness were the most prominent ones. There was also a fair amount of work on training and optimization techniques, VAEs, quantization and understanding different behaviors of networks. While the last day of the conference displayed many papers about different NLP tasks, Computer Vision felt to us like a solved problem and there were very few applicative vision papers. However, most of the work on optimization, understanding networks, meta learning etc is in fact done on images. Even though the conference focuses on learning representations, there wasn't much work on learning representations for structured or tabular data. The notes below are grouped according to the main topics. ## Reinforcement Learning There were three workshops and an entire poster session (~90 papers) dedicated to Reinforcement Learning. One of the prominent subjects discussed this year was sparse-reward or no-reward reinforcement learning. The subject was well represented with the Task-Agnostic Reinforcement Learning workshop, multiple posters and short talks, as well as a great talk by Pierre-Yves Oudeyer, discussing curiosity-driven learning, where information-gain is used as an intrinsic reward, allowing algorithms to explore their environments and focus on areas where information-gain is highest. These kinds of intrinsic rewards can later be combined with extrinsic ones to achieve tasks with almost no explicit practice. In his talk, Oudeyer demonstrated how robots learn to manipulate their environment and perform tasks with no prior practice, as well as a novel math-teaching method, where a curiosity-driven algorithm is used to personalize each child's curriculum, focusing on his/her weaker subjects. Two other workshops dealt with structure in RL: The Structure & Priors in Reinforcement Learning workshop focused on different ways to learn structure and introduce priors to RL tasks in order to allow them to learn from fewer examples and have a higher generalization power at a lower computational cost. The Deep Reinforcement Learning Meets Structured Prediction workshop suggested viewing structure prediction as a sequential decision making process and focused on papers that leverage the advances in deep RL to improve structured prediction. ## Generative Adversarial Models and Adversarial Examples Besides Goodfellow’s keynote talk, which focused on GAN’s applications to different ML research areas, there was an entire poster session dedicated to GANs and adversarial examples. These also got a lot of attention in the Debugging ML models workshop, the Deep Generative Models for Highly Structured Data workshop and in the Safe Machine Learning: Specification, Robustness, and Assurance workshop. For example, in his talk “A new Perspective on Adversarial Examples”, Aleksander Madry claimed that models learn from robust features, which capture some interpretable meaning to humans, and non-robust features that are correlative with the labels. Since the non-robust features are just as good in order to maximize model’s accuracy, models focus on them as well. Therefore, he claims, we cannot hope for post-training interpretability, since some of the model’s predictive power comes from the non-robust features. We call those features adversarial examples, but from the model’s perspective, there isn’t a real difference between the robust and non-robust features. In order to improve interpretability Madry presented ways to force models to focus only on robust features during training. ## Fairness, Accountability, Safety and AI for Social Good This year there was a lot of (not necessarily technical) talk about issues related to fairness, safety and AI for Social Good. Cynthia Dwark gave a great keynote about recent development in algorithmic fairness in which she presented two notions of fairness - group fairness, which address the relative treatment of different demographic groups, and individual fairness, which requires that people who are similar, with respect to a given classification task, should be treated similarly by classifiers for that task. She talked about the challenges related to the two notions and about recent work aiming at bridging the gap between them, with an emphasis on the notion of multi calibration - being calibrated on different groups, and fair ranking and scoring. Another talk that aimed at sparking discussion about possible dangers in ML was Zeynep Tufekci’s “While we’re all worried about the failures of ML, what dangers lurk if it (mostly) works?”. The AI for social good workshop focused on applying ML to solve problems important for society such as health care, education and agriculture. The Safe ML workshop, focused on specifying systems purpose, making them more robust and monitor them. The reproducibility in ML workshop and the Debugging ML workshop (where we presented our work on uncertanity and attention), mainly focused on fairness and interpretability. We read the high emphasis on Fairness, Accountability and Transparency (FAT) by the conference organizers as an open invitation to the community to participate in this important discussion as part of the main ML conferences. While most related papers are published in designated conferences like FAT conference and there weren’t many related papers in the main track this year, we expect this to change in upcoming years. ## Other things we liked There were many other interesting papers on topics outside the themes mentioned above. It’s hard to choose just a few out of the 500+ papers published in ICLR 19. We compiled a few for some topics we are interested in. ### Transfer Learning A research group from the University of Illinois presented a very interesting paper, "Knowledge Flow - Improve upon your teachers". The idea is a novel method of transfer learning - "Knowledge Flow" architecture. A "student" network, training on a novel task, uses the intermediate layer weights from one or more "teacher" networks, pre-trained on other tasks. By using trainable transformation matrices and weight vectors, the model learns how much each teacher contributes to the students in different parts of the network. In addition, the Knowledge Flow loss function includes the student's dependency on its teachers, which decreases as training progresses. Using this method, the teachers contribute heavily to the initial training steps, allowing for a quick "warm start" of the student network. As the student trains on the novel task, the teacher's influence is diminished, until the final training steps, when the student becomes independent of its teachers, and uses only its own weight. This works on both supervised and reinforcement-learning tasks, and achieves top results and faster convergence on several test-sets. Another transfer learning paper we found interesting was K for the price of 1: Parameter efficient multi task and transfer learning. It shows a unified framework for both transfer and multitask learning. The idea is to use model patches - a set of small trainable layers specific to each task that are interleaved with existing layers. These patches can be used for transfer learning by fine-tuning only the patch parameters on the new task while making the rest of the pretrained layers unchanged. For multitask learning it means that the models share most of the weights except for the task specific patches. ### Uncertainty (Our favourite subject!) Bias-reduced uncertainty estimation for Deep Neural Networks, shows an “overfitting”-like phenomena with uncertainty estimations where on easier cases uncertainty estimation improve at the beginning of the training but after a number of iterations they degrade. In order to prevent this, averaging different checkpoints from training or early stopping is recommended. Modeling Uncertainty with Hedge Instance Embedding: What if instead of learning deterministic embeddings we’ll learn probabilistic ones? Siamese networks are trained with constructive loss but instead of outputting z as usual, the output is expectation and standard deviation, defining the gaussian that represent the embedding. The size of the gaussian can later be treated as an uncertainty estimation. ### Why Batch Normalization Work? There we three papers worth mentioning shedding light into how batch normalization works: Towards Understanding, Regularization in Batch Normalization , A Mean Field Theory of Batch Normalization , and Theoretical Analysis of Auto Rate-Tuning by Batch Normalization ## Some thoughts on attendance The audience was a mix of academia and industry labs. A huge amount of papers came from industry research centers - DeepMind had around 50 papers, Google around 55, Microsoft and Facebook around 30 each. There was also a fair amount of work coming from industry companies. We had a lot of fun and came back with many ideas we can't wait to develop! --- ### How Intellij Tricked Me To Think Mockito Is Broken URL: https://www.taboola.com/engineering/intellij-tricked-think-mockito-broken/ Last Modified: 2025-01-14 14:37:31 ### My Little DOH! Moment Did you ever debug a piece of code, and said to yourself: “this shouldn’t happen!”? Well, me too, so when it happened to me and I learned why it happened, I thought it’s worth sharing. In my project I have two classes, let’s for simplicity call them Bouncer & EntranceChecker, Now I know that Bouncer is basically EntranceChecker, but bare with me, it’s not really important to the story. Bouncer holds EntranceChecker and every time a Person tried to enter, Bouncer calls EntranceChecker’s isInTheGuestList method, and if the person in the list Bouncer return not to block, else it blocks. public class Bouncer { private final EntranceChecker entranceChecker; public Bouncer(EntranceChecker entranceChecker) { this.entranceChecker = entranceChecker; } public boolean shouldBlock(Person person) { return !entranceChecker.isInTheGuestList(person); } } public class EntranceChecker { private final Set<String> names; public EntranceChecker(Set<String> names) { this.names = names; } boolean isInTheGuestList(Person person) { if(person == null) { return true; } return names.contains(person.getName()); } } Like any good programmer I wrote Bouncer a unit test to check it works. I used Mockito to Mock EntranceChecker. When it receives a Person that is in the list (“mockInListPerson”) it should return true, and otherwise false (“mockNotInListPerson”). Looks simple enough, right? should pass both tests, right? Want to guess what happened when I tried to run it? @RunWith(MockitoJUnitRunner.class) public class BouncerTest { @Mock private EntranceChecker mockEntranceChecker; @Mock private Person mockInListPerson; @Mock private Person mockNotInListPerson; private Bouncer bouncer; @Before public init() { Mockito.when(mockEntranceChecker.isInTheGuestList(mockInListPerson)).thenReturn(true); Mockito.when(mockEntranceChecker.isInTheGuestList(mockNotInListPerson)).thenReturn(false); this.bouncer = new Bouncer(mockEntranceChecker); } @Test public boolean shouldBlock_nullPerson_notBlock() { boolean result = bouncer.shouldBlock(mockInListPerson); Assert.assertTrue(result); } @Test public boolean shouldBlock_nullPerson_block() { boolean result = bouncer.shouldBlock(mockNotInListPerson); Assert.assertTrue(result); } } Got this: When I tried to debug it, I saw that it happens when the init() code runs and it tried to execute the line below: It enters(!) the EntranceChecker’s isInTheGuestList method, and throws NPE on the line below. It turned out The NPE was thrown because names wasn’t initialized. Which is true, I didn’t initialize it, it’s shouldn’t have been running in the first place! it’s mocked! that’s the all idea of mock! Haa I have a bug in Mockito!!! it’s broken! I broke Mockito! After way more hours than I am proud to say, I calmed down and started to think rationally again, and figured it out. EntranceChecker, was simply missing a small little public in front of isInTheGuestList. public class EntranceChecker { private final Set<String> names; public EntranceChecker(Set<String> names) { this.names = names; } public boolean isInTheGuestList(Person person) { if(person == null) { return true; } return names.contains(person.getName()); } } The moment I did it, Mockito came back to life and managed to mock my method. ### Why Was This The Problem? I don’t want to go into too many details on how mockito works, especially when Reinhard Seiler explains this so good in his blog: explanation how proxy based mock work, I strongly recommend you read it if you are writing tests and want to know how the tools you use work. I’ll just say that when you ask Mockito to mock a class for you, it basically wraps it in a wrapper class and answers what you ask of him. The problem is, that it cannot access private method and package private methods fall under that same restriction. Once I turned the method from package private methods to public, Mockito could “see” it again and mock it. when it failed to see it, the line of code simply tried to call the real method and crashed. When I went back to Intellij and looked at the metod, I remembered why I did it in the first place. Intellij suggested it to me. Doing that, didn’t break any thing and so I assumed I was OK. But Mockito is not part of that package, and so when I asked Mockito to mock the class it was blind to that method and failed. It’s a weird behavior on the part of Intellij not to be able to detect this. When I tried to pass Mockito a private method, it knew to prevent me from doing it: isTheGuestList(Person) has private access in EntranceChecker ### Bottom Line: I find Intellij suggest most helpful and try to accept them whenever I can, but this time it was the thing that gave me an headache, for more hours than I feel comfortable admitting. I can offer excuses why it took me so long to find this simple thing: - I just upgraded the Intellij version and I was worried it might have broken the mockito annotation or something. so I reverted it back, just to find out it wasn’t the issue. - I tried to rebuild the all project — more then once — and run other tests with similar code. - It was very late, and I was tired. Bottom line, after I found it, and felt really stupid that I didn’t think about it in the first place. Because I am a strong believer in the practice of sharing one stupidity with others, I decided to make it as public as possible. In the hope that, you found my stupidity funny enough that you share it with your friend and colleagues. And maybe one day you or one of your colleagues will face similar issue, have a vague memory that you already seen this problem, fix it in seconds and go on in your merry way. Thanks for reading --- ### How to conduct a good reference check? URL: https://www.taboola.com/engineering/conduct-good-reference-check/ Last Modified: 2025-01-14 14:37:31 Every candidate we recruit goes through a long process of evaluation. Near the end of the process, after we decided they fit our culture and have the skills we need, we have a reference check. Sometimes we take it as a formal phase in the process just to make sure they’re not a serial killer. Actually a reference check is one of the more important stages in the process, let me explain why. Think for a second of a recruiter that is going to recruit someone who worked with you. You know more about this person than any process. If they could peek inside your head - they will get all the knowledge they need - much more knowledge than they got from their process. Now, while you’re still in the place of the referee, think about how will you actually answer to a reference check. Most of the time, you will try to present a better picture than how it really was to work with this person. Once you’ve become a referee, you will get more than one call. After a few of those you will get tired of this process, and will try to make those calls as short as possible. So, back to the recruiter position, what should you do to avoid such a scenario? a scenario in which the reference check didn’t help you, or even worse, made you hear what you want to hear instead of the parts you need to hear. So how do we get the referee buy-in? First, make sure the referee is available and explain the rationale behind this call. For example, explain to them the importance of this call, explain that you are doing this call since you are positive about hiring the candidate. Check what is the position of the referee. What was the relationship between them and the candidate. Use this call also to validate their cultural fit and to understand how you can make their onboarding process as smooth as possible. Eventually we’re not saying if the candidate is good or not, we’re checking their fit to our organization. Ask the referee to describe the person and their work experience with them in their own words. Listen carefully and try to find out what they really think about the candidate - not just what they say, but also how they say it (with confidence / enthusiasm / motivation to give this feedback, how detailed they are). Ask for examples that demonstrate their feedback - you might interpret those examples differently. Make sure you get both pros and cons. Ask closed questions like: - Why did they leave? - Were they fired? - What are the candidate’s weaknesses? - How would you position them in the team (top /middle /bottom)? - Would you hire them again? - What they need to improve according to the last feedback you gave them? - How long did you work together? Try to clear things that you’re not sure about that came up during the process. If you’re not getting anything valuable from the referee ask the candidate for a different one. (for example, because they didn’t work really close or something like that). In case of a doubt - make sure the referee is real (you can ask them details about what they do, you can check them in LinkedIn). Last but not least - be pleasant. The person you’re talking to is an ambassador for the company. You don’t know when your paths will cross and you want to leave a good impression. So next time you’re doing a reference check, put yourself in their place, come prepared and make sure you remember this. It is the most important stage in your process. Good luck! --- ### Bucket the shuffle out of here! URL: https://www.taboola.com/engineering/bucket-the-shuffle-out-of-here/ Last Modified: 2026-01-08 10:30:25 ## Intro At Taboola we use Spark extensively throughout the pipeline. Regularly faced with Spark-related scalability challenges, we look for optimisations in order to squeeze the most out of the library. Often, the problems we encounter are related to shuffles. In this post we will present a technique we discovered which gave us up to 8x boost in performance for jobs with huge data shuffles. ## Shuffles Shuffling is a process of redistributing data across partitions (aka repartitioning) that may or may not cause moving data across JVM processes or even over the wire (between executors on separate machines).Shuffles, despite their drawbacks, are sometimes inevitable. In our case, here are some of the problems we faced: - Performance hit - Jobs run longer because shuffles use network and IO resources intensively. - Cluster stability - Heavy shuffles fill scratch disks of cluster machines. This affects other jobs on the same cluster , since they all share the same worker nodes. A single heavy job can fill-up all scratch(aka temp) disk space on all machines causing cluster unavailability. ## Where it meets us One of our many Spark jobs is to calculate some long term counters. We hold the following data: (group key, counter key, counter type) as key and (counter value, update time) as value. This job prepares input for other downstream jobs (ML, analytics etc) which makes it important for the overall progress of daily computations. We have “base” data to which we add daily aggregated changes. The daily deltas per key take up around 3% of the “base” data in terms of volume (several hundreds of gigabytes vs several terabytes). The naive approach in this scenario would be (pseudo-sql): select sum(counter_value), max(update_timestamp) from (select * from base union select * from all-deltas) group by group_key, counter_type, counter_key We started with the naive approach, it worked great and wasn't hard to implement, enabling us to deploy downstream jobs as well. However, as time passed the "base" data grew considerably. At some point the job started to fail. The job was failing with disk out of space errors. After looking at Spark UI stats we noticed huge shuffles of base data filling up all scratch disks on the application executors. At first we tried to increase the number of “spark.sql.shuffle.partitions”. Increasing the number of partitions means also reducing the size of each shuffle part. We also tried to rewrite a bit sql by pre-aggregating data by parts. Neither of the two ideas helped. The first approach might help when some executors work harder than others, but we didn’t have any skew in data. The second approach didn't help since the "base" data still needed to be reshuffled which caused the same problem. ## Partitioning of data After all previous failures, we defined our prime objective: avoid shuffles of our "base" data. Usually, this happens if spark knows how the data is partitioned. For example, joining two RDDs when only one of them has a partitioner. In this case Spark will reshuffle the second rdd using the partitioner of the first rdd. Our first approach was to partition by key while saving the dataframe (in our case the key is a compound one). However, the partitionBy method creates a subdirectory for every partition. This is effective when partition key cardinality is low and usable when the cardinality is finite. In our case the group_key has very large cardinality (millions of distinct values), which makes this approach infeasible. An example of how partitioning is used at saving: resultDf.write() .mode(SaveMode.Overwrite) .partitionBy("group_key", "counter_type", "counter_key") .parquet(outputPath) Our next idea was to use a relatively new feature called Bucketing. This feature provides kind of “partitioning by key” the same way as the standard partitioning provides. The change here is that it lets you divide the partitions into finite number of buckets(i.e. good also for high cardinality keys). However before explaining how to use it, we should talk about another optimisation we came across while working on bucketing: sql refactoring. ## SQL refactoring We have noticed that our base data is already aggregated and every key appears only once. Since our “base” data has a unique key, it doesn't need to be aggregated. On the other hand, all daily deltas may have multiple updates for the same key. As such the daily deltas need to be aggregated as described in the naive approach. Instead of using UNION on the two datasets and then aggregating both of them, we can split this process into two stages: at the beginning pre-aggregating daily deltas into a dataset that contains every key once, and then joining with “base” data by full outer join to compute updated value for every key. Replacing aggregation with a Full Outer Join (pseudo-sql): with daily_input_agg as ( select sum(counter_value), max(update_timestamp) from union (all - deltas) -- not on base anymore! group by group_key, counter_type, counter_key) -- sort + repartition) SELECT COALESCE(b.group_key, d.group_key) AS group_key, COALESCE(b.counter_key, d.counter_key) AS counter_key, COALESCE(b.counter_type, d.counter_type) AS counter_type, COALESCE(b.counter_value, 0) + COALESCE(d.counter_value, 0) AS counter_value, COALESCE(d.update_time, b.update_time, CAST (0 AS Timestamp)) AS update_time FROM base b FULL OUTER JOIN daily_input_agg d ON b.group_key = d.group_key AND b.counter_type = d.counter_type AND b.counter_key = d.counter_key Using the full outer join we can finally apply the bucketing technique to avoid the base data shuffle. ## Bucketing Bucketing is a concept that came from Hive. When using spark for computations over Hive tables, the below manual implementation might be irrelevant and cumbersome. However, we are still not using Hive and needed to overcome all gotchas along the way. This is a relatively new feature and as you will see it comes with lots of potential pitfalls. Schematically when using bucketing you'll follow several steps: - Save the data bucketed - Read data bucketed Maybe in a different SparkContext - Providing the number of buckets Writing example: resultDf.write() .option("path",outputPath) .mode(SaveMode.Overwrite) .bucketBy(numberOfOutputBuckets, "group_key", "counter_type", "counter_key") .sortBy("group_key", "counter_type", "counter_key") .saveAsTable(outputTableName); Sorted data produces fewer files - Without sorting, spark will create many files per each partition for some bucket. With sorting by same key as bucketing key only one file per partition per bucket will be created. Read the same way it was bucketed - During bucketing Spark uses hash function + modulo on the bucketing key to choose to which bucket to write data to. It's very important to preserve the same number of buckets between the reads and writes of the data. Both sides of the JOIN need to be “aligned” - in case the deltas are not partitioned with the same number of partitions both parts of the JOIN will be reshuffled. So an obvious solution in our case is to partition the “daily” data the same way we read the “base” data. daily_agg_repartitioned = daily_agg.repartition(N, col("group_key"), col("counter_type"), col("counter_key") ) No one API to bucket them all - bucketing of data is not available for the usual dataframe api, but only when using the table api. In addition, to read data while preserving bucketing information one should use table api as well. We read data by defining external table backed up by hdfs path, while specifying that the table is bucketed. example: sparkSession.sql( "CREATE TABLE base(" + " group_key string ," + " counter_type string, " + " counter_key string ," + " counter_value double," + " update_time timestamp ) " + " USING PARQUET\n" + " CLUSTERED BY (group_key, counter_type, counter_key) INTO " + numberOfInputBuckets + " BUCKETS " + " LOCATION '" + path + "'"); Partitions same as buckets - another voodoo notion we came across is that spark.sql.shuffle.partitions must be same as number of buckets, otherwise we once again get "base" shuffle. ## Results So did this help our cause? Yes. The job now finishes x8 times faster than before and cluster stability is restored. As you can see in the green boxed Spark Sql execution plan, the right branch is missing an Exchange (i.e. shuffle). When trying to optimize a spark job, use sql tab to understand if your changes helped or not(the picture attached are on test data and doesn't represent production workloads, however the implementation of removing "base" shuffles has not changed). Overall, bucketing is a relatively new technique that in some cases might be a great improvement both in stability and performance. However we found that using it is not trivial and has many gotchas. --- ### TensorFlow - The Scope of Software Engineering URL: https://www.taboola.com/engineering/tensorflow-the-scope-of-software-engineering/ Last Modified: 2025-01-14 14:37:31 ## How to structure your TensorFlow graph like a software engineer So you’ve finished training your model, and it’s time to get some insights as to what it has learned. You decide which tensor should be interesting, and go look for it in your code - to find out what its name is. Then it hits you - you forgot to give it a name. You also forgot to wrap the logical code block with a named scope. It means you’ll have a hard time getting a reference to the tensor. It holds for python scripts as well as TensorBoard: Can you see that small red circle lost in the sea of tensors? Finding it is hard... That’s a bummer! It would have been much better if it looked more like this: That’s more like it! Each set of tensors which form a logical unit is wrapped inside a named scope. Why can’t the graph be automatically constructed in a way that resembles your code? I mean, most chances are you didn’t construct the model using a single function, did you? Your code base contains multiple functions - each forms a logical unit which deserves its own named scope! Let’s say you have a tensor x which was defined by the function f, which in turn was called by g. It means that while you were writing the code, you had this logical structure in mind: g -> f -> x. Wouldn’t it be great if the model would automatically be constructed in a way that the name of the tensor would be g/f/x ? Come to think of it, it’s pretty simple to do. All you have to do is go over all your functions and add a single line of code: def f(): with tensorflow.name_scope('f'): # define tensors So what’s wrong with that approach? - The name of the function f appears twice — both in the function declaration and as an argument to tensorflow.name_scope. Maybe next week you’ll change the name of the function to something more meaningful, let’s say foo. Unfortunately, you might forget to update the name of the scope! - You have to apply indentation to the entire body of f. While it’s not that bad, personally I don’t like having high indentation levels. Let’s say f contains a for loop which contains an if statement, which contains another for loop. Thanks to calling to tensorflow.name_scope, we’re already at an indentation level of 4! We can bypass these disadvantages using simple metaprogramming - Python’s decorators to the rescue! import re def name_scope(f): def func(*args, **kwargs): name = f.__name__', f.__name__).start():] with tensorflow.name_scope(name): return f(*args, **kwargs) return func @name_scope def foo(): # define tensors How does it work? The @ is a syntactic sugar. It’s equivalent to the following: def foo(): # define tensors foo = name_scope(foo) name_scope gets a function as an argument (f) and returns a new function (func). func creates a named scope, and then calls f. The result? All the tensors that are defined by f will be created inside a named scope. The name of the scope will be the name of the original function (“foo”) - thanks to f.__name__. One small problem is that while function names might start with “_”, tensorflow scope names can’t. This is why we have to use re. ## Why is it that important? The challenge of writing clean tensorflow code is negligible compared to the research challenge of actually making the model any good. Thus, it's easy to be tempted to just focus on the research aspects of your job. However, in the long run, it's important not to neglect the maintainability and readability of your code, including those of your graph. The decorator approach make my job a little easier, and I hope you’ll benefit from it too. Do you have other tips you'd like to share? Drop a line in the comments! --- ### Beginner’s guide for naming things in your code URL: https://www.taboola.com/engineering/beginners-guide-for-naming-things-in-your-code/ Last Modified: 2025-01-14 14:37:32 “Any fool can write code that a computer can understand. Good programmers write code that humans can understand.” Martin Fowler, 2008. Names, they are everywhere in our software. Just think of the things we name, we name our packages, classes, methods, variables, in fact us programmers do so much of it, we should probably know how to do it well. In my opinion, making the code readable is just as important as making your code work. In this post I will give you 5 tips and guidelines to choose your names in order to make your code more readable. ### 1. Reveal your intent: The name you choose should answer as many questions as possible for the reader, questions like, why it exists, what it does, and how it is used. Choosing good names takes time but saves more than it takes when the going gets tough, so take care with your names and change them when you find better names. Let’s look at an Example: int d; //elapsed time in days Judging by the comment, it looks like d is some kind of integer that stores days. Wouldn’t it be much better if d name would tell us what it’s used for: int elapsedTimeInDays; int daysSinceCreation; int daysSinceModification; int fileAgeInDays; ### ### 2. Choose the right parts of speech: Classes and variables should have noun names like User, HtmlPage, Account, and AddressParser. Methods should have verb names like postPayment, deletePage, or save. Accessors, mutators, and predicates should be named for their value and prefixed with get, set, and is. Avoid words like Manager, Processor, Data or Info, they usually don’t give much information. If you are having trouble finding a name that describes what your class does, it may indicate design smell and maybe your class does more than one thing. // Bad interface DataManagerInterface {...} // Good interface EventsAggregator {...} ### ### 3. Avoid disinformation: A misleading name can cause you and your fellow co-workers a lot of time and pain. For example, do not refer to a grouping of accounts as an accountList unless it actually is a List. If the container holding the accounts is actually a Set, it may lead to false conclusions. // Bad Set accountsList; List accountsMap;   Don’t name classes as design patterns, unless they really are design patterns. // This class is clearly not implementing the builder pattern class ReqeustBuilder { public ReqeustBuilder(HttpServletRequest request) {...} public Request getRequest() {...} }   Try to not use comments if you can express yourself in the code itself. Comments tend to rot over time, because someone may change the code and keep the comment. This can result in a misleading comment that may lead to incorrect assumptions. ### 4. The scope rule: The length of your names should depend on the scope length where they are used. For variables, use short names for short scopes and long descriptive names for long scopes. Variables that are used within short scopes are usually defined very close to where they are used. For instance, it is perfectly clear what e means here: } catch (Exception e) { e.printStackTrace(); }   However, if the variable is used within a long scope, you don’t want your readers to go up 56 lines of code just to see that p means publisher. Publisher p = Publisher.byName(data.getPublisherName()); /* . . // 56 lines of code where p is not mentioned . */ if (p.shouldSendEvents()) { For methods, it’s the other way around, use short names for public methods with large scopes: public void open() {...}   And long names for private methods with small scope: private void waitForServiceThreadToStart() {...} ### ### 5. Keep it simple If names are too clever, they will be memorable only to you. Keeping the names simple and straightforward will do your code readability a favor. Pick one word for one abstract concept and stick with it. For instance, it’s confusing to have fetch, retrieve, and get as equivalent methods of different classes. Avoid encodings such as Hungarian Notation, our IDEs today are smart enough to make that notation redundant. ### ### Wrapping up Naming is a tool to communicate with your readers, bad names prevent code from clearly communicating its intent - so choose names thoughtfully and with care. Other people, including your future self, need to understand the code to be able to make changes. Nowadays, software maintenance provides the biggest day-to-day challenge. Even if you are writing something totally new, maintaining is a task you, or your fellow co-workers, are definitely going to have to deal with, sooner or later. --- ### Preparing for the Unexpected URL: https://www.taboola.com/engineering/preparing-for-the-unexpected/ Last Modified: 2025-01-14 14:37:32 Some of the problems we tackle using machine learning involve categorical features that represent real world objects, such as words, items and categories. So what happens when at inference time we get new object values that have never been seen before? How can we prepare ourselves in advance so we can still make sense out of the input? Unseen values, also called OOV (Out of Vocabulary) values, must be handled properly. Different algorithms have different methods to deal with OOV values. Different assumptions on the categorical features should be treated differently as well. In this post, I’ll focus on the case of deep learning applied to dynamic data, where new values appear all the time. I’ll use Taboola’s recommender system as an example. Some of the inputs the model gets at inference time contain unseen values - this is common in recommender systems. Examples include: - Item id: each recommendable item gets a unique identifier. Every day thousands of new items get into the system. - Advertiser id: sponsored content is created by advertisers. The number of new daily advertisers is much smaller compared to the number of new items. Nonetheless, it’s important to handle them correctly, especially since we want to support new advertisers. So what’s the challenge with OOV values? # Learning to handle OOV values An OOV value is associated with values not seen by the model at training time. Hence, if we get an OOV value at inference time, the model won’t know what to do with it. One simple solution is to replace all the rare values with a special OOV token before training. Since all OOV values are the same from the model’s point of view, we’ll replace them with the OOV token at inference time. This solution has two positive outcomes: - The model will be exposed to the OOV token while training. In deep learning we usually embed categorical features. After training, the model will learn a meaningful embedding for all OOV values . - The risk of overfitting to the rare values will be mitigated. These values appear in a small number of examples. If we learn embeddings for these values, the model might learn to use them to explain particularities or random noise found in these specific examples. Another disaster that can result with learning these embeddings is not getting enough gradient updates propagated to them. As a consequence, the random initialization will dominate the result embeddings over the signal learned through training. Problem solved... Or is it? # Handling OOV values is hard! The model uses the item id feature to memorize different information per item, similarly to the pure collaborative filtering approach. Rare items that are injected with the OOV token can’t benefit from it, so the model performs worse on them. The interesting thing is that even if we don’t use the item id at all during training, the model still performs worse on rare items! This is because they come from a distribution different than that of the general population. They have specific characteristics - maybe they performed poorly online, which caused Taboola’s recommender system to recommend them less, and in turn - they became rare in the dataset. So why does this distribution difference matter? If we learn the OOV embedding using this special distribution, it won’t generalize to the general population. Think about it this way - every item was a new item at some point. At that point, it was injected with the OOV token. So the OOV embedding should perform well for all possible items. # Randomness is the data scientist’s best friend In order to learn the OOV embedding using the general population, we can inject the OOV token to a random set of examples from the dataset before we start the training process. But how many examples will suffice? The more we sample, the better the OOV embedding will be. But at the same time, the model will be exposed to a fewer number of non-OOV values, so the performance will degrade. How can we use lots of examples to train the OOV embedding while at the same time use the same examples to train the non-OOV embeddings? Instead of randomly injecting the OOV token before starting to train, we chose the following approach: in each epoch the model trains using all of the available values (the OOV token isn’t injected). At the end of the epoch we sample a random set of examples, inject the OOV token, and train the model once again. This way, we enjoy both worlds! As was done in the previous approach, we also inject the OOV token to rare values - to avoid overfitting. To evaluate the new approach, we injected the OOV token to all of the examples and evaluated our offline metric (MSE). It improved by 15% compared to randomly injecting the OOV token before the model starts to train. # Final thoughts Our model had been used in production for a long time before we thought of the new approach. It could have been easy to miss this potential performance gain, since the model performed well overall. It just stresses the fact that you always have to look for the unexpected! --- ### Where did my cookies go?? URL: https://www.taboola.com/engineering/where-did-my-cookies-go/ Last Modified: 2025-01-14 14:37:32 If you are using web cookies to operate your online business you probably know already that just like in real life, cookies do not last long. This is an especially known fact to whoever uses online cookies to store unique user IDs. Most online marketing companies rely on cookies for that purpose, but when cookies disappear - it makes it harder for them get persistent user data. Interested to know for how long does a cookie really last? in this post I’ll try to provide some answers. #### Who is eating web cookies? Cookies can disappear for various reasons, such as: - Clearing the browser historical data by the user - Setting the browser to reject third-party cookies - Using tools that clean up your device and free up storage space - Use of VPNs, Ad Blockers and more. One very common reason cookies disappear is the use of private browsing modes such as Incognito in chrome or private window in Safari. In most browsers browsing in a private browsing mode will cause the cookie to be deleted when the session ends - effectively turning cookies into ‘session cookies’. In other words, those cookies last for only a few hours. The next time a user will browse in private browsing mode he or she will be assigned a new cookie. #### Internet users going undercover Data we’ve gathered from billions of events around the world suggests that around 17% of global internet traffic comes from users that are only seen for one session and never seen again - much of this phenomena can be explained by usage of private browsing modes. This number seems to be inline with other independent research. This number varies a bit between countries. In UK only 14% of traffic comes from cookies that are a session long while in Germany it is 24%! However, is private browsing mode the main reason for cookie churn? #### Whose party is this cookie?! An interesting way to look at cookie churn is by drawing its survival curve and calculating its half life time. This is a common method used in nuclear physics to describe radioactive decay. Half lifetime of a cookie represents the median lifetime of a cookie - 50% of all cookies have shorter lifespan and 50% have longer lifespan. We conducted a research in taboola measuring survival rate of third-party cookies and first-party cookies. Third-party and first-party cookie are terms often used in ambiguous ways, but in general they are regarded as follows: - Third-party cookie - a cookie placed by a domain the user has not interacted directly with - First-party cookie - a cookie placed by a domain the user has interacted directly with #### The survival of a cookie In the below analysis we looked only at cookies that had a lifespan of more than 1 day - meaning we excluded sessions cookies, including private browsing modes. Our research suggests that third-party cookies are very short lived with a half life time of only 14 days! First-party cookies, on the other hand, are much more persistent; 30 days after they are created almost three-quarters still survive! #### The two cookie problems The fact cookies don’t last long is not their only shortcoming. The second problem with cookies is that they represent only a shell of the user. Most people today are using at least two devices, usually mobile phone and laptop or desktop. On each device there is a different cookie with a different user ID. Not only that, on Safari browser for example, that blocks third-party cookie by default, you might find dozens of first-party cookies with different user IDs all belonging to one person! Cookies give a fragmented representation of the user digital interactions. #### A new way to think about cookies For companies that rely mainly on web cookies to track users, I suggest the following framework to think about the persistency of their user data: longevity vs. completeness. - Longevity is how long a user identifier lasts - Completeness is how much of the user interactions a user identifier captures Logged-in users using email for log-in, for example, offers high longevity, since a user’s email lasts for years, and also a high level of completeness as the user logs in across all devices with the same email. On the other hand session cookies have very low longevity since they only lasts for few hours, and they are usually only relevant for one site so they are low on completeness. First-party cookies, as we’ve seen in the chart above, have higher longevity than third-party cookies but they do not allow the cross domain tracking third-party cookies allow, so they get lower score on completeness. Lastly there are also mobile ad IDs in the app space. They are hardly ever changed by the user so their longevity is usually the lifetime of the device, however they only capture the user interactions in apps of a single device so they are not complete like an email. #### What about YOUR cookies? Cookies pose a challenge to any company that wishes to achieve persistent user data in the web space. There are various strategies to deal with this challenge which I will save discussing for a future post. However the first step in facing this challenge is understanding its magnitude. so… do you know how long your cookies last?? --- ### Think your Data Different URL: https://www.taboola.com/engineering/think-data-different/ Last Modified: 2025-01-14 14:37:32 In the last couple of years deep learning (DL) has become a main enabler for applications in many domains such as vision, NLP, audio, click stream data etc. Recently researchers started to successfully apply deep learning methods to graph datasets in domains like social networks, recommender systems and biology, where data is inherently structured in a graphical way. So how do Graph Neural Networks work? Why do we need them? # The Premise of Deep Learning In machine learning tasks involving graphical data, we usually want to describe each node in the graph in a way that allows us to feed it into some machine learning algorithm. Without DL, one would have to manually extract features, such as the number of neighbors a node has. But this is a laborious job. This is where DL shines. It automatically exploits the structure of the graph in order to extract features for each node. These features are called embeddings. The interesting thing is, that even if you have absolutely no information about the nodes, you can still use DL to extract embeddings. The structure of the graph, that is - the connectivity patterns, hold viable information. So how can we use the structure to extract information? Can the context of each node within the graph really help us? # Learning from Context One well known algorithm that extracts information about entities using context alone is word2vec. The input to word2vec is a set of sentences, and the output is an embedding for each word. Similarly to the way text describes the context of each word via the words surrounding it, graphs describe the context of each node via neighbor nodes. While in text words appear in linear order, in graphs it’s not the case. There’s no natural order between neighbor nodes. So we can’t use word2vec... Or can we? # Reduction like a Badass Mathematician We can apply reduction from the graphical structure of our data into a linear structure such that the information encoded in the graphical structure isn’t lost. Doing so, we’ll be able to use good old word2vec. The key point is to perform random walks in the graph. Each walk starts at a random node, and performs a series of steps, where each step goes to a random neighbor. Each random walk forms a sentence that can be fed into word2vec. This algorithm is called node2vec. There are more details in the process, which you can read about in the original paper. # Case study Taboola’s content recommender system gathers lots of data, some of which can be represented in a graphical manner. Let’s inspect one type of data as a case study for using node2vec. Taboola recommends articles in a widget shown in publishers’ websites: Each article has named entities - the entities described by the title. For example, the item “the cutest dogs on the planet” contains the entities “dog” and “planet”. Each named entity can appear in many different items. We can describe this relationship using a graph in the following way: each node will be a named entity, and there will be an edge between two nodes if the two named entities appear in the same item: Now that we are able to describe our data in a graphical manner, let’s run node2vec to see what insights we can learn out of the data. You can find the working code here. After learning node embeddings, we can use them as features for a downstream task, e.g. CTR (Click Through Rate) prediction. Although it could benefit the model, it’ll be hard to understand the qualities learned by node2vec. Another option would be to cluster similar embeddings together using K-means, and color the nodes according to their associated cluster: Cool! The clusters captured by node2vec seem to be homogeneous. In other words, nodes that are close to each other in the graph are also close to each other in the embedding space. Take for instance the orange cluster - all of its named entities are related to basketball. You might wonder what is the benefit of using node2vec over classical graphical algorithms, such as community detection algorithms (e.g., the Girvan-Newman algorithm). Capturing the community each node belongs to can definitely be done using such algorithms, there’s nothing wrong with it. Actually, that’s exactly feature engineering. And we already know that DL can save you the time of carefully handcrafting such features. So why not enjoy this benefit? We should also keep in mind that node2vec learns high dimensional embeddings. These embeddings are much richer than merely community belonging. # Taking Another Approach Using node2vec in this use case might not be the first idea that comes to mind. One might suggest to simply use word2vec, where each sentence is the sequence of named entities inside a single item. In this approach we don’t treat the data as having a graphical structure. So what’s the difference between this approach - which is valid, and node2vec? If we think about it, each sentence we generate in the word2vec approach is a walk in the graph we’ve defined earlier. node2vec also defines walks on the same graph. So they are the same, right? Let’s have a look at the clusters we get by the word2vec approach: Now the “basketball” cluster is less homogenous - it contains both orange and blue nodes. The named entity “Basketball” for example was colored orange, while the basketball players “Lebron James” and “Kobe Bryant” were colored blue! But why did this happen? In this approach each walk in the graph is composed only of named entities that appear together in a single item. It means we are limited to walks that don’t go further than distance 1 from the starting node. In node2vec, we don’t have that limit. Since each approach uses a different kind of walks, the learned embeddings capture a different kind of information. To make it more concrete, consider the following example: say we have two items - one with named entities A, B, C and another with D, B, E. These items induce the following graph: In the simple word2vec approach we’ll generate the following sentences: and . In the node2vec approach we could also get sentences like . If we fetch the latter into the training process, we’ll learn that E and C are interchangeable: the prefix will be able to predict both C and E. Therefore, C and E will get similar embeddings, and will be clustered together. # Takeway Using the right data structure to represent your data is important. Each data structure implies a different learning algorithm, or in other words - introduces a different inductive bias. Identifying your data has a certain structure, so you can use the right tool for the job, might be challenging. Since so many real world datasets are naturally represented as graphs, we think Graph Neural Networks are a must-have in our tool box as data scientists. --- ### Want to improve as an engineer? Face your fears and try public speaking URL: https://www.taboola.com/engineering/want-to-improve-as-an-engineer-face-your-fears-and-try-public-speaking/ Last Modified: 2025-01-14 14:37:32 Ever thought about presenting your work to others? Talking in a meetup or a conference? In the past I couldn’t even think about it, I thought that it’s not for me and I won’t get any benefit from it at all. In the last year and a half, things have started to change. In the following post I will share how the will for continuous improvement took me out of my comfort zone, and put me in places and scenarios I never imagined. I started my journey in the software development world 8 years ago. I had some knowledge, and almost no experience. I studied industrial engineering and didn’t think I would practice software development. But things changed and I found my first role as a manual QA engineer, then QA automation engineer, automation developer, and in the last 5 years DevOps / Release engineer. I was always curious and looking for how to improve as an engineer, so I did what most of us do: read tutorials, posts and watch technical videos. Still, since I came from a different background from most engineers, I felt like I had a knowledge gap that blocked me in many ways. ## I wanted to improve, but how? It bothered me being a level behind everyone. I wanted to improve, learn and practice new tools and technologies. Then I remembered that one of the best methods I had used to understand something better at university, was to explain it to others. When you explain a solution to others, you look at it differently. To provide a good explanation you need to fill all the gaps and assumptions in the solution, so it can be easily understood. I thought about explaining to or helping others, but I was afraid of giving lectures, so I started small. I opened stackoverflow and started to read some questions and answers - then I posted some answers to questions I knew. As I progressed, I found questions that I didn’t immediately know the answers to. So first I had to find a solution and then post it online. My benefit was in a number of aspects: I improved my knowledge in those areas, my stackoverflow score started to rise, and so did my confidence. ## Getting my toes wet My next step involved a few other platforms: meetups, conferences and Twitter. I had the chance to go to the Jenkins World and DevOpsDays conferences, and met some interesting people. I also attended meetups, and strengthened my technical skills as I was exposed to new material and ideas. At that time I only had about 10 tweets in my twitter account, but I figured that most of those people that were talking at those conferences and meetups also tweet and share additional knowledge. So I jumped at the opportunity, and followed anyone who seemed interesting. From here to there, I also started tweeting. Most of my tweets were articles and videos that I had read or watched, but from time to time I also tweeted about my ideas, opinions, personal stuff. Being active on Twitter has definitely improved my confidence, which I believe makes me a better engineer....but then I encountered the following tweet: https://twitter.com/mipsytipsy/status/912103246017294336 It made me start thinking about talking in public... ## Diving into the lecturing world By a weird coincidence, I was offered to talk at the Tel Aviv Jenkins Area meetup by one of its organizers (thank you Anton!). I didn’t say yes right away, but I also didn’t say no. I had two things in my mind: continuous improvement and the tweet above. I thought about it for a while, accepted the challenge and the meetup was scheduled. My talk was about Jenkins, JVM and garbage collector. Those were topics that I knew, but to be able to explain them easily I had to drill down a lot more. I was ready, I gave a good talk (I think :) ), and I felt that I could do it again. My improvement from this talk was obvious, I understood the JVM and GC much better than before, and I could come up with new solutions for related problems at my work. It took me a while...but I gave another meetup talk, this time about software delivery processes. A short while after, I was invited to do my third talk at fiverr offices, and I did that as well. Not surprisingly, guess what happened? Each time I learned new things that I skipped before, like new aspects of the different topics, my confidence kept increasing, and now I’m looking forward to my next talks. ## The journey continues I still feel that I can and want to improve much more. As I get more experienced at my job, I also have new things to present. My plan is to do talks more frequently, and although I am still a bit afraid of talking on big stages, I believe it will happen in the near future. My comfort zone has already changed, it’s no longer my desk - so why wouldn’t I change it more and take it to the main stage? There are different ways to improve yourself as an engineer. For me talks and discussions on social networks proved to be very successful. I truly think that any engineer has something that he or she can teach others. Will you take that step and embrace the challenge? I would be happy to listen... --- ### Get real life debugging using Kibana and Elastic URL: https://www.taboola.com/engineering/get-real-life-debugging-using-kibana-and-elastic/ Last Modified: 2025-01-14 14:37:33 We all have these amazing machines in our development and testing labs, and we know that our real users do not share this wonderful world. They experience our products very differently from us. These differences result in two major challenges: We do not know what the users experience We cannot debug their machines As a Video Advertisement Player team, these challenges are multiplied. Why? Our product is a third party script that serves other third party scripts for websites. ## Your code runs on different platforms As a third party web product, you do not know which websites your code runs on. Websites have a variety of frameworks, architectures and styles. Frameworks - change the browser’s core behavior, for example, redefining methods, which challenges the product’s basic behavior. Architectures - affect the website’s performance, which impacts on the product’s natural flow. Styles -manipulate the product’s look and feel. ## Running someone else’s code When your product serves other third party scripts for websites, you neither know how they will perform, nor can you tell their impact on the website. Scripts affect website performance, behavior and style. So...what can we do? ## Elasticsearch and Kibana to the rescue Elasticsearch provides a full text search engine, with an HTTP web interface and schema free JSON documents. Kibana is an open source data visualization plugin for Elasticsearch. It provides visualization capabilities, on top of the content indexed in an Elasticsearch cluster. These are big words, how do we use them? ## Elasticsearch logging We define and log our product’s stories, and their combination creates the product funnel. Stories, are your product split into separated areas and flows. Our product’s stories are defined by the different interactions the product performs: - Interacting websites - publishers - Interacting third party script - advertisers - In-product business logic That sounds great! How does it work? ## Publishers Our publishers want to increase their revenue from video ads. Platform, machine, website performance and framework existence - this data is logged while interacting with the publisher. The Video Autoplay capability is required for an advertisement to be played on a website. Logging and debugging sessions allow us to reproduce non-played video ads. The result is a browsers’ list, which requires a non-regular solution for playing video ads. Publisher and user raw data ## Advertisers Our advertisers’ interest is in displaying noticeable ads. Loading time, malicious behavior, network load and impact on the page style - this data is logged while interacting with the advertiser. Advertisers might be trying to steal the user’s attention, and they do so by hijacking the page’s scroll and forcing sound. This allows them to increase their viewability and noticeability. Logging and debugging sessions let us know how it’s done and which methods are programmatically called. The result is implementing defenses against this kind of malicious behavior. Ad interaction logged events ## In-product business logic Our interest is to keep playing the most profitable ad whenever we can and to reduce calls to the ads server. Interacting with the ads server, ads’ life-cycle and choosing which ad to play - this data is our product’s business logic. Logging and debugging sessions reveal ads' media types which are not supported. The results are improving the mechanism that chooses the next ad to play. Ads lifecycle and metrics ## Conclusion Elasticsearch and Kibana help us analyze and reproduce real life sessions. A glimpse of this was mentioned above. Log new business logic and product’s events is all you need for maintaining your data relevance. Lacking the knowledge of your users’ experience is a liability. Thus reproducing and debugging your users’ sessions in your labs is an asset.   --- ### Fibers from out of (user) space - Hands on URL: https://www.taboola.com/engineering/fibers2/ Last Modified: 2025-01-14 14:37:33 A couple of months ago my team had its first experience working with Java fibers, we needed to make our main application work asynchronously. In this 3 part series, I will share my team’s experience and how we deploy and implement Java fibers in production. In the previous part (Part 1), we talked about what fibers are in high level, how they compare to threads and why we started to explore them. In this part we’ll focus further in-depth about fibers and how they differ from threads, we’ll see how to create fibers, how to work with them, and the basic concepts of how they work. ## Threads vs. Fibers We searched for a reason why not to stay with threads. We researched the costs and performance penalties of working with threads vs. fibers. We wanted to find proof that fibers can work better than threads, or at least shine in some areas. So we did several tests and experiments to try and prove that, mostly on performance and scale. I must say, the research did not yield conclusive results like I wished it did, but instead, it taught us a lot and enabled us to control the behavior of our service via a simple configuration flag. Eventually it really depends on your specific use-case. In our case the performance differences between threads and fibers were minor but we gained a better imperative and more clean code. ## Performance Performance measurement is always a problem to do, because it is based on the conditions of the system it is running on. But let's take a look at a standard benchmark test that does allow us to get a sense of the performance we could get by working with Fibers: The thread ring problem. In the thread ring problem, we create 500 threads (can be any other amount as well), while connecting them in a ring (circle) structure so that the last thread points to the first one. Then, serially we pass a message from one thread to the other, in a circular way, 10,000 times. There are many ways to implement this, here is one example available on GitHub: https://github.com/vy/fiber-test It uses the de-facto framework to measure nano performance on the JVM - JMH. It requires a little tweaking to run, but eventually, here are the results: Environment and plan: - Testing environment, laptop: Thinkpad X1, Core i7-7500U 2.70Ghz (4 cores), Ubuntu 1.64 - 5 Warm Up iterations, 5 executions. Results: - Java Threads: 10.646 ops/s - Fibers: 103.241 ops/s Fibers show improvement of almost x10 in this case, nice, we have a potential here ! ### perf Next thing was to put our newly refactored application under perf to figure out if we gain any improvements in metrics such as branch predictions, CPU utilization, page faults and such. If you are unfamiliar with perf , it is a Swiss-army knife Linux profiler for almost everything. Our refactored application had a flag to set whether to run with fibers or threads. The following shows the differences between the 2 runs. We issued the following command to run perf: perf stat -p 90607 -a sleep 300 Environment and plan: - Testing environment, server: Intel(R) Xeon(R) CPU E5-2630 v4 @ 2.20GHz, CentOS 7.2.1511 - 40 cores - 128gb RAM - Production traffic, +/- 300 QPS - 5 minutes sampling. Results: Method/Metric  cpu-clock (msec)  context-switches  cpu-migrations  page-faults cycles  instructions branches branch-misses  Fibers  2221610.273407  19,480,439  2,957,727  485,777  5,183,396,393,765  5,867,959,049,330  1,211,078,813,490  20,758,136,733  Threads  2076849.036113  19,361,672  3,122,642  468,374  4,826,607,131,964  5,518,275,491,102  1,141,887,811,701  20,030,907,473 The results are not conclusive. One reason for this is the behavior of our application. It suffers from long business logic decisions and methods that consume a lot of CPU and do not block enough for fibers switching. In the parts where they do block, threads are doing similar job, therefore the outcome stays similar. If we reduced the number of threads of the ForkJoinPool (see Part 3) we could have a better ruling in the favor of fibers, but we couldn't due to a large amount of CPU in our code. ## Scale Next was to try and differentiate the amount of threads vs. fibers in an application. So, to demonstrate the burden on the OS when creating thousands of threads vs. the same amount of fibers, we did the following: Lets see what happens when we try to run a small program that creates 100k threads: import java.util.concurrent.TimeUnit; public class Monster extends Thread { public void run () { while (true) { System.out.println("I am monster running on: "+Thread.currentThread().getId()); try { TimeUnit.MILLISECONDS.sleep(100); } catch (InterruptedException e) { e.printStackTrace(); } } public static void main(String[] args) throws Exception { for (int i=0;i<100000;i++) { Monster monster = new Monster(); monster.start(); } } } This is the output you will see when you run it: Java HotSpot(TM) 64-Bit Server VM warning: Attempt to deallocate stack guard pages failed. Exception in thread "main" java.lang.OutOfMemoryError: unable to create new native thread at java.lang.Thread.start0(Native Method) at java.lang.Thread.start(Thread.java:717) at Test1.main(Test1.java:19) Java HotSpot(TM) 64-Bit Server VM warning: Attempt to deallocate stack guard pages failed. Java HotSpot(TM) 64-Bit Server VM warning: INFO: os::commit_memory(0x00007f6493563000, 12288, 0) failed; error='Cannot allocate memory' (errno=12) Oops, It died. It was unable to allocate and create native threads due to memory limitations. Now, let’s write a similar fiber version of this: import co.paralleluniverse.fibers.Fiber; import co.paralleluniverse.fibers.SuspendExecution; import co.paralleluniverse.strands.Strand; import co.paralleluniverse.strands.SuspendableRunnable; public class Monster implements SuspendableRunnable { public void run() throws SuspendExecution, InterruptedException { while (true) { System.out.println("I am monster running on: "+Thread.currentThread().getId()); try { Strand.sleep(100); } catch (InterruptedException e) { e.printStackTrace(); } } } public static void main(String[] args) throws Exception { for (int i=0;i<100000;i++) { new Fiber<Void>(new Monster()).start(); } } } Running the above works like a charm :) Now that we got the sense of what fibers are, in terms of performance and scale, let’s see how to create/work with them... ## Hello world - how do we create a simple fiber? In order to simplify our lives, fibers are implemented in a very similar fashion to Java threads. They have a functional interface, are Runnable like, and are launched/destroyed in the same way. In fact, their implementation in Java is such that both implement a shared class named Strand, that is an abstraction of both a thread and a fiber. This means you can design the system based on Strands, and decide by configuration whether to run on fibers or normal Java threads. At the moment, fibers are used as an external library with intentions to make them part of the JVM (see: Project Loom). Fibers require the code to be instrumented - instrumentation is a method used to inject (patch) bytecode instructions on top of existing classes that Java produces when it compiles sources to bytecode. There are 2 ways to do it: - Via dynamic instrumentation, by adding a -javaagent parameter to the VM parameters - Using static instrumentation, by building the bytecode with instrumentation using build tools such as Ant/Maven So first, to add support for fibers in your project , add the following to your Maven pom.xml: <dependency> <groupId>co.paralleluniverse</groupId> <artifactId>quasar-core</artifactId> <version>0.7.10</version> </dependency> <dependency> <groupId>co.paralleluniverse</groupId> <artifactId>quasar-actors</artifactId> <version>0.7.10</version> </dependency> <dependency> <groupId>co.paralleluniverse</groupId> <artifactId>quasar-galaxy</artifactId> <version>0.7.10</version> </dependency> <dependency> <groupId>co.paralleluniverse</groupId> <artifactId>quasar-reactive-streams</artifactId> <version>0.7.10</version> </dependency> In order to make methods in our code “fiber friendly”, we need to annotate them with the @Suspendable annotation or declare them to throw SuspendExecution. This will tell Quasar what our interruption points are, so that instrumentation will be active. Let’s write a simple fiber that runs a single fiber and calls 2 methods that print some output and sleep: import co.paralleluniverse.fibers.Fiber; import co.paralleluniverse.fibers.SuspendExecution; import co.paralleluniverse.fibers.Suspendable; import co.paralleluniverse.strands.Strand; import java.util.concurrent.TimeUnit; public class Demo1 extends Fiber<Void> { @Suspendable void method1 () throws SuspendExecution, InterruptedException { System.out.println("Hello from method1, run count: "+Fiber.currentFiber().getCurrentRun()); Fiber.sleep(100, TimeUnit.MILLISECONDS); } @Suspendable void method2 () throws SuspendExecution, InterruptedException { System.out.println("Hello from method2, run count: "+Fiber.currentFiber().getCurrentRun()); Fiber.sleep(100, TimeUnit.MILLISECONDS); } @Override protected Void run() throws SuspendExecution, InterruptedException { System.out.println("Running fiber: "+Fiber.currentFiber().getName()); method1(); method2(); return super.run(); } public static void main(String[] args) throws Exception { new Demo1().start(); } } To run it we use the following command: java -classpath . -javaagent:quasar-core-0.7.10.jar -Dco.paralleluniverse.fibers.detectRunawayFibers=false -Dco.paralleluniverse.fibers.verifyInstrumentation=false Demo1 Output: Running fiber: fiber-10000001 Hello from method1, run count: 1 Hello from method2, run count: 2 Process finished with exit code 0 We can see that interruption occurred between method1 and method2, due to the fact that the runCount increased by one on the second print, which implies that a fiber branch selection had occurred. What went on? Why was the run count increased? Here is a step by step tracing: - Fiber is first launched and started. - run method is running. - method1 is being called (run count = 1) - Fiber.sleep is being called, the fiber stops - The fiber scheduler is running and re-schedules this fiber. - run is running again - Instrumentation now jumps to method2 - method2 is being called (run count = 2) - Fiber.sleep is being called, the fiber stops - The fiber scheduler is running and re-schedules this fiber. - Instrumentation now jumps to after method2 - Fiber ends. ## What’s next In this part we got a hold on how to create simple fibers, how they shine in performance and scale (in some scenarios), and how they can help us. In the next part, we’ll deep dive into the structure of fibers, how they work behind the scenes in order to better understand them, and what we learnt when implementing them in production. Continue reading the next part .... Part III - Deeper view Or go to the previous part ... Part I - Overview Please enable JavaScript to view the comments powered by Disqus. Method/Metric  cpu-clock (msec)  context-switches  cpu-migrations  page-faults cycles  instructions branches branch-misses  Fibers  2221610.273407  19,480,439  2,957,727  485,777  5,183,396,393,765  5,867,959,049,330  1,211,078,813,490  20,758,136,733  Threads  2076849.036113  19,361,672  3,122,642  468,374  4,826,607,131,964  5,518,275,491,102  1,141,887,811,701  20,030,907,473  Method/Metric  cpu-clock (msec)  context-switches   cpu-migrations  page-faults  cycles  instructions  branches branch-misses  Fibers  2221610.273407   19,480,439  2,957,727  485,777  5,183,396,393,765  5,867,959,049,330  1,211,078,813,490  20,758,136,733  Threads  2076849.036113   19,361,672  3,122,642  468,374  4,826,607,131,964  5,518,275,491,102 1,141,887,811,701  20,030,907,473 --- ### Fibers from out of (user) space - Deeper view URL: https://www.taboola.com/engineering/fibers3/ Last Modified: 2025-01-14 14:37:33 A couple of months ago my team had its first experience working with Java fibers, we needed to make our main application work asynchronously. In this 3 part series, I will share my team’s experience and how we deploy and implement Java fibers in production. In Part 1 we talked about what fibers are in high level, how they compare to threads and why we started to explore them. In Part 2 we went further in-depth about how fibers differ from threads, how to create fibers, how to work with them and the basic concepts of how they work. In this part, we’ll discuss what's going on under the hood in fibers and deep dive into the implementation of how fibers work and what lessons we learnt during our journey working with them. We will also see how this magic happens... ## Under the hood Fibers are implemented by instrumenting our JVM bytecode instructions, and patching them in order to save and resume state. In order to instrument our code, we have to run our application with Quasar Java agent in the command line: -javaagent:quasar-core-0.7.10.jar The framework uses a ForkJoinPool executor, to run the fibers on a set of limited amount of threads (defaulted to number of cores, but configurable). ForkJoinPool is a form of thread pool that follows the concepts of divide and conquer, and works by stealing algorithms to best utilize the CPU and thread contention costs of the normal ThreadPoolExecutor. It mostly shines in use cases where tasks can be divided to subtasks, such as running fibers code blocks. Internally, it uses internal task queues for each worker thread that lowers the contention on the main executor task queue. Also, it uses a dequeue to lower the synchronization on adding items to the queue and fetching them from the tail in a stack order. The stealing part comes where a worker has ended its job and immediately goes and fetches work from other workers queues. The main benefit of this is to fully utilize the usage of the worker threads in the job cycle. ForkJoinPool is not necessarily better than ThreadPoolExecutor, it depends on the use case, since it has more overhead. If there is a known amount of work that can be evenly distributed, ThreadPoolExecutor is better if work can be divided and re-submitted and should be re-assembled, ForkJoinPool may be a better choice. What does it look like behind the scenes? Let’s take this simple fiber as an example: Fiber f1 = new Fiber(new SuspendableRunnable() { public void run() throws SuspendExecution, InterruptedException { while (true) { System.out.println(Thread.currentThread().getId()+" 1"); Fiber.sleep(100); System.out.println(Thread.currentThread().getId()+" 2"); } } }).start(); Time to dive in and inspect some bytecode. The above snippet translates into the following bytecode: GETSTATIC java/lang/System.out : Ljava/io/PrintStream; NEW java/lang/StringBuilder DUP INVOKESPECIAL java/lang/StringBuilder.<init> ()V INVOKESTATIC java/lang/Thread.currentThread ()Ljava/lang/Thread; INVOKEVIRTUAL java/lang/Thread.getId ()J INVOKEVIRTUAL java/lang/StringBuilder.append (J)Ljava/lang/StringBuilder; LDC " 1" INVOKEVIRTUAL java/lang/StringBuilder.append (Ljava/lang/String;)Ljava/lang/StringBuilder; INVOKEVIRTUAL java/lang/StringBuilder.toString ()Ljava/lang/String; INVOKEVIRTUAL java/io/PrintStream.println (Ljava/lang/String;)V LDC 100 INVOKESTATIC co/paralleluniverse/fibers/Fiber.sleep (J)V GETSTATIC java/lang/System.out : Ljava/io/PrintStream; NEW java/lang/StringBuilder DUP INVOKESPECIAL java/lang/StringBuilder.<init> ()V INVOKESTATIC java/lang/Thread.currentThread ()Ljava/lang/Thread; INVOKEVIRTUAL java/lang/Thread.getId ()J INVOKEVIRTUAL java/lang/StringBuilder.append (J)Ljava/lang/StringBuilder; LDC " 2" INVOKEVIRTUAL java/lang/StringBuilder.append (Ljava/lang/String;)Ljava/lang/StringBuilder; INVOKEVIRTUAL java/lang/StringBuilder.toString ()Ljava/lang/String; INVOKEVIRTUAL java/io/PrintStream.println (Ljava/lang/String;)V LDC 100 INVOKESTATIC co/paralleluniverse/fibers/Fiber.sleep (J)V Now, let’s have a look what the Quasar agent does to that code bytecode after instrumentation: @Lco/paralleluniverse/fibers/Instrumented;(methodOptimized=false, methodStart=39, methodEnd=42, suspendableCallSites={40, 42}, suspendableCallSiteNames={"co/paralleluniverse/fibers/Fiber.sleep(J)V"}, suspendableCallSitesOffsetsAfterInstr={106, 164}) TRYCATCHBLOCK L0 L1 L2 co/paralleluniverse/fibers/SuspendExecution TRYCATCHBLOCK L0 L1 L2 co/paralleluniverse/fibers/RuntimeSuspendExecution TRYCATCHBLOCK L0 L1 L1 null ACONST_NULL ASTORE 3 INVOKESTATIC co/paralleluniverse/fibers/Stack.getStack ()Lco/paralleluniverse/fibers/Stack; DUP ASTORE 1 IFNULL L0 ALOAD 1 ICONST_1 ISTORE 2 INVOKEVIRTUAL co/paralleluniverse/fibers/Stack.nextMethodEntry ()I TABLESWITCH 1: L3 2: L4 default: L5 L5 FRAME FULL [] ALOAD 1 INVOKEVIRTUAL co/paralleluniverse/fibers/Stack.isFirstInStackOrPushed ()Z IFNE L0 ACONST_NULL ASTORE 1 L0 FRAME FULL [] ICONST_0 ISTORE 2 L6 FRAME FULL [] GETSTATIC java/lang/System.out : Ljava/io/PrintStream; NEW java/lang/StringBuilder DUP INVOKESPECIAL java/lang/StringBuilder.<init> ()V INVOKESTATIC java/lang/Thread.currentThread ()Ljava/lang/Thread; INVOKEVIRTUAL java/lang/Thread.getId ()J INVOKEVIRTUAL java/lang/StringBuilder.append (J)Ljava/lang/StringBuilder; LDC " 1" INVOKEVIRTUAL java/lang/StringBuilder.append (Ljava/lang/String;)Ljava/lang/StringBuilder; INVOKEVIRTUAL java/lang/StringBuilder.toString ()Ljava/lang/String; INVOKEVIRTUAL java/io/PrintStream.println (Ljava/lang/String;)V LDC 100 ALOAD 1 IFNULL L7 ALOAD 1 ICONST_1 ICONST_1 INVOKEVIRTUAL co/paralleluniverse/fibers/Stack.pushMethod (II)V ALOAD 1 ICONST_0 INVOKESTATIC co/paralleluniverse/fibers/Stack.push (JLco/paralleluniverse/fibers/Stack;I)V ICONST_0 ISTORE 2 L3 FRAME FULL [] ALOAD 1 ICONST_0 INVOKEVIRTUAL co/paralleluniverse/fibers/Stack.getLong (I)J L7 FRAME FULL INVOKESTATIC co/paralleluniverse/fibers/Fiber.sleep (J)V GETSTATIC java/lang/System.out : Ljava/io/PrintStream; NEW java/lang/StringBuilder DUP INVOKESPECIAL java/lang/StringBuilder.<init> ()V INVOKESTATIC java/lang/Thread.currentThread ()Ljava/lang/Thread; INVOKEVIRTUAL java/lang/Thread.getId ()J INVOKEVIRTUAL java/lang/StringBuilder.append (J)Ljava/lang/StringBuilder; LDC " 2" INVOKEVIRTUAL java/lang/StringBuilder.append (Ljava/lang/String;)Ljava/lang/StringBuilder; INVOKEVIRTUAL java/lang/StringBuilder.toString ()Ljava/lang/String; INVOKEVIRTUAL java/io/PrintStream.println (Ljava/lang/String;)V LDC 100 ALOAD 1 IFNULL L8 ALOAD 1 ICONST_2 ICONST_1 INVOKEVIRTUAL co/paralleluniverse/fibers/Stack.pushMethod (II)V ALOAD 1 ICONST_0 INVOKESTATIC co/paralleluniverse/fibers/Stack.push (JLco/paralleluniverse/fibers/Stack;I)V ICONST_0 ISTORE 2 L4 FRAME FULL [] ALOAD 1 ICONST_0 INVOKEVIRTUAL co/paralleluniverse/fibers/Stack.getLong (I)J L8 FRAME FULL INVOKESTATIC co/paralleluniverse/fibers/Fiber.sleep (J)V GOTO L6 L1 FRAME FULL ALOAD 1 IFNULL L2 ALOAD 1 ICONST_0 INVOKEVIRTUAL co/paralleluniverse/fibers/Stack.popMethod (I)V L2 FRAME FULL ATHROW After instrumentation, we take the resulting bytecode and decompile it, in order to make it more readable: @Instrumented static final class QuasarTest1$2 implements SuspendableRunnable { @Instrumented(methodOptimized = false, methodStart = 46, methodEnd = 49, suspendableCallSites = { 47, 49 }, suspendableCallSiteNames = { "co/paralleluniverse/fibers/Fiber.sleep(J)V" }, suspendableCallSitesOffsetsAfterInstr = { 106, 164 }) public void run() throws SuspendExecution, InterruptedException { Label_0158: { Stack stack; if ((stack = Stack.getStack()) != null) { switch (stack.nextMethodEntry()) { default: { if (!stack.isFirstInStackOrPushed()) { stack = null; break; } break; } case 1: { break Label_0159; } case 2: { break Label_0160; } } } try { while (true) { System.out.println(Thread.currentThread().getId() + " 1"); final long n = 100L; if (stack != null) { stack.pushMethod(1, 1); Stack.push(n, stack, 0); stack.getLong(0); } Fiber.sleep(n); Label_0159: System.out.println(Thread.currentThread().getId() + " 2"); final long n2 = 100L; if (stack != null) { stack.pushMethod(2, 1); Stack.push(n2, stack, 0); stack.getLong(0); } Fiber.sleep(n2); Label_0160: } } catch (SuspendExecution suspendExecution) {} catch (RuntimeSuspendExecution runtimeSuspendExecution) { throw runtimeSuspendExecution; } finally { if (loadexception(java.lang.Throwable.class) == null) { throw; } stack.popMethod(0); } } } } We can see that the decompilation gives us hints of what’s going on, even though it’s a best effort decompilation and not accurate. Analyzing the bytecode, points out that there is a decision at the beginning of the method to make a jump into injected labels in the code according to the state we are in. TABLESWITCH 1: L3 2: L4 default: L5 L5 We can see that L3 is defined near the first sleep, and L4 is defined near the second sleep. The injected push calls in the decompiled code, suggest that during the method the Quasar framework is being told where we are currently in the method, before throwing an exception to quit it. Quasar uses a special internal SuspensionException in order to escape coding blocks. Indeed, each time the code runs it is being interrupted in calls to methods that are marked as @Suspended. Quasar stores the index and run count and all the stack variables, and frames in the current position so that it can resume at the exact place after the interruption triggers. ## Lessons learned 1. Adding fibers took some time but was worth it. It was mostly built from adding @Suspendable annotations to the relevant methods in the code and stabilize the application. On the way there were few challenges, there can be external libraries or dependencies that can cause problems because they may not be 100% compatible with fibers, however, most of the time it’s possible to overcome this. The transition was quite smooth in terms of code refactoring, no major changes to code were taken except adding the @Suspendable annotations. 2. During the development phase several issues were revealed. After trying to launch a server, suddenly, strange NullPointerException exceptions started to be thrown into our log files, they appeared in places where they really shouldn’t and for no good reason. Investigation had come up with the understanding that some methods in key places were not marked as @Suspendable which caused this shady behavior to happen. Key things to remember are, first, run your application with the following VM parameter in order to verify correctness instrumentation of your code: -Dco.paralleluniverse.fibers.verifyInstrumentation=true , this will try to figure out all the places that are missing @Suspendable annotations. Be aware that this flag not always 100% detects the missing pieces. Second, it is important to annotate interfaces too, if you have a class that inherits from an interface, and you mark some method in it as @Suspendable, also mark its sibling method inside the interface definition. 3. If you are using ReentrantLocks and conditions, even though they have a corresponded fibers version it is highly advised to define your condition as: AbstractQueuedSynchronizer.ConditionObject condition = (AbstractQueuedSynchronizer.ConditionObject)locker.newCondition() Otherwise, you may get some serious strange errors due to the fact the Condition interface is not @Suspendable annotated. 4. Memory leaks pitfalls. Even though ThreadLocals are managed by fibers and transparent to them, there could be some glitches mostly in external libraries. For example, this happened with Netty. Our application is using gRPC heavily, gRPC is using netty IO underneath, the library uses a memory/bytes thread pool cache to maintain fast memory and avoid GC overhead. A ThreadDeathWatcher class expects running threads to be killed and then remove them from an internal watch list. This isn’t happening under fibers where the main ForkJoinPool engine threads are always running, which will eventually cause a memory leak from a huge ArrayList that contains many ThreadDeatchWatcher instances. 5. Define the correct amount of parallelism for the ForkJoinPool by specifying the VM parameter: -Dco.paralleluniverse.fibers.DefaultFiberPool.parallelism=XXX, by default the number of cores in the system will be used, but sometimes, that is not enough. You need to measure it and play with the numbers. 6. Monitor. Fibers expose JMX metrics, so you can monitor the ForkJoinPool queues, number of fibers running on the system and much more. You can possibly also address the default ForkJoin queues and monitor their waiting queues, for example: int poolSize = ((MonitoredForkJoinPool)DefaultFiberScheduler.getInstance().getExecutor()).getPoolSize(); long qsize = ((MonitoredForkJoinPool)DefaultFiberScheduler.getInstance().getExecutor()).getQueuedSubmissionCount(); 7. Spurious Wakeups – are known behavior in threads synchronization mechanism that states that a waiting thread in a wait state, may wake up from it’s blocking state due a spurious wakeup not necessarily fired from a notify event. The reasons for that to happen are mostly due to the fact how POSIX/Win32 system blocking calls are implemented and their sensitivity to signal processing. Because of this, LockSupport.park() / LockSupport.unpark() must be used in a loop and check a condition to verify that a real change in a condition happened and not a spurious wakeup. The same goes with Fibers, however, the story with them is a little bit different, since they run in ForkJoinPool, lags may occur between calling the park() to the time they really parks, this must be checked if unpark() is called before a fiber was parked can lead the system to a deadlock. ## Summary and conclusions I hope you got a valuable glimpse into fibers and their power. There are more technical challenges and things to improve. The library is maintained in GitHub and has active contributors. The latest version 0.8.0 is required, and only runs on JVM 11, the latest for Java 8 is 0.7.10. It is planed to be part of the JVM (see Project Loom). Moving our main bidding infrastructure to work with fibers enabled us to make our code more readable, easier to maintain, and drove us to add new asynchronous capabilities in a relatively low complexity. The costs and efforts where not huge, and the modifications to the code did not require crazy refactoring, this is because of the imperative way fibers are designed. However, you should measure everything and keep track of abnormal behavior that fibers might yield when using less compatible libraries - that will be known only after trial and error. In terms of performance when comparing to threads - well, here to be honest I had more hopes, we really liked the technology and we wished it could bring also performance improvements out of the box, sadly it wasn't so clear to us and the fight against threads could be declared as a tie. ## References JMH - https://openjdk.java.net/projects/code-tools/jmh/ Fibers performance tests - https://github.com/vy/fiber-test Quasar - http://docs.paralleluniverse.co/quasar/ Project loom - https://openjdk.java.net/projects/loom/ Vert.x - https://vertx.io/docs/vertx-sync/java/ Akka - https://akka.io/ ForkJoinPool - http://gee.cs.oswego.edu/dl/papers/fj.pdf , http://www.h-online.com/developer/features/The-fork-join-framework-in-Java-7-1762357.html Kilim - https://github.com/kilim/kilim Spurious Wakeups - https://en.wikipedia.org/wiki/Spurious_wakeup Java Threads Memory - https://www.oracle.com/technetwork/java/hotspotfaq-138619.html Go to the previous part ... ### Part II - Hands on Please enable JavaScript to view the comments powered by Disqus. --- ### The Story of a Bad Train-Test Split URL: https://www.taboola.com/engineering/story-of-bad-train-test-split/ Last Modified: 2025-01-14 14:37:33 About a year ago we incorporated a new type of feature into one of our models used for recommending content items to our users. I’m talking about the thumbnail of the content item: Up until that point we used the item’s title and metadata features. The title is easier to work with compared to the thumbnail - machine learning wise. Our model has matured and it was time to add the thumbnail to the party. This decision was the first step towards a horrible bias introduced into our train-test split procedure. Let me unfold the story... ## Setting the scene From our experience it’s hard to incorporate multiple types of features into a unified model. So we decided to take baby steps, and add the thumbnail to a model that uses only one feature - the title. There’s one thing you need to take into account when working with these two features, and that’s data leakage. When working with the title only, you can naively split your dataset into train-test randomly - after removing items with the same title. However, you can’t apply random split when you work with both the title and the thumbnail. That’s because many items share the same thumbnail or title. Stock photos are a good example for shared thumbnails across different items. Thus, a model that memorizes titles/thumbnails it encountered in the training set might have a good performance on the test set, while not doing a good job at generalization. The solution? We should split the dataset so that each thumbnail appears either in train or test, but not both. Same goes for the title. ## First attempt Well, that sounds simple. Let’s start with the simplest implementation. We’ll mark all the rows in the dataset as “train”. Then, we’ll iteratively convert rows into “test” until we get the desired split, let’s say 80%-20%. How is the conversion done? At each step of the loop we’ll pick a random “train” row and mark it for conversion. Before converting, we’ll inspect all of the rows that have the same title/thumbnail, and mark them as well. We’ll continue doing so until there are no more rows that we can mark. Finally, we’ll convert the marked group into “test”. ## And then things escalated At first sight nothing seems wrong with the naive solution. Each thumbnail/title appears either in train or in test. So what seems to be the problem? First I’ll show you the symptoms of the problem. In order to be able to compare the title-only model to the model that also uses the thumbnail, we used the new split for the title-only model too. It shouldn’t really make an impact on it’s performance, right? But then we got the following results: In the top row we see what we already know: the title-only model has higher accuracy on train set, and accuracy isn’t significantly affected by the ratio of the split. The problem pops up in the bottom row, where we apply the new split method. We expected to see similar results, but the title-only model was better on test. What?.... It shouldn’t be like that. Additionally, the performance is greatly affected by the ratio. Something is suspicious… ## So where does the problem lurk? You can think of our dataset as a bipartite graph, where one side is the thumbnails, and the other is the titles. There is an edge between a thumbnail and a title if there is an item with that thumbnail and title. What we effectively did in our new split is making sure each connected component resides in its entirety either in train or test set. It turns out that the split is biased. It tends to select big components for the test set. Say the test set should contain 15% of the rows. You’d expect it to contain 15% of the components, but what we got was 4%. ## Second try What was the problem with what we did? When you randomly sample a row, the probability of getting a row from a specific component is proportional to the component’s size. Therefore, the test set ended up with a small number of big components. It may be counterintuitive, but here’s a code snippet you can try to experience it yourself: import numpy as np import matplotlib.pyplot as plt def train_test_split(component_sizes, test_size): train = component_sizes test = [] while sum(test) < test_size: convert = np.random.choice(range(len(train)), p=train.astype('float') / sum(train)) test.append(train) train = np.delete(train, convert) return train, test component_sizes = np.array(range(1, 10000)) test_size = int(sum(component_sizes) * 0.5) train, test = train_test_split(component_sizes, test_size) plt.hist(, label=, bins=30) plt.title('Distribution of sizes of components', fontsize=20) plt.xlabel('component size', fontsize=16) plt.legend(fontsize=14) The components size distribution is different between the train and test set. Now that we formalized better what we were doing by means of bipartite graph, we can implement the split by randomly sampling connected components, instead of randomly sampling rows. Doing so, each component gets the same probability of being selected for the test set. ## Key takeaway The way you split your dataset into train-test is crucial for the research phase of a project. While researching, you spend a significant amount of your time on looking at the performance over the test set. It’s not always straightforward to construct the test set so that it’s representative of what happens at inference time. Take for example the task of recommending an item to a user: you can either recommend a completely new item or an item that has been shown to other users in the past. Both are important. In order to understand how the model is doing offline in the research phase, you’ll have to construct a test set that contains both completely new items, and items that appear in the train set. What is the right proportion? Hard to say… I guess it can be a topic for another post on another day :) --- ### Predicting Probability Distributions Using Neural Networks URL: https://www.taboola.com/engineering/predicting-probability-distributions-using-neural-networks/ Last Modified: 2025-01-14 14:37:34 If you’ve been following our tech blog lately, you might have noticed we’re using a special type of neural networks called Mixture Density Network (MDN). MDNs do not only predict the expected value of a target, but also the underlying probability distribution. This blogpost will focus on how to implement such a model using Tensorflow, from the ground up, including explanations, diagrams and a Jupyter notebook with the entire source code. ## What are MDNs and why are they useful? Real life data is noisy. While quite irritating, that noise is meaningful, as it gives a wider perspective of the origins of the data. The target value can have different levels of noise depending on the input, and this can have a major impact on our understanding of the data. This is better explained with an example. Assume the following quadratic function:   Given x as input, we have a deterministic output f(x). Now, let’s turn this function into a more interesting (and realistic) function: we’ll add some normally distributed noise to f(x). This noise will increase as x increases. We’ll call the new function g(x), which formally equals to g(x) = f(x) + ?(x), where ?(x) is a normal random variable. Let’s sample g(x) for different x values:     The purple line represents the noiseless function f(x), and we can easily see the increase in the added noise. Let’s inspect the cases where x = 1 and x = 5. For both these values, f(x) = 4, and so 4 is a reasonable value g(x) can take. Is 4.5 a reasonable prediction for g(x) as well? The answer is clearly no. While 4.5 seems to be a reasonable value for x = 5, we cannot accept it as a valid value for g(x) when x = 1. If our model simply learns to predict y’(x) = f(x), this valuable information will be lost. What we actually need here is a model capable of predicting y’(x) = g(x). And this is exactly what an MDN does. The concept of MDN was invented by Christopher Bishop in 1994. His original paper explains the concept quite well, but it dates back to the prehistoric era when neural networks weren’t as common as they are today. Therefore, it lacks the hands-on part of how to actually implement one. And so this is exactly what we’re going to do now. ## Let’s get it started Let’s start with a simple neural network which only learns f(x) from the noisy dataset. We’ll use 3 hidden dense layers, each with 12 nodes, looking something like this: We’ll use Mean Square Error as the loss function. Let's code this in Tensorflow:   x = tf.placeholder(name='x',shape=(None,1),dtype=tf.float32) layer = x for _ in range(3): layer = tf.layers.dense(inputs=layer, units=12, activation=tf.nn.tanh) output = tf.layers.dense(inputs=layer, units=1)   After training, the output looks like this:   We see the network successfully learnt f(x). Now all that’s missing is an estimation of the noise. Let’s modify the network to get this additional piece of information.   ## Going MDN We will keep on using the same network we just designed, but we'll alter two things: - The output layer will have two nodes rather than one, and we will name these mu and sigma - We’ll use a different loss function Now our network looks like this:   Let’s code it: x = tf.placeholder(name='x',shape=(None,1),dtype=tf.float32) layer = x for _ in range(3): layer = tf.layers.dense(inputs=layer, units=12, activation=tf.nn.tanh) mu = tf.layers.dense(inputs=layer, units=1) sigma = tf.layers.dense(inputs=layer, units=1,activation=lambda x: tf.nn.elu(x) + 1)   Let’s take a second to understand the activation function of sigma - remember that by definition, the standard deviation of any distribution is a non-negative number. The Exponential Linear Unit (ELU), defined as:   yields -1 as it lowest value, and so ELU+1 will always be non-negative. Does it have to be ELU? No, any function that always yields a non-negative output will do - for example, the absolute value of sigma. ELU simply seems to be doing a better job. Next, we need to adjust our loss function. Let’s try to understand what exactly we are looking for now. Our new output layer gives us the parameters of a normal distribution. This distribution should be able to describe the data generated by sampling g(x). How can we measure it? We can, for example, create a normal distribution from the output, and maximize the probability of sampling our target values from it. Mathematically speaking, we would like to maximize the values of the probability density function (PDF) of the normal distribution for our entire dataset. Equivalently, we can minimize the negative logarithm of the PDF:     We can see that the loss function is differentiable with respect to both ? and ?. You’ll be surprised by how easy it is to code: dist = tf.distributions.Normal(loc=mu, scale=sigma) loss = tf.reduce_mean(-dist.log_prob(y))   And that’s about it. After training the model, we can see that it did indeed pick up on g(x):   In the plot above, the blue lines and dots represent the actual standard deviation and mean used to generate the data, while the red lines and dots represent the same values predicted by the network for unseen x values. Great success! ## ## Next steps We’ve seen how predict a simple normal distribution - but needless to say, MDNs can be far more general - we can predict completely different distributions, or even a combination of several different distributions. All you need to do is make sure to have an output node for each parameter of the distribution’s variables, and validate that the distribution’s PDF is differentiable. I can only encourage you to try and predict even more complex distributions - use the notebook supplied with this post, and change the code, or try your new skills on real life data! May the odds be ever in your favor. --- ### How Azure Data Explorer Helped Us Make Sense of 1M Log Lines per Second URL: https://www.taboola.com/engineering/azure-data-explorer-helped-us-make-sense-1m-log-lines-per-second/ Last Modified: 2025-01-14 14:37:34 As VP of IT at Taboola, my teams and I are overwhelmed with logs, pinned down by the rate and volume of them. The job of the Production Site Reliability Engineering (SRE) team in Taboola is to keep the technology running smoothly and bring in as many insights as we can from the system, making sure that any and every technical issue (that isn't self healing or contained) is dealt with quickly. We also support this torrent of incoming data to make sure that any insights that can be gleaned from this data are found. With over 1B users discovering what’s interesting and new through the Taboola Feed, we can't drop the ball or stop thinking about our logs, log management, where to store them and how to process them. This is the challenge of processing over one million lines of logs every second. To address this challenge, we engaged with the Azure Data Explorer and after the first very simple demo, it was clear to us that this technology had something unique to it, something new to the mix of storing, indexing and exploring data. We decided to test the Azure Data Explorer to see if there is a quick win. It was clear that we need to get a full day of logs into the system, not just a sample of data to see how fast it can provide us with insights and moreover, to understand if the Azure Data Explorer can ingest the data. Once we shipped one day of logs into Azure, it was clear that the insights that we can run are far and fast reaching. Machine learning? No problem. More days of data? Easy. Scale? Done. Now we can can keep our eyes on the ball. In our case, the ball is business SLA and improving the time to resolve problems. Working with our partners, we were able to get many of our logs written directly to Azure Storage so that the ingress of data was solved without any heavy lifting. The ability to ship logs directly from the CDN (and other SaaS providers) was easily done through a configuration setting. Sending logs from our on-prem servers proved to be straightforward coding. On top of that, as ingress networking is free, cost was not an issue. The data import into Azure Data Explorer was the next phase, this is a short configuration that brings all the data in. Retention policy, and access to the data are also simple configurations. With minimal coding, and mostly configurations, it was a fast path from raw logs to data insights and clear results. Working with Azure Data Explorer we were able to get fast answers to in depth queries on the data. The dashboards made sense of a lot of data with a faster update time and an intuitive UI. Furthermore, we were able to democratize the access to the data -- anyone with a need to access the data can find what they need and make it work for them. The path to scale comes not only in the ability to store, index and report vast amounts of data, but also on the ability to provide Taboola employees fast and easy access to it. The more people working on the data, the more that is accomplished. On the cost management side, the ability to provide unrestricted queries to my users without looking at the cost of each data scan is liberating. We can now have as many employees as we need access the data and not worry about the bill for each query they run. The cost of running the system on Azure has proven to have a lower TCO than our on-prem log management system. Helping us to focus on what's important, also brought added benefits of time better spent on the business impact of the SRE teams. --- ### How We Stopped Being Afraid of 3rd Party Scripts URL: https://www.taboola.com/engineering/stopped-afraid-3rd-party-scripts/ Last Modified: 2025-01-14 14:37:34 The web is full of third-party scripts. Sites use them for ads, analytics, retargeting, and more. But this isn’t always the whole story. Scripts are unpredictable, they execute code, but you don’t know what this code actually does. With Taboola’s advertising video player solution, we struggle with 3rd party scripts daily. Working with different advertisers has exposed us to a variety of malicious behavior: sound violations, auto scroll and change page DOM are just some of them. In this post we’ll take a closer look at how we detect sound violations.   ### Steer clear of 3rd party script risks A sound violation is a state in which the video plays sound without user interaction. Several advertisers do this, to ensure that the user will notice their ads. We struggled with this often and received lots of complaints from publishers. Sound violations should be prevented by the video player, but often aren’t. The main challenge is to be the first to detect them. The passive approach is to wait for the publisher to complain. A different approach is to leave your computer and speaker on, and hope to hear sound from video. We chose the proactive approach - be the first to discover your “weak spots”. We decided to develop a framework to test and detect malicious advertisers’ behavior. Our first goal was to detect sound violations before the publisher.   ### Why Selenium is not your best friend Our first framework was based on Selenium, the most popular automation tool in the industry, which is also used in Taboola. Yet, Selenium did not suit our needs. Let me explain why... Selenium was running tests on Chrome 46, while Chrome 60 was the latest version. Also, Selenium was not intuitive - each test was a struggle. Finally, we had issues with installing Selenium on both Mac and Windows.   ### All you need is Node Our conclusion was that Selenium did not suit our needs. So, we decided to characterize and develop a testing framework on our own, and chose to base our solution on Node.js. Why Node? With the help of some Node libraries, any browser can be controlled and debug. We use chrome-remote-interface, child_process, sendmail, and more. All installed using npm and regularly maintained.   ### Be proactive - DIY The first mission was to inspect and debug the most popular browsers, CDP (Chrome DevTools Protocol) is the best solution for debugging. CDP is an API for both Chrome and Safari. It’s installed using NPM, intuitive and easy to maintain. CDP allows you to control every browser aspect - DOM inspecting, network monitoring, load extensions, are examples of CDP’s abilities. It enables you to share user experience, and improve it as much as possible. The second phase was deciding on the testing environment. Remote iOS debugging is not available on Windows, so we decided on Mac Mini as our debugging machine. We needed to install Chrome, Xcode, and a real Android mobile device. By executing command line scripts, we could control browsers on all platforms. The last step was remote launching our testing environment. We set Mac Mini to act as a TeamCity agent. We created a project on TeamCity, which runs every day and operates the agent. In the end, this is what the framework looks like:   ### Sound violation fighters Now comes the best part, we were ready to develop a test to detect sound violations. CDP does not provide info about sound playing in tab. Luckily enough, sound can be detected by Chrome extensions. Developing an extension, which listens to changes in a page’s audible state, is very easy. The extension will trigger an event when the ad’s sound is changed from mute to unmute. Here is a simple example of an extension that detects sound playing tabs: Once we were done with the test, we were able to run ~1000 web pages everyday. Playing different videos from a variety of advertisers, sampling for sound violation. When the sampling is finished, team members receive a full report about exposed sound violations.   ## The sound of silence We are able to stay one step ahead of malicious 3rd party scripts, and overcome issues before our publishers find out about them. We have since developed more tests like this, and we can track more data about advertisers’ behaviors. As a result of being proactive, we have been able to catch almost every sound violation occurring in our video player. In the beginning we caught around 10 ads in every sample. Now, we have decreased that to 1 or none per sample. Today we can also watch network activity and trace large video files, and we try to decrease page load time and CPU usage. How can everyone benefit from a proactive framework? You can use it for E2E and feature tests. You have all major browsers connected, with the ability to test them. Write some simple tests, throw in some case studies, and you have full coverage of your code using only Node. Currently we use it for E2E and sanity tests on Chrome, iOS and Android - it has reduced QA test time for new versions by half. Being proactive and developing a customized framework has made our work life much easier. We improved our user experience and relationships with partners. We are able to develop a bug-free (as much as possible, we are human after all) advertising video player. Being proactive has placed us as one of the best video players in the industry. --- ### How to Engineer Your Way Out of Slow Models URL: https://www.taboola.com/engineering/engineer-way-slow-models/ Last Modified: 2025-01-14 14:37:34 So you just finished designing that great neural network architecture of yours. It has a blazing number of 300 fully connected layers interleaved with 200 convolutional layers with 20 channels each, where the result is fed as the seed of a glorious bidirectional stacked LSTM with a pinch of attention. After training you get an accuracy of 99.99%, and you’re ready to ship it to production. But then you realize the production constraints won’t allow you to run inference using this beast. You need the inference to be done in under 200 milliseconds. In other words, you need to chop off half of the layers, give up on using convolutions, and let’s not get started about the costly LSTM… If only you could make that amazing model faster! #  Sometimes you can Here at Taboola we did it. Well, not exactly… Let me explain. One of our models has to predict CTR (Click Through Rate) of an item, or in other words - the probability the user will like an article recommendation and click on it. The model has multiple modalities as input, each goes through a different transformation. Some of them are: - categorical features: these are embedded into a dense representation - image: the pixels are passed through convolutional and fully connected layers - text: after being tokenized, the text is passed through a LSTM which is followed by self attention These processed modalities are then passed through fully connected layers in order to learn the interactions between the modalities, and finally, they are passed through a MDN layer. As you can imagine, this model is slow. We decided to insist on the predictive power of the model, instead of trimming components, and came up with an engineering solution. # Cache me if you can Let’s focus on the image component. The output of this component is a learned representation of the image. In other words, given an image, the image component outputs an embedding. The model is deterministic, so given the same image will result with the same embedding. This is costly, so we can cache it. Let me elaborate on how we implemented it. # The architecture (of the cache, not the model) - We used a Cassandra database as the cache which maps an image URL to its embedding. - The service which queries Cassandra is called EmbArk (Embedding Archive, misspelled of course). It’s a gRPC server which gets an image URL from a client and retrieves the embedding from Cassandra. On cache miss EmbArk sends an async request to embed that image. Why async? Because we need EmbArk to respond with the result as fast as it can. Given it can’t wait for the image to be embedded, it returns a special OOV (Out Of Vocabulary) embedding. - The async mechanism we chose to use is Kafka - a streaming platform used as a message queue. - The next link is KFC (Kafka Frontend Client) - a Kafka consumer we implemented to pass messages synchronously to the embedding service, and save the resulting embeddings in Cassandra. - The embedding service is called Retina. It gets an image URL from KFC, downloads it, preprocesses it, and evaluates the convolutional layers to get the final embedding. - The load balancing of all the components is done using Linkerd. - EmbArk, KFC, Retina and Linkerd run inside Docker, and they are orchestrated by Nomad. This allows us to easily scale each component as we see fit. This architecture was initially used for images. After proving its worth, we decided to use it for other components as well, such as text. EmbArk proved to be a nice solution for transfer learning too. Let’s say we believe the content of the image has a good signal for predicting CTR. Thus, a model trained for classifying the object in an image such as Inception would be valuable for our needs. We can load Inception into Retina, tell the model we intend to train that we want to use Inception embedding, and that’s it. Not only that the inference time was improved, but also the training process. This is possible only when we don’t want to train end to end, since gradients can’t backpropagate through EmbArk. So whenever you use a model in production you should use EmbArk, right? Well, not always… # Caveats There are three pretty strict assumptions here. ### 1. OOV embedding for new inputs is not a big deal It doesn’t hurt us that the first time we see an image we won’t have its embedding. In our production system it’s ok, since CTR is evaluated multiple times for the same item during a short period of time. We create lists of items we want to recommend every few minutes, so even if an item won’t make it into the list because of non optimal CTR prediction, it will in the next cycle. ### 2. The rate of new inputs is low It’s true that in Taboola we get lots of new items all the time. But relative to the number of inferences we need to perform for already known items are not that much. ### 3. Embeddings don’t change frequently Since the embeddings are cached, we count on the fact they don’t change over time. If they do, we’ll need to perform cache invalidation, and recalculate the embeddings using Retina. If this would happen a lot we would lose the advantage of the architecture. For cases such as inception or language modeling, this assumption holds, since semantics don’t change significantly over time. # Some final thoughts Sometimes using state of the art models can be problematic due to their computational demands. By caching intermediate results (embeddings) we were able to overcome this challenge, and still enjoy state of the art results. This solution isn’t right for everyone, but if the three aforementioned assumptions hold for your application, you could consider using a similar architecture. By using a microservices paradigm, other teams in the company were able to use EmbArk for needs other than CTR prediction. One team for instance used EmbArk to get image and text embeddings for detecting duplicates across different items. But I’ll leave that story for another post... --- ### How I Resolved Delays in Kafka Messages by Prioritizing Kafka Topics URL: https://www.taboola.com/engineering/resolved-delays-kafka-messages-prioritizing-kafka-topics/ Last Modified: 2025-01-14 14:37:34 As a team member in the Scale Performance Data group of Taboola’s R&D, I had the opportunity to develop a mechanism which prioritizes the consumption of Kafka topics. I did so to tackle a challenge we had of handling messages that are being sent from hundreds of frontend servers around the world by our time-series based backend servers. In this post, I will focus on three things. - Why we needed such a mechanism - the problem - Code snippets showing how I put said mechanism in place - the solution - Issues I faced with Kafka - bumpers on the way to the solution Prerequisites Assume you have basic knowledge in: - Java multithreading multithreading in Java - Kafka ### Data coming from all over the world needs to be divided by minutes Taboola has a few hundred frontend servers. They serve more than half a million requests per minute – and this number grows steadily. The frontend servers send data to our backend servers using Kafka, this data is aggregated and analyzed by our backend servers. The backend servers need to have the data arranged by minutes before they can analyze it. ### Things might get complicated when dealing with big data Imagine the next scenario, we have a problem in some data center (packet lost, Kafka issues, maintenance, etc.). As a result, messages from that DC arrive to the backend server in delay. Meanwhile, messages from other DCs arrive on time. We never finish to consume all the messages of “old” minutes, but we keep opening more and more “new” minutes. The analysis server can’t process the folders, while they are in this “standby” mode. Eventually, memory is filled up and new messages can’t be consumed from Kafka. Another reason such a gap may occur, is that our topics are of different message types. Some types are lighter and faster to consume than others. We wanted to consume messages with a maximal gap of a few minutes. That way we prevent opening new buckets before closing old ones. In other words, only after all the messages from “old” minutes are consumed, will messages from “new” minutes be consumed. ### Which solution we decided to implement We decided to develop a mechanism to prioritize the consumption of Kafka topics. Such a mechanism will check if we want to process a message that was consumed from Kafka, or hold the processing for later. Using the graph below, I will describe what I am expecting the consumption of messages from Kafka to be like. ### What was the expected result The next graph shows a metric that measures every time a message is consumed from a Kafka broker. Each line represents a Kafka consumer. We measure the gap between the current time and the time that the message was sent from the frontend server (as explained above). For example, a gap of 10 hours means that we are now processing a message that was produced in the frontend server 10 hours ago. Usually and ideally, this gap should be a few minutes. Before turning prioritization on, some topics are consumed faster than others. This causes some topics to close the gap faster than others (to the right of the yellow arrow). Our mechanism should prioritize the tardy topics. Such that it will make the preliminary ones to wait (to the right of the red arrow). You can see that the gap of preliminary ones is steady (horizontal line) until tardy topics narrow the gap. ### How do we get where we want to be? ### Blocking Kafka topics we don’t want to process right now We map between the partitions and Booleans, which blocks the consuming of each partition if necessary, topicPartitionLocks. Blocking the preliminary ones, while continuing to consume from the tardy ones, creates prioritization of topics. A TimerTask updates this map and our consumers check if they are “allowed” to consume or have to wait – as you can see in the method waitForLatePartitionIfNeeded. ##### Prioritizer class public class Prioritizer extends TimerTask { private Map<String, Boolean> topicPartitionLocks = new ConcurrentHashMap<>(); private Map<String, Long> topicPartitionLatestTimestamps = new ConcurrentHashMap<>(); @Override public void run(){ updateTopicPartitionLocks(); } private void updateTopicPartitionLocks() { Optional<Long> minValue = topicPartitionLatestTimestamps.values().stream().min((o1, o2) -> (int) (o1 - o2)); if(! minValue.isPresent()) { return; } Iterator it = topicPartitionLatestTimestamps.entrySet().iterator(); while (it.hasNext()) { Boolean shouldLock = false; Map.Entry<String, Long> pair = (Map.Entry)it.next(); String topicPartition = pair.getKey(); if(pair.getValue() > (minValue.get() + maxGap)) { shouldLock = true; if(isSameTopicAsMinPartition(minValue.get(), topicPartition)) { shouldLock = false; } } topicPartitionLocks.put(topicPartition, shouldLock); } } public boolean isLocked(String topicPartition) { return topicPartitionLocks.get(topicPartition).booleanValue(); } } ##### waitForLatePartitionIfNeeded method private void waitForLatePartitionIfNeeded(final String topic, int partition) { String topicPartition = topic + partition; prioritizer.getTopicPartitionLocks.putIfAbsent(topicPartition); while(spaFileBufferPrioritizer.isLocked(topicPartition)) { monitorWaitForLatePartitionTimes(topicPartition, startTime); Misc.sleep(timeToWaitBetweenGapToTardyPartitionChecks.get()); } } We let our mechanism run for a while, and unfortunately things were not as expected. ### Avoid Rebalance Turning our prioritization, we saw a strange pattern in the “Topic Partition Gap” graph. We saw that the consumption is stuck very often. Trying to understand what was happening, we found that those breaks in consuming were a result of Kafka rebalancing. We can see this very clearly in the graph below. The horizontal lines represent a period of time where the gap stays steady. In other words, we are not consuming messages. What happened, is that whenever we paused the consumer, Kafka thought that this consumer was“dead” and started rebalancing. Moreover, when this consumer continued consuming, it was no longer registered in the broker, so Kafka had to rebalance again. This mechanism of Kafka is what makes it reliable and efficient, but in our scenario it was unnecessary. It is important to remember that if we were consuming from multiple machines this mechanism would obviously have been necessary. In order to prevent this rebalancing, we changed the next configuration in Kafka request.timeout.ms: 7300000 (~2hrs) max.poll.interval.ms: 7200000 (2hrs) These configurations will substantially lengthen the time that the broker waits for a consumer to consume, before considering it as “dead” and rebalancing. ### Monitoring Monitoring is very important, especially when working with thousands of messages consumed from Kafka every second. Using logs at such rates is irrelevant. In order to follow the processing behavior and performance, we defined metrics which measure the blocking term. In this case we defined a metric which measures the gap between the current time and the time that the message was sent from the frontend server. For example, that way I discovered the rebalance issue. ### Great Success This actually worked very well and you can see the result in the graph I showed before (to the right of the red arrow). We blocked the consumption of preliminary topics (horizontal line) until tardy topics narrowed the gap. Lastly, to summarize, 3 steps you should take to prioritize you Kafka topics - Block topics: you can define any condition you would like to for blocking topics. I used a condition about the time gap. You could use the concept of the above code. - Prevent rebalancing: change Kafka configuration as described above. - Monitor your results: follow your performance by defining a metric to measure your blocking term. That’s it for now, I hope you find this post useful ☺ --- ### Using Uncertainty to Interpret your Model URL: https://www.taboola.com/engineering/using-uncertainty-interpret-model/ Last Modified: 2025-01-14 14:37:35 As deep neural networks (DNN) become more powerful, their complexity increases. This complexity introduces new challenges, including model interpretability. Interpretability is crucial in order to build models that are more robust and resistant to adversarial attacks. Moreover, designing a model for a new, not well researched domain is challenging and being able to interpret what the model is doing can help us in the process. The importance of model interpretation has driven researchers to develop a variety of methods over the past few years and an entire workshop was dedicated to this subject at the NIPS conference last year. These methods include: - LIME: a method to explain a model’s prediction via local linear approximation - Activation Maximization: a method for understanding which input patterns produce maximal model response - Feature Visualizations - Embedding a DNN’s layer into a low dimensional explanation space - Employing methods from cognitive psychology - Uncertainty estimation methods - the focus of this post Before we dive into how to use uncertainty for debugging and interpreting your models, let’s understand why uncertainty is important. ## Why should you care about uncertainty? One prominent example is that of high risk applications. Let’s say you’re building a model that helps doctors decide on the preferred treatment for patients. In this case we should not only care about the accuracy of the model, but also about how certain the model is of its prediction. If the uncertainty is too high, the doctor should to take this into account. Self-driving cars are another interesting example. When the model is uncertain if there is a pedestrian on the road we could use this information to slow the car down or trigger an alert so the driver can take charge. Uncertainty can also help us with out of data examples. If the model wasn’t trained using examples similar to the sample at hand it might be better if it’s able to say “sorry, I don’t know”. This could have prevented the embarrassing mistake Google photos had when they misclassified african americans as gorillas. Mistakes like that sometimes happen due to an insufficiently diverse training set. The last usage of uncertainty, which is the purpose of this post, is as a tool for practitioners to debug their model. We’ll dive into this in a moment, but first, let’s talk about different types of uncertainty. ## Uncertainty Types There are different types of uncertainty and modeling, and each is useful for different purposes. Model uncertainty, AKA epistemic uncertainty: let’s say you have a single data point and you want to know which linear model best explains your data. There is no good way to choose between the different lines in the picture - we need more data! On the left: not enough data results in high uncertainty. On the right: given more data uncertainty reducesEpistemic uncertainty accounts for uncertainty in the model’s parameter. We are not sure which model weights describe the data best, but given more data our uncertainty decreases. This type of uncertainty is important in high risk applications and when dealing with small and sparse data. As an example, let’s say you want to build a model that gets a picture of an animal, and predicts if that animal will try to eat you. Let’s say you trained the model on different pictures of lions and giraffes and now it saw a zombie. Since the model wasn’t trained on pictures of zombies, the uncertainty will be high. This uncertainty is the result of the model, and given enough pictures of zombies it will decrease. Data uncertainty, or aleatoric uncertainty, captures the noise inherent in the observation. Sometimes the world itself is stochastic. Obtaining more data will not help us in that case, because the noise is inherent in the data. To understand this point, let’s get back to our carnivorous animals model. Our model can recognize that an image contains a lion, and therefore you’re likely to be eaten. But what if that lion is not hungry right now? This time the uncertainty comes from the data. Another example is that of two snakes that look the same but while one of them is venomous, the other isn’t. Aleatoric uncertainty is divided into two types: - Homoscedastic uncertainty: uncertainty is the same for all inputs. - Heteroscedastic uncertainty: uncertainty that depends on the specific input at hand. For instance, for a model that predicts depth in an image a featureless wall is expected to have a higher level of uncertainty than that of an image with strong vanishing lines. Measurement uncertainty: another source of uncertainty is the measurement itself. When the measurement is noisy, the uncertainty increases. In the animals example the model’s confidence can be impaired if some of the pictures are taken using a bad quality camera; or if we were running away from a scary hippo and as a result we only have blurry images to work with. Noisy labels: with supervised learning we use labels to train the models. If the labels are noisy, the uncertainty increases. There are various ways to model each type of uncertainty. These will be covered in the following posts in this series. For now, let’s assume we have a black box model that exposes the uncertainty it has regarding its predictions. How can we use it in order to debug the model? Let’s consider one of our models in Taboola used for predicting the likelihood of a user clicking on a content recommendation, also known as CTR (Click Through Rate). ## Using uncertainty to debug your model The model has many categorical features represented by embedding vectors. The model might have difficulties with learning generalized embeddings for rare values. A common way to solve this is to use a special Out Of Vocabulary (OOV) embedding. Think about the advertiser of an article. All rare advertisers share the same OOV embedding, therefore, from the point of view of the model they are essentially one advertiser. This OOV advertiser has many different items, each with different CTR. If we’d use only the advertiser as a predictor for CTR, we should get high uncertainty for OOV. To validate the model outputs high uncertainty for OOV, we took a validation set and switched all the advertisers embeddings into OOV. Next, we inspected what was the uncertainty before and after the switch. As expected, the uncertainty increased due to the switch. The model was able to learn that given an informative advertiser it should reduce the uncertainty. We can repeat this for different features and look for ones that result in low uncertainty when replaced with OOV embeddings. Either those features are uninformative, or something in the way we feed them to the model is not ideal. We can even go to finer granularity: some advertisers have high variability between CTR of different items, while others have items with roughly the same CTR. We would expect the model to have higher uncertainty for advertisers of the first type. A useful analysis is therefore looking at the correlation between uncertainty and CTR variability within an advertiser. If the correlation isn’t positive, it means the model failed to learn what uncertainty to associate with each advertiser. This tool allows us to understand if something went wrong in the training process or in the model’s architecture, indicating we should further debug it. We can perform a similar analysis and see if the uncertainty associated with a specific item decreases the more times we show it (i.e. show it to more users / in more places). Again, we expect the model to become more certain, and if it doesn’t - debug we will! Another cool example is the title feature: unique titles with rare words should incur high model uncertainty. This is the result of the model not seeing a lot of examples from that area of all possible titles. We can look in the validation set for a group of similar titles that are rare and estimate the model’s uncertainty on these titles. Then we’ll retrain the model using one of the titles, and see if the uncertainty has been reduced for the entire group. Indeed, we can see that’s exactly what happened: Wait a second… By exposing the model to some titles it was able to get better and be more certain about a bunch of new titles. Maybe we can use that to somehow encourage exploration of new items? Well, yes we can! More on that in a following post of the series. ## Final thoughts Uncertainty is a big deal in many domains. Identifying which uncertainty type is important is application specific. You can use them in a variety of ways once you know how to model them. In this post we discussed how you can use them to debug your model. In the next post we’ll talk about different ways to get uncertainty estimations from your model. This is the first post of a series related to a paper we're presenting in a workshop in this year KDD conference: deep density networks and uncertainty in recommender systems. --- ### Title to Image Search for Improved Thumbnail Selection URL: https://www.taboola.com/engineering/title-to-image-search-for-improved-thumbnail-selection/ Last Modified: 2025-01-14 14:37:35 ### Introduction One of the key creative aspects of an advertisement is choosing the image that will appear alongside the advertisement text. The advertisers aim is to select an image that will draw the attention of the users and get them to click on the add, while remaining relevant to the advertisement text. Say that you’re an advertizer wanting to place a new ad titled "15 healthy dishes you must try”. There are endless possibilities of choosing the image thumbnail to go along with this title, clearly some more clickable than others. One can apply best practices in choosing the thumbnail, but manually searching for the best image (out of possibly thousands that fit this title) is time consuming and impractical. Moreover, there is no clear way of quantifying how much an image is related to a title and more importantly - how clickable the image is, compared to other options. This is where our image search comes into play. We developed a text to image search algorithm that given a proposed title, scans an image gallery to find the most suitable images and estimates their expected click through rate. For example, here are the images returned by our algorithm for the query "15 healthy dishes you must try” along with their predicted Click Through Ratio (CTR): ### Method In a nutshell, our pipeline is composed of extracting title embeddings, extracting image embeddings and learning two transformations to map those embeddings to a joint space. We elaborate on each of these steps below. #### Embeddings First, let us consider the concept of embeddings. An embedding function is a function that receives an item (be it image or text) and produces a numerical vector for the item that captures its semantic meaning, such that if two items are semantically similar - then their vectors would be close. For example the vectors for the two titles “Kylie Jenner, 19, Buys Fourth California Mansion at $12M” and “18 Facts about Oprah and her Extravagant Lifestyle” should be similar, or close in the embedding space, while the vectors for the titles “New Countries that are surprisingly good at these sports” and “Forgetting things? You should read this!” should be very different, or far away from each other in the embedding space. Similarly, the vectors for following two images should be close in the embedding space since their semantic meaning is the same (dogs making funny faces), even though the specific image values are vastly different. At first glance, one might consider such an embedding function to be very explicit: for example, counting the number of faces in an image, the number of words in a sentence, perhaps creating a list of the objects in an image. Such explicit functions are in fact extremely hard to implement and in practice do not preserve semantics very well, at least not in a way which easily allows us to compare them. In practice embedding functions represent the information of the item in a much more implicit manner, as we will see next. More concretely, let A ={a0,a1,..., ak } be the space of all items to embed. An embedding function f:ARd maps each item to a d-dimensional vector such that if two items ai , aj are semantically similar, then their embeddings distance L( f(ai), f(aj) ) should be small in some predefined distance metric L (usually standard euclidean distance). ### Image embedding Pre-trained Convolutional Neural Networks have shown to be extremely useful in obtaining strong image representation . In our framework, we use a ResNet 152 model pretrained on the Imagenet Large Scale Recognition Challenge (ILSRC) dataset. Each image is fed to the pretrained network and we take the activation of the res5c layer as a 2048 embedding vector. To better illustrate the quality of the embeddings, we implement a simple image search demo on our own Taboola image dataset. Here are the 4 most similar images to several image queries (the query is the leftmost image): ### Title embedding A number of sentence embeddings methods were proposed in the literature. However, we found Infersent to be particularly useful in our application. While common approaches to sentence representation are unsupervised or leverage the structure of the text for supervision, Infersent trains sentence embeddings in a fully supervised manner. To this end, leverages the SNLI dataset which composes of 570K pairs of sentences with a label of the relationship between them (contradiction, neutral or entailment). A number of different models were trained in and a simple Max-pooling of a BiLSTM hidden states reached the best performance on a number of NLP tasks and benchmarks. Below are some examples to title queries and their nearest neighbors: Query: “15 Adorable Puppy Fails The Internet Is Obsessed With” NN1: 18 Puppy Fails That The Internet Is Obsessed With NN2: 13 Adorable Puppy Fails That Will To Make You Smile NN3: 16 Dog Photoshoots The Internet Is Absolutely Obsessed With Query: 10 Negative Side Effects of Low Vitamin D Levels NN 1: How to Steer Clear of Side Effects From Blood Thinners NN 2: 7 Signs and Symptoms of Vitamin D Deficiency People Often Ignore NN 3: 8 Facts About Vitamin D and Rheumatoid Arthritis ### Matching titles and images Given a query title, we would like to fit it with the most relevant images. For example, the query “20 new cars to buy in 2018” would return images of new and shiny high end cars. As mentioned before, our analysis of titles and images is not explicit, meaning that we do not explicitly detect the objects and named entities of the titles and match it with images containing the same objects. Instead, we rely on the title and image embeddings which implicitly capture the semantics of titles and images. Next, we need to have a better grasp of what does it mean that a title corresponds to an image. To better define it, we built a large training set composed of pairs of <advertisement title, advertisement image> from advertisements that appeared in the Taboola widget. Given such data, we can define that a given title to corresponds to an images im0 if (and only if) a title t0'that is semantically similar to to appeared in an advertisement along with an image im0' that is semantically similar to im0. In other words, we say that a title corresponds to an image (and vice versa) if they appeared in an advertisement together. Given this definition, as a naive first solution we can apply the following pipeline: given a title to, find its embedding e0. Next find the most similar title embedding to it from our gallery, denoted by eo' with corresponding title t0'. Finally, take as output the image that appeared in the original advertisement with t0'. While this is a sound solution, it does not fully leverage our training set since it uses the information from only one <title, image> training pair. Indeed, our experiments showed that the images suggested in this manner were far less suitable than images that were suggested from an approach which uses the entire training gallery. In order to leverage the entire training set of titles and images, we would like to implicitly capture the semantic similarity of titles, images and the given correspondence between them in our training set. To this end, we employ an approach that learns a joint title-image space in which both the titles and the images reside. This allows us to directly measure distance between titles and images and in particular, given a query title, find its nearest neighbor image. For example, the closest sentence to the query “30 Richest Actresses in America” is “The 30 Richest Canadians Ranked By Wealth” which was displayed along with the following images: As reference, here are the results obtained by our method which is explained next: ### Learning a joint space Assume we have m pairs of titles with their corresponding images <t1,im1>,<t2,im2>,...,<tm,imm>. For simplicity of notation, let’s assume their embeddings are already denoted as <ti,imi>. We denote the joint title-image space by V. We would like to find two mappings Ft:{ti}V from the space of title embeddings to the joint space V and Fim:{imi}V from the space of image embeddings to the same joint space, such that the distance between Ft (ti) and Fim (imi) in the joint space V would be small. Numerous methods for obtaining such mappings exist . However, we found a simple statistical method to be useful in our application. To this end, we employ Canonical Correlation Analysis (CCA) . Let T= be the set of title embeddings and M = be the set of image embeddings. CCA learns linear projections Fim and Ft such that the correlation between the set of title embeddings and image embeddings in the joint space is maximized. Mathematically, CCA finds Fim and Ft such that = corr(FtT,FimM) is maximized, where the last equation is solved by formulating it as an eigenvalue problem. Note that in our implementation Fim and Ft are linear projections. This can be further improved learning non-linear or even deep transformations and is left for future work. ### Click Through Rate estimation Now, given a proposed image, we would like to estimate it’s Click Through Rate (CTR) - the amount of clicks divided by the number of appearances. Many of the images in our gallery already appeared in campaigns, so we can use their historic data. For others, we train a simple linear Support Vector Machine regression model based on their image embeddings. ### Full image search pipeline Our final image search pipeline works as follows: First, we take a large gallery of titles and images, run the ResNet 152 model and the InferSent model to obtain image and title embeddings. Next, we apply CCA to learn two linear projections to a joint space and multiply the image embeddings by the learned image projection to map them to a joint space. All those steps are run offline. Now, given a new title text by the advertizer, we run the predefined Infersent model to obtain its embedding. Then, the embedding is multiplied by the learned sentence projection to map it to the joint title-image space. Finally, we return the 10 closest images in the space along with their predicted ctr. ### Examples: Now for the fun part. Here are some quantitative results of our pipeline using real titles that appeared in some of our recommendations. For each title, we present the top 8 images returned by our system with their predicted click through ratio. Title: “30 Richest Actresses in America” Images returned by our system: Notice this is the same title we discussed earlier when comparing our approach with the naive solution. Here we also show the predicted CTR. We assume our model mistook Steven Tyler’s for a female actress due to his long hair… Title: “Baby Born With White Hair Stumps Doctors”. Images returned by our system: Note that some of the rightmost images are a bit off since they are farther away from the title embedding in the joint space. Our algorithm also works for longer titles: “Now You Can Turn Your Passion For Helping Others Into A Career - Become A Nurse 100% Online. Find Enrollment & Scholarships Compare Schedules Now!” Title: “20 Ridiculously Adorable Pet Photos That Went Viral” Images returned by our system: Title: “Don't Miss this Incredible Offer if You Fly with Delta Ends 11/8!” Images returned by our system (notice the leftmost image): ### Wrapping up Our goal in this post was to illustrate our system for solving a real world difficulty of image search using free text, specifically to be used by advertisers to optimize which image to use for a given add. While we have developed the system for solving our needs, it is not tailored in any way to advertising and can be used for implementing image search in other domains and industries. In fact, the concepts described above are even more general (e.g. embeddings) and can be useful for a variety of applications. In its current implementation, the system returns the 10 most similar images for a given query without any other considerations. In future work we intend to add an option of returning only images with high ctr and provide the user (i.e.: advertizer) with an option to control the importance of visual similarity in the returned images relative to their ctr (e.g.: return images that might less fit the semantics of the title but have higher ctr). We also intend to add functionality of controlling how versatile the returned images will be relative to their semantic similarity with the title (e.g.: return a set of images which might fit the title less well, but are versatile and provide the advertiser with more options). #### References Dalal, Navneet, and Bill Triggs. "Histograms of oriented gradients for human detection." Computer Vision and Pattern Recognition, 2005. CVPR 2005. IEEE Computer Society Conference on. Vol. 1. IEEE, 2005. ‏ Ojala, Timo, Matti Pietikäinen, and David Harwood. "A comparative study of texture measures with classification based on featured distributions." Pattern recognition 29.1 (1996): 51-59. ‏ Csurka, Gabriella, et al. "Visual categorization with bags of keypoints." Workshop on statistical learning in computer vision, ECCV. Vol. 1. No. 1-22. 2004. ‏ Razavian, Ali Sharif, et al. "CNN features off-the-shelf: an astounding baseline for recognition." Computer Vision and Pattern Recognition Workshops (CVPRW), 2014 IEEE Conference on. IEEE, 2014. ‏ He, Kaiming, et al. "Deep residual learning for image recognition." Proceedings of the IEEE conference on computer vision and pattern recognition. 2016. ‏ Russakovsky, Olga, et al. "Imagenet large scale visual recognition challenge." International Journal of Computer Vision 115.3 (2015): 211-252. ‏ Maaten, Laurens van der, and Geoffrey Hinton. "Visualizing data using t-SNE." Journal of machine learning research 9.Nov (2008): 2579-2605. ‏ Le, Quoc, and Tomas Mikolov. "Distributed representations of sentences and documents." International Conference on Machine Learning. 2014. ‏ Kiros, Ryan, et al. "Skip-thought vectors." Advances in neural information processing systems. 2015. ‏ Hill, Felix, Kyunghyun Cho, and Anna Korhonen. "Learning Distributed Representations of Sentences from Unlabelled Data." Proceedings of NAACL-HLT. 2016. ‏ Conneau, Alexis, et al. "Supervised Learning of Universal Sentence Representations from Natural Language Inference Data." Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing. 2017. ‏ Bowman, Samuel R., et al. "A large annotated corpus for learning natural language inference." arXiv preprint arXiv:1508.05326 (2015). ‏ Chandar, Sarath, et al. "Correlational neural networks." Neural computation 28.2 (2016): 257-285. ‏ Wang, Weiran, et al. "On deep multi-view representation learning." International Conference on Machine Learning. 2015. ‏ Yan, Fei, and Krystian Mikolajczyk. "Deep correlation for matching images and text." Computer Vision and Pattern Recognition (CVPR), 2015 IEEE Conference on. IEEE, 2015. ‏ Eisenschtat, Aviv, and Lior Wolf. "Linking Image and Text With 2-Way Nets." Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2017. ‏ Vinod, Hrishikesh D. "Canonical ridge and econometrics of joint production." Journal of econometrics 4.2 (1976): 147-166. ‏ Cortes, Corinna, and Vladimir Vapnik. "Support-vector networks." Machine learning 20.3 (1995): 273-297. ‏ Li, Yingming, Ming Yang, and Zhongfei Zhang. "Multi-view representation learning: A survey from shallow methods to deep methods." arXiv preprint arXiv:1610.01206 (2016). Lai, Pei Ling, and Colin Fyfe. "Kernel and nonlinear canonical correlation analysis." International Journal of Neural Systems 10.05 (2000): 365-377. ‏ Andrew, Galen, et al. "Deep canonical correlation analysis." International Conference on Machine Learning. 2013.‏ --- ### Uncertainty for CTR Prediction: One Model to Clarify Them All URL: https://www.taboola.com/engineering/uncertainty-for-ctr-prediction-one-model-to-clarify-them-all/ Last Modified: 2025-01-14 14:37:35 In the first post of the series we discussed three types of uncertainty that can affect your model - data uncertainty, model uncertainty and measurement uncertainty. In the second post we talked about various methods to handle the model uncertainty specifically . Then, in our third post we showed how we can use the model’s uncertainty to encourage exploration of new items in recommender systems. Wouldn’t it be great if we can handle all three types of uncertainty in a principled way using one unified model? In this post we’ll show you how we at Taboola implemented a neural network that estimates both the probability of an item being relevant to the user, as well as the uncertainty of this prediction.   ## Let’s jump to the deep water A picture is worth a thousand words, isn’t it? And a picture containing a thousand neurons?... In any case, this is the model we use. The model is composed of several modules. We’ll explain the goal of each one, and then the picture will become clearer… ## Item module The model tries to predict the probability that an item will be clicked, i.e - the CTR (Click Through Rate). To do so, we have a module that gets as input the item’s features such as its title and thumbnail, and outputs a dense representation - a vector of numbers if you will. Once the model is trained, this vector will contain the important information extracted out of the item. ## Context module We said the model predicts the probability of a click on an item, right? But in which context is the item shown? Context can mean many things - the publisher, the user, the time of day, etc. This module gets as input the features of the context. It then outputs the dense representation of the context. ## Fusion module So we have the information extracted out of both the item and the context. For sure, there’s some interaction between the two. For instance, an item about soccer probably will have higher CTR in a sports publisher compared to a finance publisher. This module fuses the two representations into one, in a similar fashion to collaborative filtering. ## Estimation module At the end we have a module whose goal is to predict the CTR. In addition, it also estimates uncertainty about the CTR estimation. I guess you’re mostly uncertain about how this module works, so let’s shed some light on it. We’ll walk you through the three types of uncertainty we’ve mentioned, and show you how each one is handled by our model. First, let’s tackle the data uncertainty. ## Data uncertainty Let’s take some generic neural network trained on a regression task. One common loss function is MSE - Mean Squared Error. We like this loss because it’s intuitive, right? You want to minimize the errors… But it turns out that when you minimize MSE, you implicitly maximize the likelihood of the data - assuming the label is distributed normally with a fixed standard deviation ?. This ? is the noise inherent in the data. One thing we can do is explicitly maximize the likelihood by introducing a new node which we’ll call ?. Plugging it into the likelihood equation and letting the gradients propagate enable this node to learn to output the data noise. We didn’t achieve anything different, right? We got an equivalent result to the initial MSE based model. However, now we can introduce a link from the last layer to ?: Now we’re getting into something interesting! ? is now a function of the input. It means the model can learn to associate different levels of data uncertainty with different inputs. We can make the model even more powerful. Instead of estimating a Gaussian distribution, we can estimate a mixture of Gaussians. The more Gaussians we put into the mix, the more capacity the model will have - and more prone to overfitting, so be careful with that. This architecture is called MDN - Mixture Density Network. It was introduced by Bishop et al. in 1994. Here is an example of what it captures: We have two groups of similar items - one about shopping, and the other about sports. It turns out the shopping items tend to have more variable CTR - maybe due to trendiness. Indeed, if we ask the model to estimate the uncertainty of one item in each group (the dotted graph in the figure), we get higher uncertainty for shopping compared to sports.   So data uncertainty is behind us. What’s next? ## Measurement uncertainty This one is a bit more tricky. In the first post we explained that sometimes measurement can be noisy. This might result in noisy features or even noisy labels. In our case, our label y is the empiric CTR of an item - the number of times it was clicked so far, divided by the number of times it was shown. Let’s say the true CTR of an item is y* - that is, without measurement noise. This would be the CTR had we shown the item infinite amount of times in the context. But time is finite (at least the time we’ve got), so we showed it only a finite amount of times. We measured an observed CTR y. This y has measurement noise - which we denote by ?. Next, we assume ? is distributed normally with a ?? as the standard deviation. ?? function of r - the number of times we showed the item. The higher r is, the smaller ?? gets, which makes y more similar to y*. At the end of the day, after we spare you from the mathematical details (which you can find in our paper), we get this likelihood equation: This is the same as the likelihood we have in the MDN architecture of a mixture of Gaussians, with one difference - the error term is split to two: - data uncertainty (?i) - measurement uncertainty (??) Now that the model is able to explain each uncertainty using a different term, the data uncertainty is not polluted by the measurement uncertainty. Besides being able the explain the data in a better way, this allows us to use more data in the training process. This is due to the fact that prior to this work we filtered out data with too much noise. ## Last but not least In a previous post we discussed how to handle model uncertainty. One of the approaches we described was using dropout at inference time. Being able to estimate model uncertainty allows us to understand better what the model doesn’t know because of lack of data. So let’s put it to the test! Let’s see if unique titles are associated with high uncertainty. We’ll map each title in the training set to a dense representation (e.g. average word2vec embeddings) and expect the model to be less certain about unique titles - titles that are mapped to sparse regions of the embedding space. To test it, we calculated the sparse and dense regions by calculating KDE (Kernel Density Estimation). This is a method for estimating the PDF (Probability Density Function) of our space. Next, we asked the model to estimate the uncertainty associated with each title. It turns out that indeed the model has higher uncertainty in the sparse regions! Nice… What would happen if we show the model more titles from the sparse regions? Will it be more certain about these regions? Let’s test it out! We took a bunch of similar titles about cars and removed them from the training set. Effectively, it altered their region in the space from dense to sparse. Next, we estimated the model uncertainty over these titles. As expected, the uncertainty was high. Finally, we added only one of the titles to the training set and retrained the model. To our satisfaction, now the uncertainty has reduced for all of these items. Neat! As we saw in the post about exploration-exploitation, we can encourage exploration of these sparse regions. After doing so, the uncertainty will decrease. This will result in a natural decay of exploration of that region. ## Final thoughts In this post we elaborated on how we model all three types of uncertainty - data, model and measurement - in a principled way, using one unified model. We encourage you to think how you can use uncertainty in your application as well! Even it you don’t need to explicitly model uncertainty in your prediction, you might benefit from using it in the training process - if your model can understand better how data is generated and how uncertainty affects the game, it might be able to improve. This is the forth post of a series related to a paper we’re presenting in a workshop in this year KDD conference: deep density networks and uncertainty in recommender systems. --- ### Recommender Systems: Exploring the Unknown Using Uncertainty URL: https://www.taboola.com/engineering/recommender-systems-exploring-the-unknown-using-uncertainty/ Last Modified: 2025-01-14 14:37:35 Now that we know what uncertainty types exist and learned some ways to model them, we can start talking about how to use them in our application. In this post we’ll introduce the exploration-exploitation problem and show you how uncertainty can help in solving it. We’ll focus on exploration in recommender systems, but the same idea can be applied in many applications of reinforcement learning - self driving cars, robots, etc. ## Problem Setting The goal of a recommender system is to recommend items that the users might find relevant. At Taboola, relevance is expressed via a click: we show a widget containing content recommendations, and the users choose if they want to click on one of the items. The probability of the user clicking on an item is called Click Through Rate (CTR). If we knew the CTR of all the items, the problem of which items to recommend would be easy: simply recommend the items with the highest CTR. The problem is that we don’t know what the CTR is. We have a model that estimates it, but it’s obviously not perfect. Some of the reasons for the imperfection are the uncertainty types inherent in recommender systems, which we discussed in the first post of the series. ## The Exploitation vs Exploration Tradeoff So now we’re facing a challenging situation - one that we’re all familiar with from our day-to-day lives: imagine you’ve just entered an ice cream shop. You now face a crucial decision - out of about 30 flavors you need to choose only one! You can go with two strategies: either go with that favorite flavor of yours that you already know is the best; or explore new flavors you never tried before, and maybe find a new best flavor. These two strategies - exploitation and exploration - can also be used when recommending content. We can either exploit items that have high CTR with high certainty - maybe because these items have been shown thousands of times to similar users; or we can explore new items we haven’t shown to many users in the past. Incorporating exploration into your recommendation strategy is crucial - without it new items don’t stand a chance against older, more familiar ones. ## Let’s explore exploration approaches The easiest exploration-exploitation approach you can implement is the ϵ-greedy algorithm, where you allocate ϵ percents of the traffic to explore new items in a random manner. The rest of the traffic is reserved for exploitation. Despite not being optimal, this method is easy to understand. It can serve as a solid baseline for more sophisticated approaches. So how can we look for good items in a wiser manner? looking for good items in a wise manner A more advanced approach - Upper Confidence Bound (UCB) - uses uncertainty. Each item is associated with its expected CTR, and confidence bound over that CTR. The confidence bound captures how uncertain we are about the item’s CTR. The vanilla UCB algorithm keeps track of the expected CTR and confidence bound by using empirical information alone: for each item we keep track of the empirical CTR (what percent of similar users have clicked on it), and the confidence bound is calculated by assuming a binomial distribution. Take for example the plain chocolate flavor you always order. You know it’s good - you give it 8 stars out of 10. Today a new flavor has arrived. You have no empiric information about it, which means it can be anything from 1 to 10 stars. Using this confidence interval, if you would want to explore you’d go with the new flavor, since there’s the chance it’ll be a 10 stars flavor. That strategy is exactly what UCB is all about - you choose the item with the highest upper confidence bound value - in our case confidence bound over CTR estimation. The motivation behind this strategy is that over time the empiric CTR will tend towards the true CTR, and the confidence bound will shrink to 0. After enough time, we’ll explore everything. Another popular approach is the Thompson Sampling method. In this approach we use the entire estimated distribution over the item’s CTR instead of only a confidence bound. For each item we sample a CTR out of its distribution. These approaches might work well when the number of available items is fixed. Unfortunately, at Taboola every day thousands of new items enter the system, and others become obsolete. By the time we get a reasonable confidence bound for an item it might leave the system. Our efforts would have been in vain. It’s like doing a world tour, each day visiting a new town with a vast amount of ice cream flavors to explore. The horror! We need an approach that can estimate the CTR of a new item without showing it even once. We need some food critic magazine that will guide us through the buffet of content recommendations. Consider a new type of chocolate flavor that just arrived. Since you know you love chocolate you have a pretty good guess that you’ll like the new flavor too. In the vanilla UCB approach (no, that’s not a name of a flavor), you won’t be able to infer it - you rely on empirical information only. In a future post we’ll elaborate on how we use a neural network to estimate the CTR of a new item, as well as the level of uncertainty. Using this uncertainty, we can apply the UCB approach in order to explore new items. Unlike the vanilla UCB that relies on empirical data, here we can use the model’s estimation to avoid showing items with low CTR. We can gamble on the horses we think will win. ## Online metrics and results How can we know how well we explore new items? We need some exploration throughput metric. In Taboola we have A/B testing infrastructure supporting many models running on different shares of traffic. Back to ice cream! Let’s say you brought your friends to help you explore the different flavors. Obviously if one of your friends randomly picks flavors, he has the best exploration throughput, but not the smartest. The other friend that orders the flavor the others have found tasty enjoys the most, but contributes nothing to the exploration effort. At Taboola we measure exploration throughput as follows: for each item that has been shown enough times, and in enough different contexts (e.g - different web sites) we declare that item to have crossed the exploration phase. Next, we analyze which models contributed to this successful effort. In order to count, a model has to show that item enough times. Using this perspective, the throughput of a model is defined to be the number of items it contributed to. Using this metric, we were able to assert that indeed showing items randomly yields the best throughput, but with a tendency for bad items. A model that doesn’t use the UCB approach shows good items, but has worse throughput. The model with UCB is somewhere between in terms of throughput, and shows only slightly worse items compared to the non-UCB model. Hence, we conclude our UCB model has a good tradeoff between exploring new items, and choosing the good items. We believe this tradeoff is worthwhile in the long term. ## Final thoughts The exploration-exploitation problem is an exciting challenge for many companies in the recommender systems domain. We hope our advancements will serve others in their journey of providing the best service to their users. We believe this is a small step in a big journey yet to be fulfilled, and we’re intrigued by the thought of what shape this field will take in the following years. In the next post of the series we’ll elaborate on the model we used for estimating CTR and uncertainty, so stay tuned. This is the third post of a series related to a paper we’re presenting in a workshop in this year KDD conference: deep density networks and uncertainty in recommender systems. --- ### Neural Networks from a Bayesian Perspective URL: https://www.taboola.com/engineering/neural-networks-bayesian-perspective/ Last Modified: 2025-01-14 14:37:36 Understanding what a model doesn’t know is important both from the practitioner’s perspective and for the end users of many different machine learning applications. In our previous blog post we discussed the different types of uncertainty. We explained how we can use it to interpret and debug our models. In this post we’ll discuss different ways to obtain uncertainty in Deep Neural Networks. Let’s start by looking at neural networks from a Bayesian perspective. ## Bayesian learning 101 Bayesian statistics allow us to draw conclusions based on both evidence (data) and our prior knowledge about the world. This is often contrasted with frequentist statistics which only consider evidence. The prior knowledge captures our belief on which model generated the data, or what the weights of that model are. We can represent this belief using a prior distribution p(w) over the model’s weights. As we collect more data we update the prior distribution and turn in into a posterior distribution using Bayes’ law, in a process called Bayesian updating: This equation introduces another key player in Bayesian learning - the likelihood, defined as p(y|x,w). This term represents how likely the data is, given the model’s weights w. ## Neural networks from a Bayesian perspective A neural network’s goal is to estimate the likelihood p(y|x,w). This is true even when you’re not explicitly doing that, e.g. when you minimize MSE. To find the best model weights we can use Maximum Likelihood Estimation (MLE): Alternatively, we can use our prior knowledge, represented as a prior distribution over the weights, and maximize the posterior distribution. This approach is called Maximum Aposteriori Estimation (MAP): The term logP(w), which represents our prior, acts as a regularization term. Choosing a Gaussian distribution with mean 0 as the prior, you’ll get the mathematical equivalence of L2 regularization. Now that we start thinking about neural networks as probabilistic creatures, we can let the fun begin. For start, who says we have to output one set of weights at the end of the training process? What if instead of learning the model’s weights, we learn a distribution over the weights? This will allow us to estimate uncertainty over the weights. So how do we do that? ## Once you go Bayesian, you never go back We start again with a prior distribution over the weights and aim at finding their posterior distribution. This time, instead of optimizing the network’s weights directly we’ll average over all possible weights (referred to as marginalization). At inference, instead of taking the single set of weights that maximized the posterior distribution (or the likelihood, if we’re working with MLE), we consider all possible weights, weighted by their probability. This is achieved using an integral: x is a data point for which we want to infer y, and X,Y are training data. The first term p(y|x,w) is our good old likelihood, and the second term p(w|X,Y) is the posterior probability of the model’s weights given the data. We can think about it as an ensemble of models weighted by the probability of each model. Indeed this is equivalent to an ensemble of infinite number of neural networks, with the same architecture but with different weights. ## Are we there yet? Ay, There’s the rub! Turns out that this integral is intractable in most cases. This is because the posterior probability cannot be evaluated analytically. This problem is not unique to Bayesian Neural Networks. You would run into this problem in many cases of Bayesian learning, and many methods to overcome this have been developed over the years. We can divide these methods into two families: variational inference and sampling methods. ## Monte Carlo sampling We have a problem. The posterior distribution is intractable. What if instead of computing the integral over the true distribution we’ll approximate it with the average of samples drawn from it? One way to do that is the Markov Chain Monte Carlo - you construct a markov chain with the desired distribution as its equilibrium distribution. ## Variational Inference Another solution is to approximate the true intractable distribution with a different distribution from a tractable family. To measure the similarity of the two distribution we can use KL divergence: Let q be a variational distribution parameterized by θ. We want to find the value of θ that minimizes the KL divergence: Look at what we’ve got: the first term is the KL divergence between the variational distribution and the prior distribution. The second term is the likelihood with regards to qθ. So we’re looking for qθ that explains the data best, but on the other hand is as close as possible to the prior distribution. This is just another way to introduce regularization into neural networks! Now that we have qθ we can use it to make predictions: The above formulation comes from a work by DeepMind in 2015. Similar ideas were presented by graves in 2011 and go back to Hinton and van Camp in 1993. The keynote in NIPS Bayesian Deep Learning workshop had a very nice overview of how these ideas evolved over the years. OK, but what if we don’t want to train a model from scratch? What if we have a trained model that we want to get uncertainty estimation from? Can we do that? It turns out that if we use dropout during training, we actually can. Professional data scientists contemplating the uncertainty of their model - an illustration ## Dropout as a mean for uncertainty Dropout is a well used practice as a regularizer. In training time, you randomly sample nodes and drop them out, that is - set their output to 0. The motivation? You don’t want to over rely on specific nodes, which might imply overfitting. In 2016, Gal and Ghahramani showed that if you apply dropout at inference time as well, you can easily get an uncertainty estimator: - Infer y|x multiple times, each time sample a different set of nodes to drop out. - Average the predictions to get the final prediction E(y|x). - Calculate the sample variance of the predictions. That’s it! You got an estimate of the variance! The intuition behind this approach is that the training process can be thought of as training 2^m different models simultaneously - where m is the number of nodes in the network: each subset of nodes that is not dropped out defines a new model. All models share the weights of the nodes they don’t drop out. At every batch, a randomly sampled set of these models is trained. After training, you have in your hands an ensemble of models. If you use this ensemble at inference time as described above, you get the ensemble’s uncertainty. ## Sampling methods vs Variational Inference In terms of the bias-variance tradeoff, variational inference has high bias because we choose the distributions family. This is a strong assumption that we’re making, and as any strong assumption, it introduces bias. However, it’s stable, with low variance. Sampling methods on the other hand have low bias, because we don’t make assumptions about the distribution. This comes at the price of high variance, since the result is dependent on the samples we draw. ## Final thoughts Being able to estimate the model uncertainty is a hot topic. It’s important to be aware of it in high risk applications such as medical assistants and self-driving cars. It’s also a valuable tool to understand which data could benefit the model, so we can go and get it. In this post we covered some of the approaches to get model uncertainty estimations. There are many more methods out there, so if you feel highly uncertain about it, go ahead and look for more data :) In the next post we’ll show you how to use uncertainty in recommender systems, and specifically - how to tackle the exploration-exploitation challenge. Stay tuned. This is the second post of a series related to a paper we’re presenting in a workshop in this year KDD conference: deep density networks and uncertainty in recommender systems. --- ### The Hitchhiker’s Guide to Hyperparameter Tuning URL: https://www.taboola.com/engineering/the-hitchhikers-guide-to-hyperparameter-tuning/ Last Modified: 2025-01-14 14:37:36 Now that more than a year has passed since our first deep learning project emerged, we have had to keep moving forward and delivering the best models we can. Doing so has involved a lot of research, trying out different models, from as simple as bag-of-words, LSTM and CNN, to the more advanced attention, MDN and multi-task learning. Even the simplest model we tried has many hyperparameters, and tuning these might be even more important than the actual architecture we ended up using - in terms of the model’s accuracy. Although there’s a lot of active research in the field of hyperparameter tuning (see 1, 2, 3), implementing this tuning process has evaded the spotlight. If you go around and ask people how they tune their models, their most likely answer will be “just write a script that does it for you”. Well, that's easier said than done... Apparently, there are a few things you should keep in mind when implementing such a script. Here, at Taboola, we implemented a hyperparameter tuning script. Let me share with you the things we learned along the way... ### Let’s start simple Sometimes using scikit-learn for hyperparameter tuning might be enough - at least for personal projects. For long term projects, when you need to keep track of the experiments you’ve performed, and the variety of different architectures you try keeps increasing, it might not suffice. ### Bare bones The first version of the script was simple, but encompassed most of our needs. The requirements were: #### Easy to run You are going to run this script many times. So, it should be as easy as possible to specify what experiments you want to run. We ended up with the following JSON format: { “architecture”: “lstm-attention”, “date-range”: ], “parameters”: { “num_of_attentions”: , “attention_hidden_layer_size”: , “attention_regularization”: } } - Architecture: architecture you want to tune, assuming your code supports multiple types - Date-range: list of tuples, each defines a time range of data that will be used to train a model. Each experiment will be executed once per date-range - Parameters: values to try for each hyperparameter. Our initial implementation only supported a finite set of values (grid search) The script randomly generates experiments out of this JSON. We created a Jenkins job that runs the script on one of our machines with GPUs, thus freeing us from the need to use SSH. #### Enrich experiments with metrics Our training process generates many metrics, such as MSE, loss and training time. You can choose whichever you want, and they will show in the results. #### Save results to the cloud Results are saved as a CSV file in Google Cloud Storage, which enables us to launch the script from any machine, and watch the results from our laptops. The results are continuously uploaded, so we don’t have to wait for all the experiments to finish to start inspecting the results. The models themselves are also saved. ### So what have we learned so far? This first implementation was the most important one. Being simple, the script didn’t do anything smart for you. It didn’t know which experiments to perform - you had to manually define the JSON input for every run. After running the script many times, you start to understand which values work better than others. It’s one of the most important things we got from the script - a more in depth understanding of our models. Using this script for the first time got us a big improvement - the MSE improved by more than 10%. ### Are we sure it’s real? When you run hundreds of experiments, the best ones usually have negligible differences. How can you know it’s statistically significant? One way to tackle this problem, assuming you have enough data, is to train the same model on several date-ranges. If one model is better than the others in all of the date-ranges, you can be more confident it’s real. Let’s say today is June, and you run the script. Here are the date-ranges the script will choose: Note that the script chooses a new set of date-ranges if you run it on a different month. This is important, since otherwise you could accidentally overfit your models due to extensive hyperparameter tuning. ### I don’t have time for this Who has the time to run every experiment three times? It’s nice that you can get reliable results, but it means you’ll end up running fewer different experiments. The next version of the script tackled this problem by supporting a new mode of operation: - Only one date-range is used - The date-range contains only one month of data - Training is limited to fewer epochs But are the results correlated with what we would get if we used more data and epochs? To answer this, we performed some experiments, each using a different amount of data ranging from one week to three months. We found out that one month had a good tradeoff between MSE and training time. (Did you notice we used the hyperparameter tuning script to tune the hyperparameter script? How cool is that?) To investigate what the right number of epochs would be, we analyzed the MSE on TensorBoard. Each plot in the graph represents a different model trained on a different amount of data. After 20 epochs all of the models have almost converged, so it’s safe to stop there. ### C'mon script, do the job for me! At this point we decided the script should choose the hyperparameter values for you. We started with the learning rate related hyperparameters: initial learning rate, decay factor, number of epochs with no improvement for early stopping, etc. Why learning rate? - It significantly affects training time. We should first settle on a good learning rate, both in terms of accuracy and training time, before tuning all the rest - Some suggest it controls the effective capacity of the model in a more complicated way than other hyperparameters, so it might be better to start with that. The new mode of the script uses hard-coded ranges of values that are reasonable for our models. Then came the next demand: The next version of the script did just that: provided with an architecture name, it automatically generates experiments for you. You don't need to specify any hyperparameter values. ### Let’s go random Although being worse than random search in some cases, grid search is easier to analyze: every value is used by multiple experiments, so it’s easy to spot trends. Since we had already gained intuition on what values work better, it was time to implement random search. Doing so helped the script find better hyperparameters. ### Some final thoughts In the research phase of any machine learning project, hyperparameter tuning can be done manually. However, when you want to take the project to the next level, it’s highly effective to automate the process. In this post I described some of the small touches we implemented into the automation process. Some might be useful for you, some might not. Drop a line in the comments if you found any other exciting things to be helpful. --- ### 5 Simple tips for boosting your Jenkins performance URL: https://www.taboola.com/engineering/5-simple-tips-boosting-jenkins-performance/ Last Modified: 2025-01-14 14:37:36 Do you know that feeling when you’ve finished working on a feature, pushed the code, but then your CI system refuses to respond? Lagging or slow responsiveness is very common among Jenkins users, and you can find many reported issues on it. Slow CI systems are frustrating, they make you develop slower and waste your time. I have worked on several Jenkins systems with various versions (1.x and 2.x), tens of slaves and hundreds of builds per day. I have managed to improve the performance of those systems using a few simple guidelines. In the following post I will share 5 tips that can make your Jenkins better, and put a smile on your developers’ faces. ## Tip 1: Minimize the amount of builds on the master node The master node is where the application is actually running, this is the brain of your Jenkins and, unlike a slave, it is not replaceable. So, you want to make your Jenkins master as “free” from work as  you can, leaving the CPU and memory to be used for scheduling and triggering builds on slaves only. In order to do so, you can restrict your jobs to a node label, for example:  In the case of pipeline jobs, put the label when allocating the node, for example: stage("stage 1"){ node("ParallelModuleBuild"){ sh "echo \"Hello ${params.NAME}\" " } } In those cases, the job and node block will only run on slaves labeled with ParallelModuleBuild.   ## Tip 2: Do not keep too much build history When you configure a job, you can define how many of its builds, or for how long they, will be left on the filesystem before getting deleted. This feature, called Discard Old Builds, becomes very important when you trigger many builds of that job in a short time. I have encountered cases where the history limit was too high, meaning too many builds were kept on the filesystem. In such cases, Jenkins needed to load many old builds - for example, to display them in the history widget, - and performed very slowly, especially when trying to open those job pages. Therefore, I recommend limiting the amount of builds you keep to a reasonable number. ## Tip 3: Clear old Jenkins data In continuation to the build data from the previous tip, another important thing to know is the old data management feature. As you probably know, Jenkins keeps the jobs and builds data on the filesystem. When you perform an action, like upgrading your core, installing or updating a plugin, the data format might change. In that case, Jenkins keeps the old data format on the file system, and loads the new format to the memory. It is very useful if you need to rollback your upgrade, but there can be cases where there is too much data that gets loaded to the memory. High memory consumption can be expressed in slow UI responsiveness and even OutOfMemory errors. To avoid such cases, it is best to open the old data management page (http://JenkinsUrl/administrativeMonitor/OldData/manage), verify that the data is not needed, and clear it. ## ## Tip 4: Define the right heap size This tip is relevant to any Java application. A lot of the modern Java applications get started with a maximum heap size configuration. When defining the heap size, there is a very important JVM feature you should know. This feature is called UseCompressedOops, and it works on 64bit platforms, which most of us use. What it does, is to shrink the object’s pointer from 64bit to 32bit, thus saving a lot of memory. By default, this flag is enabled on heaps with sizes up to 32GB (actually a little less), and stops working on larger heaps. In order to compensate the lost space, the heap should be increased to 48GB(!). So, when defining heap size, it is best to stay below 32GB. In order to check if the flag is on, you can use the following command (jinfo comes with the JDK): jinfo -flag UseCompressedOops <pid> ## ## Tip 5: Tune the garbage collector The garbage collector is an automatic memory management process. Its main goal is to identify unused objects in the heap and release the memory that they hold. Some of the GC actions cause the Java application to pause (remember the UI freeze?). This will mostly happen when your application has a large heap (> 4GB). In those cases, GC tuning is required to shorten the pause time. After dealing with these issues in several Jenkins environments, my tip contains a few steps: - Enable G1GC - this is the most modern GC implementation (default on JDK9) - Enable GC logging - this will help you monitor and tune later - Monitor GC behavior - I use http://gceasy.io/ - Tune GC with additional flags as needed - Keep monitoring - Read this post and this one - these are great posts about GC tuning ## Now it’s your turn Jenkins responsiveness issues are very common, but can be dealt with. In one of the environments I worked on, I had a configuration page which took over a minute to open. After implementing the above tips, especially G1GC and old data, this page opened in a split second. If you got this far, you are probably familiar with this issue. So, try my tips and share what worked for you and what hasn’t. There is a good chance your Jenkins will speed up too. You are welcome to tweet me for any question :)   --- ### Using Word2Vec for Better Embeddings of Categorical Features URL: https://www.taboola.com/engineering/using-word2vec-for-better-embeddings-of-categorical-features/ Last Modified: 2025-01-14 14:37:36 Back in 2012, when neural networks regained popularity, people were excited about the possibility of training models without having to worry about feature engineering. Indeed, most of the earliest breakthroughs were in computer vision, in which raw pixels were used as input for networks. Soon enough it turned out that if you wanted to use textual data, clickstream data, or pretty much any data with categorical features, at some point you’d have to ask yourself - how do I represent my categorical features as vectors that my network can work with? The most popular approach is embedding layers - you add an extra layer to your network, which assigns a vector to each value of the categorical feature. During training the network learns the weights for the different layers, including those embeddings. In this post I will show examples of when this approach will fail, introduce category2vec, an alternative method for learning embedding using a second network and will present different ways of using those embeddings in your primary network. ## So, what’s wrong with embedding layers? Embedding layers are trained to fit a specific task - the one the network was trained on. Sometimes that’s exactly what you want. But in other cases you might want your embeddings to capture some intuition about the domain of the problem, thus reducing the risk of overfitting. You can think of it as adding prior knowledge to your model, which helps it to generalize. Moreover, if you have different tasks on similar data, you can use the embeddings from one task in order to improve your results on another. This is one of the major tricks in the Deep Learning toolbox. It’s called transfer learning, pretraining or multi-task learning, depending on the context. The underlying assumption is that many of the unobserved random variables explaining the data are shared across tasks. Since the embeddings try to isolate these variables, they can be reused. Most importantly, learning the embeddings as part of the network increases the model’s complexity by adding many weights to the model, which means you’ll need much more labeled data in order to learn. So it’s not that embedding layers are bad, but we can do better. Let’s see an example. ## Example: Click Prediction Taboola’s research group develops algorithms that suggest content to users, based on what they’re currently reading. We can think about it as a Click Prediction problem: given your reading history, what is the probability that you will click on each article? To solve this problem, we train deep learning models. Naturally, we started with learning the embeddings as part of our network. But a lot of the embeddings we got didn’t make sense. ## How do you know if your embeddings make sense? The simplest way is to take embeddings of several items and look at their neighbors. Are they similar in your domain? That can be pretty exhausting, and doesn’t give you the big picture. So, in addition you can reduce the dimensionality of your vectors using PCA or t-SNE, and color them by specific characteristic. The advertiser of an item is a strong feature, and we want similar advertisers to have similar embeddings. But this is what our embeddings for different advertisers looked like, colored by the language of that advertiser: Ouch. Something is clearly not right. I mean, unless we assume that our users are multilingual geniuses that read one article in Spanish and then effortlessly go and read another one in Japanese, we would probably want similar advertisers in our embedding space to have content in the same language. This made our models harder to interpret. People were upset. Some even lost sleep. Can we do better? ## Word2Vec If you’ve ever heard about embeddings you’ve probably heard about word2vec. This method represents words as high dimensional vectors, so that words that are semantically similar will have similar vectors. It comes in two flavors: Continuous Bag of Words (CBOW) and Skip-Gram. CBOW trains a network to predict a word from its context, while Skip-Gram does the exact opposite, predicting the context based on a specific target word. Can we use the same idea to improve our advertisers’ embedding? YES WE.. well, you get the idea. ## From Word2Vec to Category2Vec The idea is simple: we train word2vec on users’ click history. Each “sentence” is now a set of advertisers that a user clicked on, and we try to predict a specific advertiser (“word”) based on other advertisers the user liked (“context”). The only difference is that, unlike sentences, the order is not necessarily important. We can ignore this fact, or enhance the data set with permutations of each history. You can apply this method to any kinds of categorical features with high modality, e.g, countries, cities, user ids, etc.. See more details here and here. So simple, yet so effective. Remember that messy visualization of embeddings we had earlier? This is what it looks like now: Much better! All advertisers with the same language are clustered together. Now that we have better embeddings, what can we do with them? ## Using embeddings from different models First, note that category2vec is just one example of a general practice: take the embeddings learned in task A and use them for task B. We could replace it with different architectures, different tasks, and in some cases, even different datasets. That’s one of the great strengths of this approach. There are three different ways to use the new embeddings in our model: - Use the new embeddings as features for the new network. For example, we can average the embeddings of all the advertisers the user clicked on, to represent the user history and use it to learn the user’s tastes. The focus of this post is Neural Networks but you can actually do it with any ML algorithm. - Initialized the weights of the embedding layer in your network, with the embeddings we just learned. It’s like telling the network - this is what I know from other tasks, now adjust it for the current task. - Multitask Learning - take two networks and train them together with parameter sharing. You can either force them to share the embeddings (hard sharing), or allow them to have slightly different weights by “punishing” the networks when their weights are too different (soft sharing). This is usually done by regularizing the distance between the weights. You may think about it as passing knowledge from one task to another, but also as a regularization technique: the word2vec network acts as a regularizer to the other network, and forces it to learn embeddings with characteristics that we are interested in. ## OK, let’s recap. Sometimes we can get great results simply by taking some existing method, but applying it to something new. We used word2vec to create embeddings for advertisers and our results were much more meaningful than the ones obtained with embedding layers. As we say in Taboola - meaningful embeddings = meaningful life. --- ### How Re2 Shattered My Bottleneck URL: https://www.taboola.com/engineering/how-re2-shattered-my-bottleneck/ Last Modified: 2025-01-14 14:37:37 One pleasant morning I got to work, thinking this day couldn’t get any better. But as Murphy would have it, there was my boss walking frantically toward me. It turned out that almost over night one of the main data pipeline systems had become a major bottleneck for the company, and a solution was needed, Fast! Usually in a startup, let alone a company moving as fast as Taboola, these things can occur on a weekly basis. I needed to find some quick wins to relieve some of the bottlenecks inside the system. Luckily Re2 was there to the rescue - in this post I will share how to find the bottlenecks using Gprof2dot beautiful image rendering, and of course, what Re2 is and how to use it. * Note that this article addresses a pain I had in a Python framework, but because there are Re2 implementations to all major languages you should still find value reading this article. ### The strategy to relieve bottlenecks is pretty straightforward Step one: Recognize were it hurts the most Step two: Reduce the pain Step three: Iterate back to step one until you get the desired result Sounds simple enough, right? Well it is, in theory. Usually the implementation is where it gets you. ### Step one: Recognize where it hurts the most Like anyone who ever saw a guy talking about A.A. meetings on TV, will say: “The first step in dealing with your problem is recognizing that it exists”, in our case that meant measurements and finding the bottlenecks. The system in question is a pretty straightforward ETL system written in Python, so I decided to start by looking at the transform part first. I spent a couple of hours extracting the main part of the code into a single independent process, and when it was done I started a Jupyter notebook http://jupyter.org/. Using cProfile built-in python package and gprof2dot, I ran the following code to test it - This is the image Gprof2dot produced: A short explanation if this is the first time you encountered the Gprof2dot library: The image that I got was a profile that measured our code, and determined, for each major block, how much time our code visited that block and how much time in total each block contributed to the overall running time. The dark blue ones are the “cold” areas, the interesting parts in our investigation. The path starting from the red box, and “cooling down” all the way to the leaves are the interesting paths to investigate. Each box has the following format: In the next image below, I looked at the leaf at the end of the path. If you look at the upper box, you can see that it’s a method called doesTextFit in class GenericTextInTextSearch. The system spent 32.05% of it’s runtime in that code, but only 4.94% of it is in the actual code of the method, the other ~95% of the time was spent in code called from doesTextFit. The 2,221,386 is the amount of times this code was called. The more interesting part is that doesTextFit calls another method, where it spent ~26% of the time. That was the method I decided to focus my attention on. * Note that it might happen in the leaf box, that the self time - in the parenthesis - and the total time - above the self time - don’t add up. This happens when the execution time is too short, causing the round-off errors to be large. Don’t worry about it. “Method ‘Search of’ _sre.SRE_Pattern object“ is from Python built-in library - the Regex library re. It turned out that the system that had a lot of moving parts, spent ~26% of the time on a single method on a single library call. So now I set out to see if I could find a replacement for the Python built-in method. It turns out there is. And I was completely surprised by it. ### Step two: Reduce the pain I came across these two articles: - Python, Catastrophic Regular Expressions - Regular Expression Matching Can Be Simple And Fast They basically told me, I was living a lie - I always thought that Regex is the most efficient way to match a string. It turns out that it might have started like that, but somewhere along the way we sacrificed - in all the major programing languages - performance for features. The salvation was going back to the fundamentals: State Machine Based Regex! ### What’s wrong with the built in Regex engine? TL;DR Backtracking. When the built-in Regex tries to match a string, it follows a greedy algorithm pattern. Trying to match the full string first, and when it fails trying to match minus one characters on the full string and the leftover char separately. Then on the full string minus two characters, and on the two leftover characters separately, and so on and so forth. If there is no match, the engine is destined to fail, it simply doesn’t know that before it iterates through all the different combinations. This process is time and space consuming. ### So what’s the alternative? Introducing Ken Thompson. you can read more about him in his wiki page, but here is the TL;DR for the list of his achievements: - Unix - B (programming language) - ancestor to C - UTF-8 - Go (programming language) - State Machine Based Regex! The last bullet is best represented in this comparison graph: The left one is for Perl, but represents the major programing languages, the right one is the Thompson method. Make sure you notice that the left one is in seconds and the right one in microseconds(!). How did he do that? state machine. Think of a regular expression as a state machine built from the expression you want to match. Each char the engine encounters, propels it to make the next decision, if by the end of its journey it reaches a “final” state. Then the engine found a match. If not, then there is no match. The space complexity of a solution like that, is the amount of different states the regular expression has. And the time complexity is the length of the text being checked. With that in mind, I went out to see if I could find an implementation of the Thompson engine, and I found the good people of Google with their Re2 implementation in C++. To integrate Google Re2 in Python, I had to use a third party wrapper py-re2 , but for Java there is Google Re2/J project or Brics, we in Taboola, use Brics in our Java code, you can read comparison of the two projects in this great blog: brics-vs-re2j. The integration in the code is that simple: It is important to remember that not all the functionality that exists in re, exists in re2. So what I did, and what you should do if you plan to use this solution, is to analyze all your Regex calls, and make sure that you get the expected result. Although the graph above promised a performance boost of hundreds of percent, the performance boost I managed to get was just a ~40%-50% boost. I didn’t complain, it was not too bad for a few days work. ### Step three: Iterate back to step one till you get the desired result I never got to this part, and hopefully, if you use this engine you won’t either… ### No Silver Bullet The specific circumstances of the system i described here, is that the majority of the Regex comparisons finish without a match. So it worked really well to have a new Regex engine that returns “no match” faster than the regular - backtracking - one. You will need to test it in your unique scenario to see how much performance improvement you will get. Second thing to check is the Regex you use, if it needs Backtracking to complete it’s comparison it won’t work, and you will get Segfault. But, it shouldn’t be too hard to rewrite them to a simpler syntax. Third thing to monitor, is the state explosion problem when building deterministic automata, that can happen in some of the Re2 implementations - Brics for example but not in Google Re2. The cost is in performance. The last thing you need to check is the return values, there might be a case that the expected outcome has changed from what you had before. For example to get a null were before you got a False. The bottom line is that you will need to do the same evaluation process that you would have done before upgrading to a new version of a library. Things might change, so you need to be carefull. Having said that, even if it is only a few percent improvement, the fact that it is so easy to integrate makes it a “money on the floor” scenario. Good luck, and feel free to contact me for help and guidance. --- ### DIY: Engineering Your Company Culture URL: https://www.taboola.com/engineering/diy-engineering-your-company-culture/ Last Modified: 2025-01-14 14:37:37 This is a tale of heroism, of overcoming obstacles and hardships. This is a tale of ingenuity, of originality and thinking outside the box. This is a tale... of how I was too lazy to go and look if someone was already playing table tennis in the game room.     Hiking Across the Office is a Drag   Taboola’s Israeli office in Tel Aviv, houses about 350 people, spread over five large floors. The game room, however is smack dab in the middle of them. In smaller companies, if you wanted to know if the game room was available, all you had to do was to look slightly over your monitor and you would have your answer. Here, it takes 60 seconds and 110 steps, including one flight of stairs, to get from my workstation all the way to the game room - believe me, I counted. Unfortunately, due to the unbending laws of physics, it takes exactly the same time to go back, frustrated that it was already being used. And so, after one too many of these time-wasting, context-switching, hope-shattering trips, I started asking myself one simple question - what if we could just ask if someone was already using the game room? #### #### Weapon(s) of Choice - Slack Bot, PIR Sensor and a Raspberry Pi Salvation came in the form of a Passive Infrared (PIR) motion sensor connected to a Raspberry Pi. The sensor data is transmitted to the Raspberry Pi, which in turn saves the data to a local DB. The Slack bot user then queries the DB, to find out if actual movement has been detected in the room recently.   The entire system looks like this: #### I Sense a Disturbance in the Room   A PIR sensor is just a fancy name for your run-of-the-mill motion detector, the ones you usually see mounted on walls of houses. Basically, the way it works is you have a sensor that measures infrared light, radiating from objects in its field of view. The term passive in this instance refers to the fact that PIR devices do not generate or radiate energy for detection purposes, they work entirely by detecting infrared radiation, emitted by or reflected from objects (find more information about PIR sensors here). #### #### Easy as Pi   Raspberry Pi is a series of small single-board computers designed to promote the teaching of basic computer science in schools and in developing countries. It achieves this goal by being highly versatile, easy to setup and most important - very, VERY affordable, starting at only $5 (!) for the most basic model and $35 for the latest model. It has a lot of cool features, but maybe the most important one is the General Purpose Input/Output (GPIO) connector. These pins are a physical interface between the Pi and the outside world. This is what is going to be used in order to connect the motion sensor output to the Raspberry Pi. #### #### A Tale of Two Scripts Raspberry Pi can run all sorts of images on its MicroSDHC card, but the most prevalent one is its own flavor of Unix -Raspbian, which is what I installed to run the Python interpreter and the various dependencies and tools required for this project. The solution was divided into two separate Python scripts, which run on startup - the motion detection and the slackbot script. A local MySQL database is used to store information about the activity in the room, meaning a list of all previous sessions and a current active one, should one exist. This is to allow for persistency and other possible future features, such as to determine when is the optimal time to go and play, or how long someone has already been using the room. The motion detection script listens for movement (or lack thereof) and updates the database accordingly - opens a new game session if one is not already opened, extends active sessions and closes sessions after a certain period of non-movement. The slackbot script handles Slack integration and queries the database to provide information about the room’s status, by checking if there are currently any active open sessions. #### #### Halt! Who Goes There?   I used an HC-SR501 motion sensor, which you can get online for $1. It only has three pins that you need to connect to the Raspberry Pi - GND (Ground), VCC (5V) and OUT (Data) - follow the diagram above to connect these (see the pin layout here). The motion detection script communicates with the GPIO channels by listening on the data pin - marked as OBSTACLE_PIN in the code. This is done utilizing the RPi.GPIO package, which allows you to access the various pins on the board and get their current state (see documentation here). Once the motion detection script starts running it goes in an endless loop, which is divided to one second intervals, and in each iteration the sensor input is read. Querying the sensor gives a binary response - a movement was detected or not. Since the motion detector samples motion every second, it needs to be able to determine that an actual movement occurred not only in a particular second, but several times during a short period of time, to avoid false positives, i.e. detecting movement when there was none, and false negatives, i.e. not detecting movement when it should have. So, in every iteration a “detection frame” is examined, which is basically a time window in which the number of seconds a movement was detected is counted. The number of seconds with movement, out of the total number of seconds in the time window, is compared to a predefined threshold. If the number exceeds the threshold, then a movement was detected in the room. This allows control of the sensitivity of the detection, in addition to being able to adjust it physically on the sensor itself. The following code demonstrates this logic: # The physical pin number on the board connected to the DATA output pin of the motion sensor OBSTACLE_PIN = 7 # The maximum time in seconds allowed for a session to be inactive MAX_SESSION_TIME_IN_SECONDS = 60 # The time window in seconds we are looking for consecutive movement DETECTION_FRAME_LENGTH_IN_SECONDS = 10 # The threshold that determines if a movement was detected DETECTION_INCIDENTS_THRESHOLD = 0.4 def setup(): print_log('initializing sensor') GPIO.setmode(GPIO.BOARD) GPIO.setup(OBSTACLE_PIN, GPIO.IN) print_log('initializing db') time.sleep(10) # preparing tables, cleaning existing open sessions setup_database() def is_motion_detected(): return GPIO.input(OBSTACLE_PIN) def loop(): iteration_counter = 0 detection_counter = 0.0 print_log('started monitoring') # start monitoring for movement while True: # if the detection window is over if (iteration_counter >= DETECTION_FRAME_LENGTH_IN_SECONDS): # if we saw enough movement during that window # i.e. seconds with movement out of total window time is larger than the threshold if (detection_counter / iteration_counter >= DETECTION_INCIDENTS_THRESHOLD): handle_motion_detected() else: handle_no_motion(DETECTION_FRAME_LENGTH_IN_SECONDS, MAX_SESSION_TIME_IN_SECONDS) # reset counters for next detection window iteration_counter = 0 detection_counter = 0.0 # if a movement is detected in this particular second if (is_motion_detected()): detection_counter += 1 print_log('motion ' + str(detection_counter) + ' detected') iteration_counter += 1 time.sleep(1)   #### Slacking Off Below is a code snippet from the slackbot script, outlining the interaction with the aptly named PingPongBot - you will need to add a new bot user to your Slack team (see this link for more information), after which you will receive a token to be used in your client: # these can be obtained when creating a new slack bot user # see instructions on how to do that here - https://my.slack.com/services/new/bot SLACK_BOT_TOKEN = '<your slack bot token here>' SLACK_BOT_NAME = '<your slack bot name here>' # aux method to get the slack bot id, required for authentication def get_slack_bot_id(slack_client, bot_name): if __name__ == "__main__": api_call = slack_client.api_call('users.list') if api_call.get('ok'): users = api_call.get('members') for user in users: if 'name' in user and user.get('name') == bot_name: print("Bot ID for '" + user + "' is " + user.get('id')) return user.get('id') else: print('could not find bot user with the name ' + bot_name) return None # set up connection to Slack client slack_client = SlackClient(SLACK_BOT_TOKEN) BOT_ID = get_slack_bot_id(slack_client, SLACK_BOT_NAME) AT_BOT = '<@' + BOT_ID + '>' # the actual command to look for when addressing the bot in chat EXAMPLE_COMMAND = 'free' # the actual logic when receiving a message in the chat def handle_command(command, channel): response = 'Hello! I am the ping pong bot! Ask me if the table is free by typing - \"free?\"' # parse the chat message and check if it contains the desired command if EXAMPLE_COMMAND in command.lower(): if is_open_session_exists(): response = 'Sorry, someone is playing right now... Try again later!' else: response = 'Free to play! enjoy :)' # return the response in the chat according to the result from the database slack_client.api_call('chat.postMessage', channel=channel, text=response, as_user=True) # aux method to receieve messages from the chat def parse_slack_output(slack_rtm_output): output_list = slack_rtm_output if output_list and len(output_list) > 0: for output in output_list: if output and 'text' in output and AT_BOT in output: return output.split(AT_BOT).strip().lower(), output return None, None if __name__ == "__main__": READ_WEBSOCKET_DELAY = 1 # establishing connection to bot if slack_client.rtm_connect(): print(SLACK_BOT_NAME + ' connected and running') # start listening for messages while True: command, channel = parse_slack_output(slack_client.rtm_read()) if command and channel: handle_command(command, channel) time.sleep(READ_WEBSOCKET_DELAY) else: print('connection failed, invalid Slack token or bot ID?') You can find the full project code on my github page. #### I Love It When a Plan Comes Together This is what the sensor looks like taped to the floating ceiling of our game room (the Raspberry Pi is hidden from view): And finally, this is how it looks like when you ask the PingPongBot if the game room is free:   #### #### Not Only for the Office Hopefully by now you realize how simple and easy it is to implement an entire motion detection system using accessible and affordable parts. There are many other sensors that you can incorporate in your next project, turning your office/home into a smart one. For example, a temperature and humidity sensor can be used as input to control your AC, or a rain detection module to let you know when to turn off your sprinklers or close the electric windows. Well, writing this post really made me parched. Hmm... I wish there was a way to know if there are beers left in the kitchen...     --- ### Calculating Git Version URL: https://www.taboola.com/engineering/calculating-git-version/ Last Modified: 2025-01-14 14:37:37 Hello Git user. In this blog post I will discuss a technique for a unique version calculation for every Git commit. You may ask why we need this, after all every commit in Git is identified by a unique sha1 hash. That’s right, let’s take 2 commits, 4bd92c9 and f5fc029, use their sha1 hash as a version and perform a simple A/B test. The test showed that 4bd92c9 is preferred to f5fc029. If this is the case, how can we tell: - Which version is newer? - If 4bd92c9 is included in f5fc029, or vice versa? - What branch they were built from? It seems we need an alternative. The common standard for the versioning is a SemVer scheme. We will use its parts as follows: - Major - manual increment - Minor - every released feature will increment the minor - Patch - will always be 0 Now let’s take a look at our Git graph: We will give a version number for every mentioned commit: - M1 - 1.0.0: our init major version - M2 - 1.1.0: some new feature released - M3 - 1.2.0: feature B released - M4 - 1.3.0: feature A released But what about A1,B1? We need to give them a version number as well in order to identify these builds for various purposes like automation test, deploying on GA environment, etc. According to the semver rules we need to mark these versions as a next minor pre-release, so we get: - A1 - 1.1.0-A-5 - B1 - 1.2.0-B-3 Let’s put the version numbers in the graph: Now let’s automate it by using the following logic. If you are on the master branch calculate the version by: latest=$(git tag -l --merged master --sort='-*authordate' | head -n1) latest=$(git tag -l --merged master --sort='-*authordate' | head -n1) semver_parts=(${latest//./ }) major=${semver_parts} minor=${semver_parts} patch=${semver_parts} version=${major}.$((minor+1)).${patch} Put the tag on a HEAD If you are on the feature branch just calculate the version (without putting a tag) by: latest=$(git tag -l --merged master --sort='-*authordate' | head -n1) latest=$(git tag -l --merged master --sort='-*authordate' | head -n1) semver_parts=(${latest//./ }) major=${semver_parts} minor=${semver_parts} patch=${semver_parts} branch=$(git rev-parse --abbrev-ref HEAD) count=$(git rev-list HEAD ^${latest} --ancestry-path ${latest} --count) version=${major}.${minor}.${patch}-${branch}-${count} Combining all these together we get the following script: #!/bin/bash branch=$(git rev-parse --abbrev-ref HEAD) latest=$(git tag -l --merged master --sort='-*authordate' | head -n1) semver_parts=(${latest//./ }) major=${semver_parts} minor=${semver_parts} patch=${semver_parts} count=$(git rev-list HEAD ^${latest} --ancestry-path ${latest} --count) version="" case $branch in "master") version=${major}.$((minor+1)).0 ;; "feature/*") version=${major}.${minor}.${patch}-${branch}-${count} ;; *) >&2 echo "unsupported branch type" exit 1 ;; esac echo ${version} exit 0 --- ### Optimizing a Homepage Widget with AB Testing URL: https://www.taboola.com/engineering/optimizing-a-homepage-widget-with-ab-testing/ Last Modified: 2025-06-30 11:00:22 As part of the optimization team at Taboola, we are constantly working with publishers and conducting A/B tests, to find the best user interface (UI) for our widget. In doing so, we improve sponsored content (SC) revenue per mille (RPM). When we dealt with a very large news site in India, as part of an ongoing optimization, we needed to get a bit creative. We were already implementing our best practices for our widget on the publisher homepage — so we came up with a different solution. We tested a native UI, one that would help blend the widget into the design of the page. ### User Behaviour on the Homepage Unlike article pages, where users come to read an article, and then sometimes leave after finish reading it, a user who enters a homepage directly will sometimes scan it from top to bottom to find something to read. When they finish reading everything interesting a Taboola widget with additional reading material at the bottom of the homepage could be a good idea. This behaviour of scanning the homepage until reaching the bottom of the page, can be related to our brain's tendency to need to finish what it starts, as Russian psychologist Bluma Zeigarnik talks about in a 1927 study. Our widget location on the client’s homepage was the 3rd row from the bottom of the page — a sometimes unattractive location that can be easily missed when scrolling down. So, our first step was to test placing our widget at the far bottom of the page. That way users will see it when they reach the bottom. This generated a conclusive 4 percent uplift in SC RPM — but we didn’t stop there... ### AB Testing a Dark Background Atop the new location, there was a photo gallery surrounded by a dark background. Since a dark background on a white page causes high contrast that attracts the eye, we decided it could be interesting to add the dark background to our widget as well. We wanted to mimic the exact UI, font style, text color, etc. This spiked the SC RPM by 40 percent. ### Trying Out a Hero Image After a few months, we were suddenly faced with a new challenge — The client changed the entire homepage design in order to bring in more content cards. The dark gallery above was completely changed, and our widget was a complete stranger on the site. Since mimicking site design had worked previously, we again A/B tested a design, like the unit that was placed above ours — a unit with one large image, commonly referred to in the industry as a Hero image, and 6 smaller images to the right. There was also a subtle gray shadow around it. This generated a 20 percent uplift. ### Size matters? Adding a control variant After seeing the SC RPM increase, and examining the homepage layout, we realized we had missed an important step along the way. By mimicking the Hero unit, we increased the size of our widget by almost 50 percent. So, what caused the RPM uplift? The size, or the UI? If we tested Taboola’s best practice widget, but with the same size as the Hero unit, would it perform better? To measure this, we added two additional variants: A regular 4x2 widget — almost the same size as the Hero unit — and a variation of it, with a light shadow surrounding the widget, mimicking the shadow surrounding the client’s unit. Below is an example: #### Variant 1: 4x2 widget - generated 6 percent uplift #### Variant 2: 4x2 widget, with surrounding shadow to mimic the client’s unit - generated 7 percent uplift Both variants, though successful in results, gave a smaller uplift compared to the variant that looked most like the site’s design. This proved that it wasn’t just increasing the size of the widget that helped. ### Outcome: After viewing the results, the client agreed to change the previous, dark background unit to the new Hero design. Before: After:   After keeping the old UI on 5% traffic in order to measure the long term effect of the new UI, we still saw a conclusive and positive uplift of 12% ### AB Testing Conclusion Using A\B testing to compare between two different designs is an amazing tool that can help Taboola explore and identify what works best for publishers and what should be avoided, thus improving performance. Don’t be afraid to take chances and test new and innovative designs. You’ll never know unless you try. --- ### Scaling Out Jenkins Based CI with Docker and Nomad URL: https://www.taboola.com/engineering/scaling-out-jenkins-based-ci-with-docker-and-nomad/ Last Modified: 2025-01-14 14:37:38 The existing CI processes in Taboola are quite demanding - a full product build includes 150 maven modules accounting for about 20000 unit tests. Beside running builds from master branch there are also builds for all feature branches and patch releases. All in all accumulating to more than a hundred builds per day. Some of the executed tests are pretty heavy on CPU and memory, performing resource-intensive data crunching. As you can imagine - all this requires substantial CI infra horse power. Which it definitely possesses: Taboola’s Jenkins cluster currently has 35+ Jenkins slaves, each with 20 to 40 CPU cores and at least 100 Gb memory. Each slave runs 5 to 10 executors. All in all - a powerful CI/CD factory. But even with all this power - there are limitations. On the day of the weekly release the volume of builds peaks and we sometimes find ourselves starving for resources. Builds line up in queues wasting valuable developer time and causing unnecessary stress. Further enhancing the problem is the fact that some of the build slaves have specific system configuration and can’t be used for all kinds of processes. We love and nurture our pet servers but treating them as cattle is so much more productive. So there’s room for improvement! Add to this the fact that Taboola has been growing fast and we can only anticipate the build volume to continue accelerating in the future - and finding that improvement becomes a necessity. Taboola development infra (should we say DevOps?) team is constantly looking at ways of enhancing the overall productivity and eliminating bottlenecks in the delivery process. Smart test execution algorithms have been developed breaking down the tests into variable size chunks to optimize the load distribution. But we didn’t want to stop there. On a quest to maximize effective resource utilization we started looking at using containers. It’s no secret that in the last couple of years they’ve become the single hottest piece of technology everyone is raving about. They bring all the benefits of workload portability, resource isolation and deterministic deployments. And yes - they are a perfect match for CI/CD - allowing to quickly spin up and tear down pristine and versioned build and test environments. Moreover - they can help with turning our Jenkins slave cluster into a real herd of cattle, where all we need from the slave nodes is to have Docker installed - all the rest of configuration will be taken care of by container images. As with all improvement initiatives involving new tech - we started out small. Installed the Jenkins Docker plugin and set out to create the build slave image equipped with all the needed tools. The physical build slaves are managed by Foreman+Puppet - so puppet manifests were a great help in getting this configured fast. Once the slave image was ready we uploaded it to Taboola’s Artifactory integrated Docker registry and tried running builds from Jenkins with the use of Jenkins Docker plugin. Another day of tweaking was spent trying to get the ssl certificates working but in the end we achieved great success - the build and tests were passing! Unsurprisingly it took hours to complete the full cycle as at this stage we were only using one physical Docker host for our games. As depicted in the following diagram: Now that we’ve proven the build could be executed in a container the time came to decide on a container scheduling solution. Container schedulers (sometimes also referred to as orchestrators) are responsible for receiving container workload tasks and distributing them across a cluster of container hosts. A number of solutions exist with the most renowned being Docker Swarm (by Docker Inc), Kubernetes (from Google) and Marathon (from Mesosphere). But we didn’t go with one of those. Instead we chose to try a tool that comes from a company we’ve learnt to rely on - Hashicorp. Taboola has been using their service discovery tool - Consul in production for quite some time, and we heard good things about their scheduler/cluster manager named Nomad. Beside coming from Hashicorp, it also boasts the following attractive features (as stated on the official website): - Flexible Workloads: Nomad can schedule containers, but also standalone applications and batch tasks across a cluster of hosts. - Operational Simplicity: Nomad ships as a single binary, both for clients and servers, and requires no external services for coordination or storage. Nomad is distributed and highly available, and combines resource management and scheduling into a single system for simplicity. - Built for Scale: Nomad was designed from the ground up to support global scale infrastructure. Nomad is distributed and highly available, using both leader election and state replication to provide availability in the face of failures. Nomad is optimistically concurrent, enabling all servers to participate in scheduling decisions which increases the total throughput and reduces latency to support demanding workloads. The decision to go with Nomad was taken in collaboration with Taboola’s production operations team. They were also looking for a cluster management solution and it was decided we will all be better off by joining our research efforts. Additional motivation was provided by this blog post which showed us integration between Nomad and Jenkins was already taken care of: http://www.ivoverberk.nl/scalable-ci-cd-with-nomad-and-jenkins/ All we had to do is connect the dots! We’ve created a Nomad cluster, installed the Jenkins plugin and tried to schedule our first Nomad-based docker slave. Now it looked like this: But of course it’s never that easy. One of the issues we had to deal with was the fact that current builds rely on 1) shared workspaces and 2) configuration written to an NFS mount. Nomad in its current configuration doesn’t allow mounting specific host folders into containers. This makes some sense - if you want real dynamic cluster scheduling you don’t want to depend on data sitting on the host. But that also meant we had to mount the NFS from inside the container at bringup. While fully possible in Nomad - this wasn’t doable through the Jenkins plugin. But that’s what’s so great about open source. You need functionality - you’re free to build it! We extended the plugin to support alternative container bringup commands, verified it was working and submitted a pull request to the official Github repo. The maintainer - Ivo Verberk, was quick to react - he merged the change, added some improvements of his own and released a new version. In the meantime we encountered a new issue. Containers weren’t getting scheduled fast enough. Sometimes Jenkins would idly wait for up to a couple of minutes before deciding to schedule a new slave. Now multiply this by 50 test jobs and you find yourself waiting for hours… The investigation of Jenkins slave scheduling strategy wasn’t easy - it is not really documented and the examples online are rare and sparse. The real breakthrough came from reading the code of Yet-Another-Docker-Plugin. It revealed that there is a better way - extending Jenkins NodeProvisioner.Strategy . A couple of days of tweaking and voilà - we had slaves coming up in seconds, not minutes - just as one would expect from containers. This has also been merged upstream and is now part of jenkins-nomad-plugin release 0.3 So the end result is a cluster of 4 Nomad nodes capable of running 15-20 Jenkins slaves whenever needed. We’re still sorting out some stability and performance issues with this setup. As mentioned - Taboola testing suite is very CPU and memory-intensive and we occasionally see some jobs strangling others when scheduled on the same node. Probably comes down to careful resource quota tuning. We still need to find the perfect balance between stability and efficient resource utilization. Of course any tips on this will be most welcome. The conclusion: Nomad and Docker are still young technologies. Together they can provide great benefits for your software delivery tooling, but there’s tweaking involved. As always with open source - you can enjoy it the most if you’re prepared to give back. We’ll be happy if you decide to use Jenkins with Nomad and benefit from our contribution to the plugin. Technology grows faster with community support - so do let us know if we can help you, or if you can help us. Building software together is much more fun. After all that’s what DevOps is all about. Happy delivering! --- ### More than one Graph - Code Reuse in TensorFlow URL: https://www.taboola.com/engineering/more-than-one-graph-code-reuse-in-tensorflow/ Last Modified: 2025-01-14 14:37:38 Large production pipelines in TensorFlow are quite difficult to pull off. Training small models is easy, and we mostly do this at first, but as soon as we get to the rest of the pipeline, complexity rapidly mounts. One reason is that the "Computation Graph" abstraction used by TensorFlow is a close, but not exact match for the ML model we expect to train and use. How so? Typically a model will be used in at least three ways: - Training - finding the correct weights or parameters for the model given some training data. Often done periodically as new data arrives. - Evaluation - calculating various metrics during training on a different data set to evaluate training quality or for cross validation. - Serving - on-demand prediction for new data There could be more modes. For example we could re-train an existing model or apply the model to a large amount of data in batch mode. While the conceptual model is the same, these use cases might need different computational graphs. For example, if we use TensorFlow Serving, we would not be able to load models with Python function operations. Another example is the evaluation metrics and debug operations like `tf.Assert` - we might not want to run them when serving for performance reasons. Turns out we need 3-5 different graphs in order to represent our one model. The are a couple of ways to do this, and picking the right one is not straightforward. ## Can't we just build a graph and update it as we go? TensorFlow graphs in Python are append-only. TensorFlow operations implicitly create graph nodes, and there are no operations to remove nodes. Even if we try to overwrite with tf.Session() as sess: my_sqrt = tf.sqrt(4.0, name='my_sqrt') # override my_sqrt = tf.sqrt(2.0, name='my_sqrt') #print all nodes print sess.graph._nodes_by_name.keys() TensorFlow will just add a suffix to the operation name: What's done cannot be undone, so to speak. So what can we do? We can try to create more than one graph. ## Creating multiple graphs with the same code This is the method that we usually find in the documentation. A dropout example might look like this: if is_training: activations = tf.nn.dropout(activations, 0.7) TensorFlow even ships with tools like `tf.variable_scope` that make creating different graphs easier. This approach has a big drawback however - the serialized graph can no longer be used without the code that produced it. Even a small change (like changing a variable name) will break the model in production so to revert to an older model version, we also need to revert to the older code. This is not always practical with larger repositories and in any case requires some operations effort. In addition, training and evaluation don't use the same graph (even if they share weights) and require awkward coordination to mesh together. So we still want one graph, but we want to use it for both training and evaluation. And maybe serving... But definitely training and evaluation. ## Using one graph with conditional logic TensorFlow does have a way to encode different behaviors into a single graph - the `tf.cond` operation. is_train = tf.placeholder(tf.bool) dropout = tf.nn.dropout(activations, 0.7) activations = tf.cond(is_train, lambda: dropout, lambda: activations) The big advantage is that now we have all of the logic in one graph, for instance, we can see it in TensorBoard. Now our serialized models work for training and evaluation. There are two sources of complexity that make the picture less rosy though - laziness and queues. ### Laziness and the conditional operator Let's say we have an expensive operation we would only like to run during evaluation. If we put it behind a conditional operator, we would expect it to only run at evaluation time. This is also true if we mutate something as result of a condition such as the case with batch normalization. This is not always how it works in TF. The `tf.cond` operation is like a box of laziness, but it protects only what's inside. So this code works correctly: # Good - dropout inside the conditional is_train = tf.placeholder(tf.bool) activations = tf.cond(is_train, lambda: tf.nn.dropout(activations, 0.7), lambda: activations) But this will run the dropout even if `is_train == False`. # Bad - droupout outside the conditional evaluated every time! is_train = tf.placeholder(tf.bool) do_activations = tf.nn.dropout(activations, 0.7) activations = tf.cond(is_train, lambda: do_activations, lambda: activations) ## Queues and the conditional operator Queues are the preferred (and best performing) way to get data into TensorFlow. Typically, train and evaluation will be done simultaneously on different inputs, so we might want to try the approach above to get them into the same graph. tf.cond(is_eval, lambda: tf.train.shuffle_batch(eval_tensors, 1024,100000,10000), lambda: tf.train.shuffle_batch(train_tensors,1024,100000,10000)) It doesn't work however. TensorFlow would inform us that `operation has been marked as not fetchable` and crash. The issue is that TensorFlow, does not allow us to enqueue conditionally. However, `shuffle_batch` operation creates the queue and dequeue operation together. To avoid this we need to split the operation into a conditional part that creates the queue, and conditional part that pulls from the correct queue. Here is an example (the full code is quite long so I only left the relevant parts) def create_queue(tensors, capacity, ...): ... queue = data_flow_ops.RandomShuffleQueue( capacity=capacity, min_after_dequeue=min_after_dequeue, seed=seed, dtypes=types, shapes=shapes, shared_name=shared_name) return queue def create_dequeue(queue, ...): ... dequeued = queue.dequeue_up_to(batch_size, name=name) ... return dequeued def merge_queues(self, is_train_tensor, train_tensors, test_tensors, ...): train_queue = self.create_queue(tensors=train_tensors, capacity=... ) test_queue = self.create_queue(tensors=test_tensors capacity=... ) input_values = tf.cond(is_train_tensor, lambda: self.create_dequeue(train_queue, ...), lambda: self.create_dequeue(test_queue, ...) So we can get what we want with the conditional operator, but the code is more complex and harder to understand. Operations should be easier though - we have simple serialized graphs and monitoring. Could we avoid conditions entirely and somehow work around the append-only limitation? ## Working with saved graphs Most pipelines serialize graphs, if only for serving. One very important thing we can do with model is to serve serialized representations. TensorFlow Serving is outside the scope of this post, but the general idea is that to get a full featured server we just need to run: bazel-bin/tensorflow_serving/model_servers/tensorflow_model_server --model_name=my_model --model_base_path=/my/model/path Running our training graph in TensorFlow Serving is not the best idea however. Performance is hurt by running unnecessary operations, and `tf.py_func` operations can't even be loaded by the server. Luckily, the serialized graph is not like the append only graph we had when we started. It is just a bunch of Protobuf objects so we can create new versions. As an example, below is a simplified and annotated version of the `convert_variables_to_constants` function in `graph_util_impl.py` that (unsurprisingly) converts variables into constants. It's useful because this can be faster when serving in some cases. def convert_variables_to_constants(sess, input_graph_def, output_node_names, variable_names_whitelist=None, variable_names_blacklist=None): inference_graph = extract_sub_graph(input_graph_def, output_node_names) ... # Here we find the variable we want to convert for node in inference_graph.node: if node.op in : # Compute a list of variables found ... #Here we create a new graph with the variables replaced by constants output_graph_def = graph_pb2.GraphDef() ... for input_node in inference_graph.node: output_node = node_def_pb2.NodeDef() if input_node.name in found_variables: # Make output_node into a new constant with the variables weights else: output_node.CopyFrom(input_node) output_graph_def.node.extend() print("Converted %d variables to const ops." % how_many_converted) return output_graph_def TensorFlow actually ships with a few ways to manipulate saved graphs. An example is the `quantize_graph` tool and the `freeze_graph` tool which uses the code in the example above. You can use them if they fit your needs, but make sure that they work with your serialization format. ## Serialization formats At least at the moment (TF is at 1.2), there are many different serialization formats used by TensorFlow and serialized graph manipulation is quite complicated. The formats currently used are: - `GraphDef`- only defines the graph structure without weights or serving considerations. Graph freezing code is written for this format. - `Saver` - adds operations to load weights from files to a Graph - `SessionBundle` - adds a signature for serving most graph manipulation code is written for this format - `SavedModel` - the future of graph storage (for ML models). Allows versioning, signatures, the works. Support for manipulation is not great for now. As of Tensorflow 1.2, most graph manipulation tools work with `GraphDef` objects and serving works with `SavedModel` objects. Older versions use `SessionBundle` more. Documentation usually points to the correct input format. ## Our approach here at Taboola In order to simplify operations, and make experiments easy to reason about, we try to avoid the code-model dependencies. Our general strategy is to create a super-graph that can train, evaluate, already at at train time. For code reuse between train and evaluation we use conditional operations, and we prepare the graph for serving using serialized graph manipulation. Online experiments are a big part of our workflow, and we have serving machines around the world, so lowering operational complexity is well worth the effort. This approach has worked well for us. So far we have shipped two big pipelines, training, evaluating and deploying dozens of models. --- ### Choosing a Reliable Synthetic Monitoring Solution: A True Story URL: https://www.taboola.com/engineering/choosing-a-reliable-synthetic-monitoring-solution-a-true-story/ Last Modified: 2025-01-14 14:37:38 Synthetic monitoring is something that we all do. It’s almost something that you don’t think about. You set up a monitor and it just tells you if the service is up or down, most times with just a simple GET. There are the giants in this field (lately consolidated under the Keynote brand as part of AppDynamics) and the new comers like Catchpoint, ThousandEyes, Pingdom (now part of SolarWinds) and WorldPing. All solutions have the same basic concept, pull website information from different agents around the world and provide visibility for the web site operator on uptime, response times and other metrics. But what happens with you have a failure, and no alert? These tools have become so widespread and have such long usage history, that it almost seems pointless to compare. This is a solved problem, no? Just take the cheapest one out there and you're done. Here at Taboola Engineering, we decided to take a second look, as we had some glitches with our monitoring provider and as with every IT issue, you want to learn and improve to better the service for the next time. This is the journey we took to migrate off the big brands. The story starts in mid 2015, I was on the train to meet friends after hours and got a call from the VP Marketing. “Is our web site down?” he asked me. My initial response was no, obviously not, how could it be? Not one alert fired, not one call from the NOC, and not one PagerDuty incident was created; surely the site must be up. But, as with all such calls, the easiest thing was to just browse the web site from my phone. Now at this point, you know the feeling, the train rattles on, the scenery around you blurs and field of vision narrows down to the small screen of the phone in front of you. And the phone screen stubbornly remains white. The website isn’t loading and with each second that ticks by you know the site is down, and the monitoring failed you. I brought back the phone to my ear and told the marketing exec, “yes, you're right”, something is wrong, although I don’t know yet what that is. Time to dive deep and bring the site back. “Thanks for calling, I’m on it” was the closing line and the lead into the troubleshooting session. My first move was to login to our monitoring service, one of the old respectable services. It was immediately clear that the site was down, and for more than a few hours. The fix was easy (the problem was with a bug in the landing page code, an easy roll back) and within 15 minutes the site was back online. But why didn’t I get any alert? Looking into the configuration of the monitoring system, it was not immediately clear what had gone wrong. The configuration is long, the GUI is very old and the system is sluggish at best, but hey, this is the de-facto industry standard for web monitoring. Eventually, with help from my team we found that the default configuration for alerting reads that most types of errors (DNS, connection timeout, error 500 and many many more) should send an alert, except one - “misc error”. It seems that the service took the failure to load a page and match text as a miscellaneous error, so that is the only alert that doesn’t get sent via email or API. Missing the alert was the tipping point for our search in the path to find simpler and faster synthetic monitoring services. I am not saying in any way that missing the downtime is to be blamed on the monitoring service. We could have better engineered the system. Checking for lower boundary and not only top boundary alerts. Adding to that a simple one statistic of uptime and pulling it with an API call to a collection system or some internal observability system. But there is also something to be said for over engineering a solution. The simple fact is that we were looking to offload this task to an external provider, and we were now in the search for some solution that would be simple, fast, flexible, cater to a simple API, have a plugin or add-on community (or market) and above all would find incidents faster. At the time, worldPing were just at closed beta stage but it was clear that there is something different in this products approach. Simplicity, efficiency, full API integration, no compromise GUI. ### The New Approach - Short Intervals The first thing to notice is the very low interval of the testing cycle. We now had a service that can test our site and SaaS in 10 second intervals. The ability to get 20 agents pulling every 10 seconds provides many data points and the ability to alert on them very fast. At the basic level, if you wish to wait for 3 agents to fail 3 times for an alert, you need only wait for 30 seconds. For services that wish to be highly available shortening the time of alerting is an important place to shave off valuable time in the MTTR. (obviously it is best to be proactive and resolve issues before they develop to downtime). This also allows for fast updates when issues are resolved. When running a troubleshooting session and you want to see if a change worked, or at least made a difference, there is very little time to wait, as the metrics come in fast and the system can display them in good timing. Test interval of 10 seconds per agent: ### The New Approach - Easy API Access Next is the API access. As we already have our solution for dashboards (Grafana) there was little to no work for integration, but nonetheless, pulling stats from the service is easy. Providing a good source of information for multiple internal clients to view the status of the website or service was a great win. Other integrations for automatic alerting directly for the person on call was the next logical phase. Bringing issues to the attention of the on-duty engineer further helps reduce MTTR. So the fact the alert comes in fast is the first piece, and reaching the right person is the second. ### Testing the Product For the next 7 months we tested the solutions neck to neck, alerts were sent in parallel, both systems updated with the new endpoints to be monitored. It was clear from the first alert that one system is fast and the other is not. No false alarms were sent, no real event missed. But changing synthetic monitoring systems is something that one doesn’t take lightly. We have been using the legacy system for many years, and personally I have been using them for over a decade. When the renewal for the legacy system came in, it was clear that this would not happen. We have not logged into the system in months and have moved all our troubleshooting sessions and visibility to the new system (worldPing). One specific feature stood out - Scatter Plot. While line graphs are all nice and pleasing on the eye, there is a lot of information missing. The more data you have, the harder it is to render the scatter plot and some of the providers put a cap on the data allowed in the graph (truncating it or right down denying access to more than 5 days). With WorldPing we found the response time of the scatter plot pleasing and the ability to customize any report, and the scatter plot among them a power data visualization tool. 15 minute graph: 24 hour graph: 7 day graph: We now have a fast responding SaaS provider with coverage for all the features we were looking for: - Fast test intervals on multiple regions of agents - The ability to add private agents and add to the data set our own measurements (internal or external) - Main HTTP protocol path covered (network connectivity, DNS lookup, HTTP and HTTPS) - Keyword search in the results - Custom header for the agent, to filter out in the logs - Fast alerting While you might want additional “full browser” tests for some cases, for our business this was more than enough. The flexibility of the solution with the ease of use and speedy setup allowed answered most of what we needed. With the team working in Grafana on a daily basis the actual use of the tool was much higher, as you now have the same user interface as Grafana. The legacy solutions were measured and found wanting. No more high pricing for synthetic solutions that when we had a downtime didn’t alert. No more solutions that take 5 min to send off the alert for of an issue. Now I know of issues way before I get the call, most times, I know of them before they become a major issue. --- ### Testing with Selenium: Covering an Enterprise Web Application URL: https://www.taboola.com/engineering/testing-with-selenium-covering-an-enterprise-web-application/ Last Modified: 2025-01-14 14:37:38 ### Regression Testing a Complex UI Web applications are a mischievous bunch. When they are born they are usually small, clean and orderly, but they may grow up to become complex and error prone monsters. Each feature added to the mix increases the chance of a new bug appearing, a promise that is usually fulfilled. When developing a large and complex web application, we need to be able to continually check regressions and verify that everything that worked until now is still working. So that at least we won’t break more than we need to. Taboola is are a web company, and must deliver on a very rapid pace. We ship many features on a continuous basis. Under such circumstances no matter how big your QA team is, it will never manage to cover the entire system for every delivered feature. This means an extensive testing automation framework is a must. Entering the stage: Selenium Selenium is a robust browser automation framework, allowing us to write web tests with many different languages: Java, C#, Python, etc. Simply said: it provides an API that allows us to simulate user behavior on browsers. Selenium supports all major browsers, and provides a relatively simple API to test everything UI. ### Testing in Taboola Taboola’s publishers and advertisers use a large enterprise web application called Backstage. Backstage is mainly composed of forms and reports (charts and tables). A peek into a Backstage formBackstage has grown over time. Features are continuously being developed and deployed each week, many of these by different teams that may inadvertently break each other’s functionalities. At each change we need to make sure we haven’t broken anything. Even with QA teams working around the clock, we need to have the best coverage possible. Sometimes a small fix in one report can create havoc on a different form, and we need to catch that in time. So how can we test this behemoth of an application? To cover as much as possible, we use different test frameworks: - Mockito for server unit tests - Karma for JS unit tests - Wraith for visual (screenshot comparison) testing - Selenium for client functional and acceptance testing - REST-assured for API testing We strive to cover each possible flow that may occur from user interaction with Selenium. Testing the flows actually becomes part of our server testing, since it validates all integration points. Since it’s almost impossible to cover everything, we cover most of the relevant use cases. Testing coverage is then increased over time with these rules: - As part of fixing each bug, a developer has to add a coverage test verifying the same bug won’t return - As part of developing a new feature, a developer has to add tests covering the new behavior. The feature is only approved for production after manual QA and all Selenium tests have passed Over time this has helped us create a dense coverage of the Backstage application. ### Taboola’s Selenium Solution The following diagram shows our solution: Let’s examine the different parts. ### The WebDriver Framework To be able to add tests easily, we have created a framework that represents the actual pages of the application. Each element is represented by a Selenium component, allowing tests to interact with the website in our tests, as though we were a user. For Selenium to be able to simulate user behavior, we needed to create a representation of all the components in the site. Each element is represented by a Selenium component, allowing tests to navigate and interact with the UI. We call this the WebDriver Framework. ### Application Components For each web component type we created a Selenium representation that knows how to react and read data from this component. It receives the WebElement Selenium object that identified it on the web page, and uses it as a base for all behaviors. For example, a checkbox component: The Checkbox’ constructor receives the WebElement that wraps the actual element. For example, if a checkbox can be located by its id, it can be passed like this: @FindBy(id=”myCheckboxId”) private WebElement myCheckboxElement; …. Checkbox myCheckbox = new Checkbox(myCheckboxElement); The actual components are combined in the application structure section. ### Structure After we have all the necessary components, we gather them to the different web pages in our application. A Page class represents a single web page, and its different implementations allow us to represent all the pages we have and their relations. ### Writing the Tests Taboola’s Selenium solution is powered by JUnit, providing the power of assertions and test focused development. Their structure is in JUnit fashion, with @Before annotation for test preparation, @After for test tear up, and @Test for the different tests. We have divided the tests into classes, according to the section they cover in Backstage. This is an example of package structure for our tests: - Campaigns <- All Campaign related tests Management <- All Campaign Management related tests Form <- All Campaign Create/Edit related tests Create <- All Campaign Creation tests - Edit <- All Campaign Edit tests - Validate <- All Validation tests (successful and errors) - Misc - Inventory <- All Campaign Inventory related tests - Table <- All Campaign Management Table report tests - Reports <- Other campaign report tests - RSS <- RSS Inventories tests - SelfService <- Self-service Flow tests - approving and rejecting campaigns - Editorial Tools BlockContent - LookBackSettings - ….. Each package has a base class that is extended by all classes in subpackages. These base classes provide all the necessary logic to generate entities related to this section. In order to avoid super long waiting time for tests to conclude, a lot of thought was put into the test preparation stages (the @Before and @BeforeClass methods). Since similar tests are gathered in the same classes, a BeforeClass preparation method generates all needed entities that can be reused. While a Before method takes care of the buildup that can’t be shared (logging in, reaching the page, etc.). To facilitate easy entity generation we’ve created a utility class, DatabaseUtils, which is used for all preparation steps. ### Tips for Writing Selenium Tests After a lot of pain and gain, I’d like to share some important points to be aware of when writing Selenium tests. - Copy/Paste is your enemy. Since many tests can repeat the same flows with small changes, identify all similar flows and bundle them into utility functions. Test code can become very large very fast, and if you’re not paying attention maintainability can become an impossible task. - Don’t use the same entities and data for different tests, if they are being modified during tests, tests will run in parallel and may break each other’s data. Only generate entities that will remain immutable in BeforeClass methods. - You don’t need to test everything. For component or small page element behavior, you can use Karma tests. Selenium tests are slow - use them for user flows and not for every component aspect. - That being said - assert everything possible. While interaction is slow, assertion is fast. Assert existence of elements, their visibility, their states - whatever comes to mind. - Remember that tests take time to run because they emulate user behavior. Aspire to generate lean and specific flows with most of build up done to the database and not through the test itself. - Don’t panic! It’s not specific for Selenium, I just think it’s a good tip for everything you do. ### Summary Selenium is a massive testing framework, and I have only touched the tip of the iceberg in this post. The most important thing to keep in mind when building these tests is: KISS. Keep It Simple, Slugger. You can start by building a partial cover of your web application only, no need to head on with a full blown solution. This will keep some of the bugs at bay, and that’s more than you had before. I would love to hear what hurdles you’ve come across in your implementation, and of course about the successes as well. Do you already implement a Selenium testing framework? What are the lessons you’ve learned from your implementation? --- ## Marketing Hub ### AI Ad Creative on the Open Web: A Performance Advertiser’s Guide URL: https://www.taboola.com/marketing-hub/ai-ad-creative/ Last Modified: 2026-07-13 11:20:52 Performance marketing teams are using open web advertising to discover massive new audiences and cost-effective inventory, as they move beyond oversaturated search and social channels. The open web offers massive scale, but it also brings fragmented ad formats and volume requirements that can quickly stymie a small team. Without the help of artificial intelligence (AI) resources, performance advertising teams can easily fall behind. Incorporating AI ad creative can save time, as well as helping break through creative fatigue and testing effective copy and creative in a fraction of the time, versus strictly manual work. ## What Is AI Ad Creative? AI ad creative refers to the use of artificial intelligence tools to generate, optimize, and score advertising assets. These might include images, video, copy, and any other assets included in a performance marketing AI strategy. Done well, AI ad creative can take an advertising team from using slow, manual design processes to data-driven workflows built for high conversion output. It allows teams to scale in a way that isn’t possible otherwise. AI ad creative tools include AI ad generators that provide all-in-one ad creation, along with video generation, image creation and editing, performance analysis and prediction, copywriting, and more. As heavily used social and search channels show diminishing returns, advertisers are exploring broader options, and AI tools can help performance marketers meet the open web opportunity. The open web — sites beyond the walled gardens of Google, Meta, and others — includes independent websites, news outlets, and other channels with massive reach. AI’s capabilities make it easy for marketers to scale ad creative quickly, which wasn’t previously possible for small, human-only teams. AI can support creating thousands of different ad sizes for these varied placements. ## Gauging the Core Capabilities of AI Ad Creative Generators When deciding on an AI ad creative generator for your particular needs, it’s essential to look at how a particular tool can ingest brand guidelines, then produce ready-to-launch campaigns, with the format types needed. Keep these capabilities in mind when evaluating AI ad creative generators: ### 1. Rapid Visual and Video Production AI features can transform product photos into studio-quality static ads or create dynamic ad variations, such as AI video ads, to ensure synergy across channels and formats. Brands can use these features to produce high-end content without long production cycles or expensive photo shoots. These features can also be useful in creating or editing imagery to fit holiday or event campaigns, or to target a particular audience. ### 2. Data-Driven Copywriting and Hooks High-converting ad copy is a top performance marketing goal across industries, and AI models can be particularly helpful here. AI ad generators can create headlines and CTAs based on high-converting frameworks, then use data to refine and test variations. AI tools analyze the psychological tone of the associated visual, taking into account brand guidelines, then create a cohesive, persuasive message. ### 3. Predictive Creative Scoring Predictive creative scoring helps marketers get to the ideal creative to reach audiences, but it can also save significant budget and time. With predictive creative scoring, AI and ML can evaluate newly generated creative work against historic performance metrics like CTR, brand recall, and others. The AI analyzes the potential impact of the ad. With that data, teams can make sure to only allocate budget to those assets with the best statistical probability of success. ## Beating Creative Fatigue With Infinite Variations Modern digital audiences often look past ads they’ve seen before. This creative fatigue can be a real barrier to success for performance advertisers. AI tools can go beyond the creativity and available time of advertising teams to instantly produce brand-new, data-backed variations of successful ads and campaigns. These quickly created ads can extend campaign lifespans without requiring a lot of extra resources. ## Adapting Ads for Diverse Open Web Placements The open web requires multiple creative formats and sizes, which can easily consume a marketing team’s time. AI is able to automatically resize and reformat work so that the visual in a standard display banner can become a native content widget in seconds. This allows advertisers to adapt quickly and test campaign variations in hours, instead of months. ## Using AI for Competitor Analysis Beyond creating images, video, and copy for ads, AI platforms offer features that can identify and then reverse-engineer top-performing ads across particular industries. These can save a ton of time in competitor research and building a strategy. Consider exploring these capabilities depending on your business and marketing goals: ### Social Media and Sentiment Analysis AI tools can monitor the engagement rates, hashtags, and overall user sentiment of competitors to show what’s resonating. ### Automated Monitoring and Alerts Set up AI agents to continually track competitor websites for changes, then alert at certain thresholds or updates. ### SEO and Content Strategy Analysis AI tools can analyze keyword rankings, content gaps, and traffic sources of competitors to help you make regular updates. ## Exploring A/B Testing Strategies for AI-Generated Assets Using an AI ad generator brings a lot of power to your performance marketing team, particularly when it comes to ad creative testing. It also brings a new surge in the sheer volume of ad variations you can test. It’s easy to get overwhelmed by the possibilities, so make sure you bring a systematic approach to this multivariate testing, following these steps: - Define the goal of which metric you’re optimizing for and a hypothesis of which variant will perform better. - Use a tool like the GenAI Ad Maker to create multiple high-contrast variations of different ad elements, like headlines, images, and CTAs. Or, start with broad visual concepts if you don’t already have a sense of what’s generally successful with your audience. - Run the ads you want to test simultaneously to two equal audiences against a high-performing control version from your advertising platform. As with any A/B testing, focus on just one variable at a time, like the style of the image or the tone of the headline or CTA copy. - Once you have a statistically significant sample size, see what your ad platform or other AI-powered tools can tell you about the winning variant. - Use the AI tool to make smaller optimizations, like color tweaks. - Allocate budget accordingly to the winning ad. ## Integrating AI Into Your Performance Marketing Workflow The possibilities of using AI in performance marketing are nearly endless, and the number of product options on the market can support teams’ goals whatever their size, budget, or industry. Keep these tips in mind when adopting AI and integrating it into your performance marketing workflow: - Prioritize high-quality, organized, and compliant data from the start to ensure on-brand outputs and accurate insights. - Automate repetitive tasks first with AI (e.g., campaign monitoring) to learn quickly and continue automating workflows. - Use a “human in the loop” model: set guidelines for AI usage and review all AI output for brand safety and consistency. ## Measuring the ROI of AI Ad Creatives Many AI ad generation platforms will assist with measuring ad creative ROI, but make sure you’re tracking your preferred key performance indicators (KPIs) when you’re paying to use AI-generated ad creative tools. Some foundational metrics include click-through rate (CTR), cost per acquisition (CPA), and return on ad spend (ROAS). When using AI ad creatives, remember that a core value metric is the reduction in creative production turnaround times. ## Key Takeaways AI capabilities have matured so quickly that performance marketers can now choose from a variety of platforms, depending on their needs. This maturity curve aligns nicely with the potential of open web advertising, which offers scale and incremental growth beyond the walled gardens of search and social. Modern performance marketers have to include AI automation in their strategy to overcome creative fatigue, save time and money, and successfully scale on the open web. ## Frequently Asked Questions (FAQs) ### How does AI improve ad creative performance? Beyond simply making different types of ad creative, AI ad generators are able to analyze vast amounts of historical performance data to predict which visual and textual elements will drive the highest engagement. With this information, ad generators can then create optimized variations for specific audiences. Ideally, you’ll then see increased CTR and lowered CAC. ### Can AI generate video ads as well as static images? Yes, AI ad generators can produce video ads as well as static images to provide consistency across formats. AI-generated videos can be very realistic, with AI avatars, voiceovers, product interactions, and more, all without the overhead expenses of a studio. AI can also write effective copy. ### Does using AI ad creative mean replacing human designers entirely? No, it doesn’t. AI is an efficiency tool for performance advertising teams. It can cut out repetitive and time-consuming tasks like resizing, versioning, reformatting, repurposing, and ideating for testing and campaign expansion. Human designers and strategists can use AI as a tool to support their work of managing and using brand voice, building campaigns and setting goals, and developing creative guardrails for AI tools to follow. ### Why is AI crucial for campaigns beyond search and social? Search and social have served performance marketing teams well for many years, but their oversaturation and cost have become prohibitive. The open web offers less expensive opportunities, but the thousands of featured display and native ad placements require specific asset dimensions and guidelines. Work that would take humans many hours can be done in minutes or less with AI, which scales and adapts one core concept into every specific asset requirement. This makes multi-channel expansion an option for even very small teams. --- ### AI Campaign Optimization: How to Scale Performance on the Open Web URL: https://www.taboola.com/marketing-hub/ai-campaign-optimization/ Last Modified: 2026-07-13 11:12:25 You’ve maxed out your search budget. You’ve squeezed every dollar out of social. And yet, scaling feels harder than it was two years ago — because it is. Walled garden platforms are crowded, and crowded means expensive. That’s where artificial intelligence (AI) changes the equation. By applying machine learning to real-time data across the open web, advertisers can automate bidding, personalize creative, and identify high-intent audiences that the walled gardens of search and social simply can’t reach. The result is a fundamentally different approach to performance marketing, one that doesn’t just optimize for today, but continuously learns, adapts, and improves over time. ## What Is AI Campaign Optimization? At its core, AI campaign optimization means using machine learning algorithms to manage and improve digital advertising without relying on constant human intervention. Instead of manually adjusting bids, rotating creative, or reviewing performance reports, AI-driven marketing campaigns handle all of that automatically, far faster than any team could. This shift matters, as modern digital advertising generates more data than any person can reasonably process in real time. AI campaign optimization moves beyond manual management by continuously analyzing large datasets to predict outcomes and improve return on investment (ROI), making adjustments in the moment rather than waiting for an end-of-week report. What separates this from basic automation is the learning component. Traditional rules-based systems execute fixed instructions: “If cost per acquisition (CPA) exceeds a threshold, pause the ad group,” for example. AI-driven systems work on a continuous loop to: - Sense: Collect live data from connected platforms. - Analyze: Detect patterns, identify trends, and diagnose causes. - Decide: Determine the best optimization action based on your defined goals. - Act: Execute the change, evaluate the result, and feed outcomes back into the loop. That cycle runs without waiting for a human to run a report, and it gets sharper with every iteration. ## Core Capabilities of AI-Driven Campaigns The value of AI in performance marketing comes from a set of functional capabilities that work together. On walled garden platforms, those capabilities are limited by what the platform chooses to share. Those platforms also optimize for their bottom line as much as yours. On the open web, AI operates across a far broader data landscape, executing strategies at speeds and scales no human team can match. ### Autonomous Optimization and Real-Time Adjustments Autonomous campaign optimization doesn’t wait for a human to review results and decide on next steps. The system monitors performance signals continuously, making real-time ad adjustments to bids, budgets, and creative rotation to keep campaigns aligned with defined CPA and return on ad spend (ROAS) targets. Think of it as a campaign manager who processes every data point instantly and never needs a check-in. Instead of discovering a week later that a campaign has been underperforming, the AI identifies the issue and acts within minutes. ### Continuous Learning and Predictive Intelligence Every impression, click, and conversion becomes an input that sharpens the system’s predictive intelligence. Over time, the AI builds a picture of which behaviors and signals correlate with high-value conversions, and uses that picture to inform future targeting decisions. Instead of repeating yesterday’s wins, continuous campaign learning allows the system to predict tomorrow’s opportunities and prepare campaigns to capture them. This is an important distinction. A rules-based system gets better only when a human updates it. A predictive system gets better on its own, continuously, as long as the data keeps flowing. ### Full-Funnel Visibility and Analytics One of the persistent challenges in performance marketing is connecting upper-funnel activity to actual revenue outcomes. AI-driven marketing campaigns address this by pulling together acquisition, activation, and retention data across channels, then analyzing the complete picture to identify which touchpoints actually drive business outcomes. With full-funnel visibility, advertisers can see how open web advertising contributes to conversions — not just as an isolated metric, but as part of a complete customer journey. That clarity makes budget decisions sharper and ROI reporting more defensible. ## Advanced Audience Targeting Without Walled Gardens Google and Meta offer powerful targeting options, but they’re built around data those platforms own. Predictive audience targeting on the open web takes a different approach: using behavioral signals and contextual relevance to identify high-intent prospects across thousands of independent publishers. Rather than relying on demographic categories alone, machine learning for advertisers can analyze how people are behaving in real time — what they’re reading, researching, and engaging with — to gauge receptivity. Compared to walled garden targeting, this approach offers several advantages: - Broader signal set: Behavioral and contextual data across the open web isn’t filtered through a single platform’s reporting lens. - Discovery-moment targeting: Ads reach users while they’re actively researching relevant topics, not just because they match a demographic profile. - Lookalike modeling without platform lock-in: AI identifies traits common to your best converters and expands to similar audiences — no platform required. The result is a targeting approach that scales without the cost premiums that come with saturated walled garden inventory. ## Smart Budget Allocation and Bidding Strategies Manual bid management has always been a game of catch-up. By the time a human analyst identifies a high-performing placement and reallocates budget toward it, the window may have already passed. Cross-channel budget allocation powered by AI eliminates that lag. The system continuously evaluates which publishers and placements are driving the best results, shifting spend in real time to maximize efficiency. Instead of a static media plan that locks in allocations for weeks, advertisers get a dynamic allocation engine that responds to live performance data. This is especially important on the open web, where inventory quality varies significantly across publishers. AI-driven campaigns can surface which independent publishers consistently deliver strong returns, prioritizing those placements without requiring marketers to review site-level performance manually. ## Dynamic Ad Creation and Rapid Experimentation Testing used to be slow. You’d run two or three ad variants for a few weeks, pick a winner, and move on. Generative AI has changed that cycle entirely. Dynamic creative testing allows marketers to generate hundreds of ad variations, then let the optimization engine determine which combination of headlines, images, and copy resonates most with specific audience segments. AI-powered optimization delivers a 20%–30% improvement in campaign performance compared to periodic manual optimization. Much of that lift comes from this kind of continuous experimentation that no human team could sustain manually. The key advantage is that testing and optimization happen simultaneously. While the system identifies winning creative, it’s already shifting spend toward top performers. There’s no waiting period between learning and action. ## Overcoming Data Quality and Integration Challenges AI is only as strong as the data it works with. Before any of the capabilities above can deliver valuable results, advertisers need to address what might be the least glamorous part of the process: data readiness. Fragmented data is the most common obstacle. Advertising data lives in one platform, customer relationship management system (CRM) records live in another, and site analytics sit in a third. Without a unified tracking framework that consolidates these sources, AI models are working with an incomplete picture, and optimizing toward incomplete signals. These principles can help: - Define clear conversion events and ensure they’re tracked consistently across all touchpoints. - Consolidate data sources into a single reporting environment before scaling AI-driven campaigns. - Audit existing data for gaps, particularly around attribution windows and cross-device tracking. - Take data privacy considerations seriously, ensuring your approach to audience data is compliant with applicable regulations. Think of clean, unified data as the foundation. Miss this step, and no amount of AI sophistication will deliver useful results. ## Steps to Implement AI-Powered Strategies on the Open Web Getting started with AI campaign optimization doesn’t have to be overwhelming. A phased approach reduces risk while building the infrastructure needed for long-term performance. Here’s how: - Define your key performance indicators before touching a single platform setting: Whether you’re optimizing for CPA, ROAS, or lead volume, the system needs clear success criteria to work toward. - Audit your data sources to ensure tracking is complete and consistent: Gaps here will compound quickly once the system starts learning. - Choose a platform with native AI capabilities built for open web advertising: Look for a platform that integrates bidding, creative testing, and audience modeling in a single environment. - Launch pilot campaigns with limited budgets to validate performance before scaling: This initial phase feeds the AI with the conversion data it needs to begin learning. - Give the system time to calibrate before drawing conclusions: Early performance may be uneven as the model learns. Optimizing too aggressively in the early weeks can interrupt that process. - Expand reach: Once you’ve established a baseline, use predictive audience tools to expand to lookalike segments and scale what’s working. ## The Future of AI Marketing and Web Performance Performance marketing AI is moving fast, and the direction is clear: the days of optimizing for known audiences are giving way to something more dynamic. Campaigns now discover new audiences continuously, while creative strategies evolve in real time, rather than waiting on periodic reviews. Multi-agent systems are emerging as the next stage of this evolution. Specialized AI models work in parallel, each handling a unique function. One manages creative, another adjusts bids, and a third analyzes sentiment. Together, they share contextual insights and collaborate to achieve consistent optimization across platforms. Ad campaign orchestration at this level shifts the marketer’s role from hands-on campaign manager to strategic director, setting objectives, defining guardrails, and interpreting outcomes while handling AI execution. Over time, this brings a compounding advantage for open web advertisers. The earlier you build AI-powered infrastructure and start generating conversion data, the more sophisticated your targeting and optimization capabilities become. Early movers aren’t just ahead today. They’re widening the gap. ## Key Takeaways AI campaign optimization is no longer optional for performance advertisers looking to scale beyond saturated search and social platforms. By replacing manual bid management and creative decisions with autonomous, real-time optimization, AI-driven campaigns on the open web continuously learn and improve with every conversion event. For advertisers, the strategic case is straightforward: the earlier you build this infrastructure, the wider your performance advantage grows, and the harder it becomes for competitors who lag behind. ## Frequently Asked Questions (FAQs) ### How does AI campaign optimization improve marketing ROI on the open web? AI algorithms analyze real-time performance data across thousands of independent publishers, automatically adjusting bids, reallocating budgets, and rotating creative to prioritize what’s working. By acting on high-intent behavioral signals and eliminating manual guesswork, performance campaigns on the open web consistently improve return on ad spend without requiring constant human oversight. ### Why should performance advertisers focus on expanding beyond Search and Social? Walled garden platforms are increasingly saturated and expensive. The open web offers access to vast, untapped inventory, and today’s AI-powered tools can target that inventory with the same efficiency as traditional networks, often at a significantly lower cost per acquisition. ### Can AI handle ad creation as well as campaign optimization? Yes. Modern generative AI tools can rapidly produce multiple ad variations across headlines, copy, and creative formats. The optimization engine then tests those variations in real time, ensuring the right creative reaches the right audience at the right moment, automatically shifting spend toward top performers. ### What is the biggest challenge when implementing an AI optimization strategy? Data quality and integration are the biggest challenges in AI optimization implementation. AI models require clean, consolidated data to generate useful predictions. Establishing consistent tracking, unifying data sources, and auditing for gaps before launching AI-powered campaigns is essential, and it’s often the step advertisers underinvest in. --- ### 6 Best Practices for Successful CTV Advertising Campaigns URL: https://www.taboola.com/marketing-hub/ctv-advertising-best-practices/ Last Modified: 2026-06-30 11:19:08 Connected TV (CTV) has spent years being sold as the premium awareness play. Big screen. Big reach. Big brand moments. But the conversation is shifting, and it’s shifting fast. Advertisers are no longer content with impressions as the endpoint. They want to know what happened after someone watched their ad. Did they visit the site? Did they buy something? The era of CTV performance advertising is here, and the brands winning right now are the ones who’ve stopped treating connected TV like a billboard and started treating it like a conversion channel. Here are some best practices to build campaigns that actually deliver. ## 1. Lay the Right Foundation Before You Buy a Single Impression Most CTV campaigns fail before a single ad runs. The reason is almost always structural. The teams involved didn’t align early enough, the identity data was inconsistent across planning and buying, or the budget was inherited from last year’s media mix without questioning whether it still made sense. If you’re serious about programmatic CTV best practices, start by getting your house in order. ### Align Teams Around Clear KPIs Before launch, everyone touching the campaign — brand, agency, media partner, analytics — should be working from the same objective. “Awareness” and “performance” require different buying approaches, creatives, and measurement frameworks. If your brand team wants impressions and your performance team wants conversions, you’ll get neither. ### Build Your Budget From Zero Instead of taking last year’s TV spend and shifting a percentage toward CTV, build from current viewer behavior. Where are your audiences actually watching? How do their device habits map to your conversion funnel? Zero-based budgeting forces you to ask these questions instead of inheriting the assumptions baked into last year’s plan. ### Get Your Identity Right CTV measurement is only as reliable as the identity spine underneath it. Ideally, you’re using a consistent household and people-level identity source across planning, buying, and measurement. When these don’t match, attribution gaps appear, and campaign data becomes unreliable. Household IP targeting, in particular, requires clean first-party data to work well. Without it, you’re connecting dots that don’t belong together. ### Define Your Supply Path by Goal Awareness campaigns can justify premium direct deals with major streaming platforms. But performance marketing for TV requires something more flexible. Private Marketplace deals with outcome-based pricing, where you’re buying inventory tied to actual results rather than guaranteed eyeballs, and are usually a better fit for CTV audience targeting with a conversion objective. ## 2. Bridge the Gap Between the TV and the Web The single biggest structural problem in CTV advertising has always been that when someone watches your ad on the big screen, nothing happens after. The session ends. They pick up their phone and see something else. The connection between the TV exposure and what happens next online is completely lost. This is where the TV-to-Web conversion strategy becomes the difference between CTV that earns its budget and CTV that gets cut. The idea is straightforward. When a viewer is exposed to your CTV ad, their household IP address becomes the connective tissue between the big screen and their other devices. From there, connected TV retargeting lets you follow up with relevant ads across the open web — on the news sites, content platforms, and editorial environments that viewers browse throughout the day. You can use the audience exposed to your CTV campaign as a seed for lookalike modeling, identifying similar users who haven’t seen the ad yet but match the behavioral and contextual signals of people who convert. The reach extends beyond your initial CTV exposure, and the attribution connects back to the original campaign. The practical implication is that your CTV buy should never be planned in isolation. Your media plan needs to account for the web touchpoints that follow. What content will viewers see after they watch the TV ad? What action are you driving them toward? Answering these questions before you launch is what separates a CTV campaign from a CTV strategy. ## 3. Use AI to Find the Audiences Most Likely to Convert Basic demographic targeting, like age, gender, and household income, was a reasonable starting point when CTV was mostly about reach. It’s not enough now. To get the best results from CTV audience targeting, you should move beyond demo buckets to behavioral and intent-based signals. The question to ask is: “Who, right now, is most likely to take the action we want?” and not “Who fits our target profile?” This is where AI-driven approaches are making a meaningful difference. Tools like Taboola Realize CTV use open web signals such as browsing behavior, content engagement patterns, and real-time intent indicators to identify predicted converters. Instead of targeting based on who someone is, you’re targeting based on what they’re doing and where they are in the decision process. That’s a different category of precision. Realize builds this into its partnerships with major CTV platforms. Paramount Advertising integrated Realize technology into its Paramount Ads Manager to build a “Performance Multiplier” product, giving advertisers, including small and medium-sized businesses, access to premium streaming inventory with the ability to track purchases, sign-ups, and other conversion events.  LG Ad Solutions took a similar approach, partnering with Realize to turn smart TV inventory into a performance channel connected directly to the open web. What both partnerships illustrate is a model that’s becoming the new standard. CTV platforms that don’t just sell reach, but offer the closed loop that lets advertisers prove what that reach actually drove. For advertisers, the practical implication is this: When evaluating CTV partners, ask how they handle post-exposure behavior. Do they offer connected TV retargeting across devices? Can they model predicted converters from your exposed audience? Is there a direct path from the TV ad to a measurable web action? If the answer to any of these is vague, you’re working with a reach product, not a performance product. ## 4. Optimize Creatives for the Attention Economy There’s a temptation to take a linear TV spot, upload it, and call it a CTV campaign. It’s quick, cheap, and tends to underperform. CTV isn’t television. The viewing context is similar — lean-back, big screen, household setting — but the technical environment is fundamentally different. Performance CTV creative needs direct response elements baked in from the start. QR codes are the clearest example. They let a viewer scan from the couch and land on a purchase page without any friction, turning a passive viewing moment into an active conversion event. They’re the bridge between the big screen and the bottom of the funnel. Beyond QR codes, the more durable creative framework for CTV performance is what you might call “Story + Action.” The TV spot does the heavy lifting on the emotional and contextual fronts. It builds recognition, establishes the offer, and creates intent. The follow-up native ads on the open web then close the loop, hitting that same viewer with a targeted, action-oriented message once they’ve picked up their phone or opened their laptop. Each element plays a different role. When they’re built together as a sequence rather than independently, the combined effect is consistently stronger than either one alone. A few other things worth keeping in mind: ### Don’t Repurpose, Rebuild Even if the underlying message is the same as your linear campaign, the execution should account for CTV-specific formats. Non-skippable inventory requires different pacing. Interactive elements need to be built in, not bolted on. The earlier creative teams are involved in campaign planning, the better. ### Test for Fraud and Brand Safety CTV fraud is real and growing. Invalid traffic, domain spoofing, and fake impressions exist across programmatic channels, and CTV is no exception. A forensic mindset, in which third-party verification tools and anomalous delivery data are treated as signals worth investigating, should be part of any serious campaign setup to help ensure brand safety. ## 5. Measure What Actually Matters CTV ROI measurement is where many otherwise well-run campaigns fall apart. The problem is that many advertisers are still defaulting to reach and frequency metrics when their actual goal is conversions. Impressions tell you your ad ran. Cross-screen attribution tells you what happened because of it. A few principles that hold up across most CTV performance campaigns: ### Don’t Rely on a Single Attribution Method Marketing Mix Modeling gives you a macro view of how channels contribute over time, but it’s slow and can’t capture short-term conversion signals from CTV. Geo holdout tests, where you run the campaign in some markets and withhold it from others, give you cleaner incrementality data. Transaction match-backs connect ad exposures to actual purchase records. Used together, these methods triangulate on CTV’s real contribution rather than relying on a single imperfect proxy. ### Use Data Clean Rooms Where You Can Clean rooms let you match your first-party customer data against a partner’s media exposure data without either side exposing raw records. That means you can tie a CTV impression to an in-store purchase, a subscription sign-up, or a lead form submission, connecting events that would otherwise sit in separate silos. This is increasingly how large advertisers are solving cross-screen attribution at scale. ### Get the Right Data Into the Right Hands Post-campaign reporting that sits in a deck and doesn’t feed back into planning is wasted. Build the workflow so that what you learn about audience performance, creative performance, and channel contribution actually informs the next buy. That feedback loop is what turns individual campaigns into a compounding CTV program. ## 6. Start Broad Before You Narrow One of the most common mistakes in CTV audience targeting is over-segmenting too early. Advertisers build tight audience definitions before the campaign has any data to work with, and the AI engine never gets enough signal to optimize toward anything meaningful. The better approach is to start broader than feels comfortable and let the algorithm learn. When Taboola Realize CTV and similar AI-driven tools run against a wide audience pool, they’re actively scanning for behavioral patterns that predict conversion, not just demographic proxies. That learning process takes impressions to work. Cutting the audience too narrowly upfront starves the model before it can identify which segments are actually driving results. Once the data starts coming in, usually after a few weeks of meaningful volume, you’ll see which audience clusters are converting at higher rates. That’s the right time to shift budget toward those segments and tighten targeting. If you skip the broad phase and go narrow from day one, you often end up with campaigns that look efficient on paper — high click-through rate (CTR), low cost per mille (CPM) — but deliver almost no actual conversions, because they’ve optimized for the wrong thing before they had real data. Start broad. Let the AI learn. Then optimize toward what the data tells you, not what you assumed going in. ## Key Takeaways The infrastructure for TV performance marketing now exists in a way it didn’t before. Measurement gaps are narrowing, and we now have sharper tools for connecting TV exposure to web action. Taboola Realize represents what this new model looks like in practice. CTV inventory that’s connected to open-web retargeting, AI-driven audience prediction, and closed-loop attribution. Campaigns that start on the big screen and follow the viewer through the full conversion journey. The brands investing in that infrastructure now are building a durable advantage. Those still treating CTV as a brand-only channel will find it increasingly hard to justify the spend. ## Frequently Asked Questions (FAQs) ### How do I track conversions from a CTV ad if users can’t click the TV screen? Through cross-device attribution. When someone watches your ad, the platform logs their household IP address. If a device on that same network later visits your site or completes a purchase, the platform connects the two events and credits the TV exposure. It’s not a perfect signal, but it’s the closest thing CTV has to a click. ### What’s the minimum budget for a performance CTV campaign? Traditional TV required millions to enter. Programmatic CTV is a different story. Tools like Taboola Realize let advertisers start in the low thousands, test what’s working, and scale from there. The entry point is low enough that testing is a legitimate strategy, not just a luxury. ### Can I use my existing social media video assets for CTV? Yes, with one hard rule: Vertical video doesn’t work on a widescreen format, so you need landscape assets. Beyond that, shorter formats (15 seconds or 30 seconds) often outperform long-form TV spots on CTV. Viewers are accustomed to the pace of social video, and concise creative tends to hold attention better than traditional 60-second commercials. --- ### Best Performance Platforms for Pixel Audience Management URL: https://www.taboola.com/marketing-hub/best-performance-platforms-for-pixel-audience-management/ Last Modified: 2026-06-30 11:14:45 Successful performance marketing depends on signal quality. If your pixel data is incomplete, inconsistent, or trapped in the wrong platform, you end up optimizing to the wrong events, retargeting the wrong users, and wasting budget on people who have already converted. Pixel audience management solves that problem by turning raw site and app behaviors into audiences you can reliably activate across channels, with governance that keeps tracking accurate as your site changes. Page depth, lead submits, checkout steps, purchases, and offline outcomes are only useful when they’re synthesized into panoptic signals pointing you toward profitable results. Read on to see a side-by-side comparison of the best performance platforms for pixel audience management, followed by practical guidance on how to set up retargeting pixels, when to use a dedicated platform, and what pricing typically looks like. ## 10 Best Performance Platforms for Pixel Audience Management Platform Why It’s Essential Core Use Cases and Features Best for (Performance Advertisers) Pricing Model (Indicative) 1. Realize Pixel‑linked audience activation and performance optimization. Unified pixel data activation, automated audience updates, performance‑based triggers. Performance advertisers relying on real‑time event signals tied to return on investment (ROI). Performance-based model; campaigns billed on CPC basis, or CPM for programmatic. 2. Google Tag Manager (GTM) Centralized deployment of tracking pixels and tags. Pixel governance, event tracking, conversion triggers, cross‑platform tagging. Advertisers managing complex event tracking without direct code changes. Free. 3. Google Analytics and Enhanced Conversions Turn pixel events into enriched audience signals. Event tracking, audience segments, remarketing lists for ad campaigns. Advertisers maximizing conversion signal quality and audience targeting. Free/paid upgrade. 4. Meta Pixel and Conversions API (CAPI) First‑party event capture for improved audience quality. Web/app pixel events and server‑side API for accurate conversion signals. Performance teams focused on social conversion audiences. CPC/CPA/spend via platform. 5. Adobe Experience Platform (AEP) Enterprise audience activation from pixel data. Real‑time edge profiles, event streams, identity resolution. Large advertisers needing unified audiences across channels. Custom/enterprise pricing. 6. Segment (Twilio Segment) Customer data infrastructure to collect and sync pixel events. Event routing, audience unification, destination connectors. Teams needing clean event pipelines for activation in ad systems. Usage/subscription. 7. Tealium iQ and AudienceStream CDP Pixel/event management and real‑time audience building. Event governance, identity stitching, audience feeds to ads. Performance advertisers scaling multi‑channel audiences. Custom/subscription. 8. mParticle Modern customer data platform with strong event and pixel control. Unified event ingestion, audience segmentation, destination rules. Performance teams needing clean, consistent audience signals. Subscription/usage. 9. Snowplow Analytics Event‑level data tracking and audience construction. Raw event collection for deep audience insights. Advertisers requiring advanced audience analytics from pixel streams. Usage‑based cloud pricing. 10. Oracle Unity Customer Data Platform Enterprise audience management from event/pixel data. Real‑time customer profiles, predictive segments, cross‑channel activation. Large enterprises with complex audience needs. Custom/enterprise pricing. ### 1. Realize Why it’s essential: Realize is a performance-first advertising platform that utilizes its proprietary pixel to capture real-time user signals, enabling brands to build, manage, and activate high-intent audience segments across the open web. It serves as a central hub for bridging the gap between website interactions and media buying, allowing advertisers to turn first-party site data into actionable targeting strategies that drive lower CPAs and higher conversion rates. The platform is used to track deep-funnel behaviors (such as page views, leads, and purchases) to train its Performance AI for smarter matchmaking. By managing these pixel-based audiences, marketers can re-engage previous visitors, or exclude converted users to ensure their budget is always focused on the most valuable prospects in a privacy-compliant manner. Showcased features:  - Pixel audiences: Build owned audience segments from pixel data to precisely personalize re-engagement efforts, or exclude existing customers to reduce waste. - Optimize for engagement: Leverage pixel signals like time on site and session depth to qualify audiences and move prospects further down the funnel. - Predictive audiences: Mirror the actions of high-value pixel events to discover new, incremental users with a statistically higher likelihood of converting. - Codeless conversions: Set up event-based tracking for button clicks and page visits directly in the dashboard, without needing developer or technical support. - Tracking test tool: Validate pixel event firing in real time to ensure audience segments are being populated accurately, before launching campaigns. - Server-to-server (S2S) tracking: Complement pixel data by reporting offline or CRM-based conversions to provide a more comprehensive view of the customer journey. Best for: Realize is a standout choice for conversion-focused organizations — particularly in the D2C and financial services sectors — that need to turn anonymous site visitors into loyal customers. Realize excels in scenarios where a brand wants to retarget users based on their specific progress through a multi-step funnel, such as re-engaging “Add to Cart” users while suppressing those who have already completed a purchase. Pricing model: Performance-based model; campaigns billed on CPC basis, or cost per thousand impressions (CPM) for programmatic. Pros:  - Users re-engaged through pixel-based targeting are 70% more likely to convert than first-time site visitors. - Activating predictive audiences derived from pixel data can lead to a 23% uplift in conversion rates and a 13% improvement in CPA. - The Codeless Conversions tool allows non-technical marketers to manage complex tracking setups without touching website code. Cons:  - Advanced features like the Performance Simulator and Predictive Audiences require a baseline of pixel data to function at peak accuracy. - New audience segments created from the pixel may take time to populate before they reach enough scale for effective targeting. - Several high-value pixel management tools, such as the Performance Simulator and Maximize Value bidding, are currently in limited beta. ### 2. Google Tag Manager (GTM) Why it’s essential:  Google Tag Manager is the control center for pixel governance. It lets performance teams deploy, edit, and quality assurance (QA) tracking tags without repeatedly shipping site code. In pixel audience management, GTM is where clean data begins because it standardizes how events fire, ensures parameters are consistent, and reduces the “tracking drift” that happens when landing pages, forms, or checkout flows change. Instead of treating pixels as one-off scripts, GTM lets you build a durable measurement layer. Triggers, variables, consent checks, and QA workflows keep audiences accurate across every destination. Showcased features:  - Tag governance and version control: Centralized management, workspaces, approvals, and rollback. - Event triggers and variables: Define consistent events (e.g., lead_submit, purchase) across pages. - Cross-platform tagging: Deploy multiple ad/analytics tags off the same event definitions. - Consent-aware firing: Control when tags fire based on user consent states. - Preview/Debug mode: Validate event firing before publishing changes. Best for: Advertisers managing complex funnels, frequent landing page iterations, or multi-platform pixel stacks who need agility without engineering bottlenecks. Pricing model: The core Google Tag Manager client-side product is free to use for most users, offering essential tag management functionalities without hidden charges or subscription fees. Pros:  - Fast iteration on tracking without code releases. - Improves event consistency across platforms. - Strong QA and governance for teams. Cons:  - Still requires a thoughtful event taxonomy (otherwise you automate chaos). - Misconfigured triggers can create duplicate or missing events. - Client-side tagging alone can be fragile in privacy-restricted environments. ### 3. Google Analytics and Enhanced Conversions Why it’s essential: Google Analytics (GA4) turns the universe of onsite behavior into structured audiences, then makes those audiences actionable through segmentation and remarketing integrations. Enhanced Conversions (where applicable) improves signal quality by helping match conversion data more reliably, which can strengthen audience building and measurement when browser-based identifiers are degraded. In practice, GA is often the “audience truth layer”: it’s where teams sanity-check funnel drop-off, validate event counts, and build behavioral segments that can be used for activation. This is even more true when it’s paired with clean GTM implementation. Showcased features:  - Event-based measurement: Standardized event model for funnel tracking. - Audience builder: Create segments from event sequences and user properties. - Remarketing list creation: Publish audiences to connected ad platforms (where supported). - Enhanced conversions: Improve match quality for conversion signals to support optimization. - Attribution and path exploration: Diagnose what’s actually driving conversions. Best for: Performance teams that want one consistent measurement system for funnel analysis and audience segmentation, especially in Google-centric stacks. Pricing model: Free, with paid and enterprise upgrades available depending on product tier and needs. Pros:  - Strong for funnel diagnostics and behavioral segmentation. - Useful “single source” to validate pixel health and conversion trends. - Enhancements can improve conversion measurement resilience. Cons:  - Audience activation depends on integrations and configuration quality. - Identity and cross-channel resolution are limited, compared to enterprise platforms. - Misaligned event schemas can result in misleading audiences. ### 4. Meta Pixel and Conversions API (CAPI) Why it’s essential:  Meta’s Pixel and Conversions API (CAPI) are designed to preserve conversion measurement and audience quality for paid social, especially when browser-side tracking becomes inconsistent. The Pixel captures client-side events including page views, view content, “Add to Cart,” and purchases, while CAPI sends matching events server-to-server, improving reliability and reducing signal loss. For pixel audience management, the big win is redundancy and accuracy. Better event coverage means superior custom audiences, better lookalikes, and more stable optimization, particularly for deep-funnel events where every missing conversion degrades learning. Showcased features:  - Dual capture (browser + server): More complete conversion event coverage. - Event deduplication: Avoid double-counting when Pixel and CAPI both send events. - Custom audiences: Retarget based on event depth and recency. - Value-based optimization: Optimize toward purchase value (when implemented). - Diagnostics and event quality tools: Identify payload and match issues. Best for: Teams running meaningful Meta spend who need resilient conversion measurement and high-quality social retargeting/seed audiences. Pricing model: Meta’s tracking tools follow a tiered pricing model where the software itself is free, but implementation and maintenance incur costs based on the chosen setup method. Pros:  - More reliable conversion signals than pixel-only setups. - Strong retargeting and lookalike foundations. - Reduces performance volatility from tracking loss. Cons:  - Requires careful implementation to avoid duplicates and mismatched parameters. - Server-side setup can demand engineering resources or a partner tool. - Still primarily optimized for the Meta ecosystem. ### 5. Adobe Experience Platform (AEP) Why it’s essential: Adobe Experience Platform (AEP) is built for enterprises that want to unify customer identity and activate audiences in real time across channels. Instead of relying on isolated pixels, AEP ingests event streams, resolves identities, and builds “edge” profiles that can be used for personalization and advertising activation with stricter governance. For performance advertisers, AEP becomes the high-scale audience brain. It’s how you connect anonymous and known behaviors from sources on the web, apps, your CRM, and offline into segments that remain consistent across campaign platforms, without losing control of data definitions and consent. Showcased features:  - Real-time customer profiles: Persistent, unified profiles that update continuously. - Event streaming and edge activation: Use real-time signals to build and refresh audiences. - Identity resolution: Stitch identities across devices and systems (where permitted). - Governance and privacy tooling: Policy controls for data usage and consent. - Activation connectors: Push segments to downstream channels. Best for: Large advertisers with complex identity needs, multiple business units, and strict governance requirements. Pricing model: AEP uses a customized, subscription-based pricing model tailored to an organization’s specific requirements, data volume, and desired digital capabilities. Pros:  - Powerful identity and real-time segmentation. - Strong governance for regulated or complex organizations. - Enables consistent cross-channel activation. Cons:  - Heavy implementation and organizational lift. - Cost and complexity may be overkill for mid-market teams. - Requires strong data strategy to realize value. ### 6. Segment (Twilio Segment) Why it’s essential:  Segment acts like the plumbing for pixel events, collecting behavioral data and routing it to every destination that needs it, from analytics tools and ad platforms to customer data platforms (CDPs) and warehouses. For pixel audience management, Segment reduces fragmentation by enforcing a single event taxonomy and preventing each platform from becoming its own silo. Instead of “we have five different definitions of ‘lead,’” Segment helps you define events cleanly and distribute them reliably, so audiences built in downstream systems are based on the same source-of-truth behaviors. Showcased features: - Single event collection layer: Capture events once across web/app. - Destination connectors: Send standardized events to ad and analytics tools. - Event governance: Naming conventions, schema controls, and data quality checks. - Identity tools (where configured): Map users across devices and sessions. - Warehousing/Reverse ETL compatibility: Build audiences from warehouse data when needed. Best for: Performance teams that need clean, consistent event pipelines feeding multiple activation endpoints. Pricing model: Twilio Segment employs a usage-based, tiered subscription model focused on monthly tracked users (MTUs). Plans include a free tier (1,000 MTUs), a Team plan (starting at $120 per month for 10,000 MTUs), and a custom-priced Business plan for higher volumes. Pros:  - Standardizes tracking across many platforms. - Reduces duplicated implementation work. - Improves audience consistency downstream. Cons:  - Not a full ad platform; activation depends on connected destinations. - Requires discipline on schemas and governance to avoid event sprawl. - Costs can scale with volume. ### 7. Tealium iQ + AudienceStream CDP Why it’s essential: Tealium iQ tag management plus the AudienceStream CDP combines pixel deployment with real-time audience construction. It’s built for teams that need both the ability to govern event collection and to transform those events into audiences that can be activated across channels. Events become segments quickly, and those segments can be pushed to ad destinations with identity stitching and governance controls baked in. The value for performance advertisers is immediacy and control. Showcased features:  - Enterprise tag management (iQ): Scalable governance and deployment. - Real-time audience building (AudienceStream): Segments based on behaviors and attributes. - Identity stitching: Unify users across sessions and devices where possible. - Connector ecosystem: Push audiences to ad/marketing destinations. - Data governance and consent tooling: Control collection and activation rules. Best for: Teams scaling multi-channel audience programs who need real-time segmentation and strong governance. Pricing model: Tealium iQ and AudienceStream use a customized, enterprise SaaS pricing model based primarily on annual event volume, the number of data sources/properties, and the number of active integrations. Pros:  - Strong combination of collection and activation. - Real-time segmentation supports agile retargeting. - Good governance and enterprise controls. Cons:  - Implementation can be complex. - Costs typically require enterprise budget justification. - Requires operational maturity to maintain clean schemas. ### 8. mParticle Why it’s essential:  mParticle is a modern CDP designed to keep event data clean, consistent, and activation-ready. Instead of wrestling with platform-by-platform pixel quirks, mParticle helps teams define a reliable behavioral layer and turn it into segments that flow wherever performance media runs. The platform incorporates events from web and app sources, normalizes them, applies rules to route audiences, and signals to downstream platforms. For pixel audience management, mParticle is a strong choice when your biggest bottleneck is event consistency across products, devices, and destinations. Showcased features:  - Unified event ingestion: Capture and normalize web/app behaviors. - Audience segmentation: Build cohorts from behavioral sequences and attributes. - Destination rules and controls: Route signals to ad/analytics tools with governance. - Identity resolution (configured): Improve cross-device coherence where possible. - Data quality tooling: Reduce noisy/duplicative events that pollute audiences. Best for: Performance organizations with multi-platform web and app products that need consistent events powering activation. Pricing model: mParticle uses a value-based pricing (VBP) model, which is a consumption-based system centered around pre-purchased credits. Unlike traditional models that charge flat fees per user profile or event, this unbundled approach allows you to pay based on how data is used and stored. Pros:  - Strong event normalization and governance. - Scales well across web/app ecosystems. - Improves quality of downstream audiences. Cons:  - Requires upfront schema and tracking strategy. - Activation depends on destination availability and setup. - Ongoing maintenance is needed to prevent event drift. ### 9. Snowplow Analytics Why it’s Essential:  Snowplow is less “plug-and-play retargeting” and more “build the event foundation that makes every audience smarter.” Teams that want raw, event-level ownership can collect behavioral data at a granular level, store it in their own environment, and then construct audiences from that first-party event stream. For pixel audience management, Snowplow shines when you need deep customization including custom events, uncommon funnels, or advanced attribution logic that packaged analytics tools can’t handle. Showcased features:  - Raw event collection: Highly flexible event schemas and tracking detail. - Behavioral analytics: Deep funnel analysis and custom user journeys. - First-party data ownership: Control how data is stored and used. - Audience construction: Build cohorts from event-level logic. - Warehouse-native workflows: Pair with modern data stacks for activation. Best for: Advanced performance advertisers and data teams that want maximum control over event data and audience logic. Pricing model: Snowplow Analytics utilizes a tiered pricing model that separates the cost of the software/service from the infrastructure required to run it. Their commercial offering, Snowplow BDP (Behavioral Data Platform), primarily uses event-based pricing. Pros:  - Extremely flexible and detailed event tracking. - First-party control supports privacy and customization goals. - Strong foundation for sophisticated audience logic. Cons:  - Requires data engineering investment. - Not primarily an ad-activation user interface (UI); needs downstream tooling to push audiences. - Time-to-value can be longer than packaged CDPs. ### 10. Oracle Unity Customer Data Platform Why it’s essential: Oracle Unity Customer Data Platform (CDP) is designed for enterprises that need real-time customer profiles, predictive segmentation, and cross-channel activation tied to identity resolution. From a pixel audience standpoint, Unity helps consolidate behavioral and customer data into governed profiles that can fuel consistent activation across large organizations with multiple brands, regions, or business lines. It’s the “system of audience record” for companies where audience definitions must be centralized, auditable, and scalable. Showcased features:  - Real-time customer profiles: Unified views that update with new events. - Predictive segmentation: Build cohorts using propensity-like modeling. - Cross-channel activation: Push segments to marketing and advertising destinations. - Identity resolution: Map customers across sources and devices where permitted. - Enterprise governance: Controls for data usage, permissions, and compliance. Best for: Large enterprises with complex audience needs, strict governance, and multi-channel activation requirements. Pricing model: Oracle Unity CDP uses a consumption-based pricing model centered on the value derived from unified customer data. Pros:  - Strong enterprise-grade identity and segmentation. - Predictive capabilities can improve audience efficiency. - Centralized governance and consistency. Cons:  - Implementation and integration complexity. - Higher cost threshold. - Requires strong internal data ownership to operate effectively. ## More About Pixel Audience Management in Performance Campaigns ### How to Set Up Audience Pixels for Retargeting To set up a retargeting campaign, start by defining the handful of funnel events that actually represent intent for your business. These are typically milestones like viewing key content, submitting a lead form, adding a cart item, and purchasing, plus any unique steps that matter in your journey. Once those events are chosen, it’s imperative to standardize how they’re named and what parameters they carry. Audience quality depends on consistent definitions. Some common definitions are event names, product or lead identifiers, values, currency, and any category fields you’ll need to use for segmentation. From there, implement your tracking through a tag manager whenever possible so triggers, variables, and updates live in one governed place. This reduces the risk of broken tags when pages change. When you build retargeting audiences, segment by both intent and recency so you can prioritize high-intent users while keeping audiences fresh. A practical approach is to target recent “Add to Cart” users while excluding anyone who purchased within a longer window. Excluding recent conversions — and possibly existing customers — prevents wasted spend and keeps reporting honest. Suppression should be the default to keep your ad spend in bounds. Finally, treat QA as ongoing maintenance by using preview/debug tools and platform diagnostics to catch duplicate firing, missing parameters, and tracking gaps that appear after landing page iterations or site releases. ### The Benefits of Using a Dedicated Audience Management Platform Using a dedicated audience management platform typically improves performance because it keeps your conversion signals clean and consistent, which makes optimization algorithms learn faster and behave more predictably. A dedicated platform also speeds up iteration by letting marketers adjust audiences, rules, and triggers without waiting on engineering every time a landing page changes or a new funnel step needs to be tracked. When your audience logic is centralized, a lead or high-intent visitor is defined the same way across every destination, reducing mismatches between platforms and reporting. These tools can also make tracking more resilient in a privacy-restricted environment, by supporting server-side event delivery and identity tooling that helps reduce signal loss. Finally, they add a layer of governance and compliance, with consent-aware controls and auditable change histories that help organizations manage risk while still moving quickly. ### Cost of Pixel Audience Management Platforms The cost of pixel audience management platforms usually falls into four broad buckets. At the low end, you have free foundations like Google Tag Manager and basic analytics tools, which don’t charge licensing fees, but still come with real costs in the form of implementation time, ongoing QA, and the internal effort required to keep your event taxonomy clean as your site evolves. Next are media-tied systems, where the platform cost is effectively bundled into your ad spend. In setups like Meta’s Pixel and Conversions API, the audience tools come with the ecosystem, and what you’re really paying for is performance efficiency: better signals can reduce CPA, but weak implementation can quietly inflate it. Then there are subscription or usage-based customer data platforms such as Segment, mParticle, and Snowplow, where pricing often scales with the volume of events you collect, how many sources you ingest, and how many downstream destinations you send data to. These platforms can start reasonably but become more expensive as data volume and activation complexity grow. Finally, enterprise suites like Adobe Experience Platform, Oracle Unity, and Tealium are typically sold on custom contracts, with pricing driven by factors such as total data volume, the specific modules you need, the number of business units or properties, and the level of support and services required to implement and maintain the system. ## Key Takeaways Pixel audience management is fundamentally a signal quality and activation issue: capture the right events, keep them consistent, then build audiences that map to intent. It’s best to start with governance before adding CDPs, so that clean events compound value across every platform. For performance teams, the best setups combine retargeting, suppression, server-side resilience, and audience expansion. Choose tools based on where your bottleneck is, whether that’s implementation agility, identity resolution, cross-channel activation, or enterprise governance. ## Frequently Asked Questions (FAQs) ### What is pixel audience management in performance advertising? It’s the process of collecting behavioral events via pixels and event streams, turning them into defined audience segments, and activating those segments to improve CPA, return on ad spend (ROAS), and funnel efficiency. ### Why are first-party pixel audiences more important now? Because third-party identifiers are less reliable, platforms increasingly depend on first-party events to build audiences and optimize delivery. The cleaner your first-party signal, the more stable your performance. ### How does server-side tagging affect pixel audience management? Server-side tagging can reduce event loss and improve match quality by sending conversion and behavioral signals directly from your server environment (or a server-side container) to ad platforms, making audiences and optimization more resilient when browser-side tracking is degraded. --- ### Four Types of Bidding Strategies for the Open Web URL: https://www.taboola.com/marketing-hub/open-web-bidding-strategies/ Last Modified: 2026-06-30 08:49:23 Performance advertising on the open web has changed more in the last five years than in the previous 15. The technology driving that change is bidding: specifically, the shift from manual bid management to automated models that can evaluate millions of impressions in real time and optimize toward outcomes a human team could never hit at that speed or scale. This guide covers how automated bidding strategies work, which ones fit which campaign goals, and how to get through the setup process without burning budget while your algorithms learn. ## What Is a Bidding Strategy? A bidding strategy is the set of rules or algorithms that controls how much you pay for a given ad placement, click, or conversion in a real-time auction. Every time a user loads a page, a programmatic auction runs in milliseconds to decide which ad wins that impression. Your bidding strategy is what determines how you show up in that auction — what you’re willing to pay, which impressions you compete for, and how those decisions get made. Some strategies put a human in charge of those decisions. Others hand them to an algorithm. Either way, the underlying goal is the same: win the right impressions at a price that delivers a return, and avoid paying more than an impression is worth. Get that balance right, and your budget works harder. Get it wrong, and you’re either overpaying for placements that don’t convert, or under-bidding on ones that would have. ## 4 Core Types of Bidding Strategies for Performance Marketers Different campaign objectives call for different bidding approaches. Here’s a breakdown of the four strategies most relevant to performance advertisers on the open web. ### 1. Target Cost per Acquisition (tCPA) Target CPA (tCPA) is the conversion-focused default for most performance marketers. You set an average cost you’re willing to pay per acquisition — a lead, a sale, a form fill — and the algorithm does the work of hitting that target across a campaign. The key word is average. TCPA doesn’t mean every conversion will cost exactly your target number. The algorithm bids aggressively when it predicts a high probability of conversion and conservatively when it doesn’t. Some days will run above target, some below, and the goal is to hit your average across the campaign’s lifetime. Taboola’s target CPA guide covers this well: TCPA delivers the strongest results for advertisers with clearly defined cost-per-lead or cost-per-sale goals and the data volume to support consistent optimization. In practice, it requires a minimum conversion threshold — typically 30 to 50 conversions per month — for the model to have enough signal to optimize reliably. TCPA works best for lead generation, direct-response campaigns, subscription services, and any vertical where acquisition cost is a hard business constraint. ### 2. Target Return on Ad Spend (tROAS) / Value-Based Bidding TROAS is tCPA’s more sophisticated sibling. Rather than optimizing for the number of conversions, it optimizes for the value of those conversions. Instead of trying to acquire any type of customer, tROAS focuses on acquiring the most valuable ones. This requires passing revenue or value data back to the platform. When you tell the algorithm that a $200 purchase happened on a given click, it learns to find more users likely to generate $200-plus transactions and bid accordingly. The result is a shift away from volume-based thinking toward value-based bidding that accounts for the actual revenue each customer generates. For e-commerce, this is especially powerful. A campaign optimizing for conversion volume might chase lower-order transactions at a lower CPA. A tROAS campaign, properly configured, pursues high-order-value customers even at a higher individual acquisition cost — because the math works out better at the campaign level. The trade-off is that tROAS requires more data and a more complex conversion-tracking setup. You need to pass reliable revenue values to the platform so the model has something to learn from. TROAS works best for e-commerce brands with variable order values, subscription services with known lifetime value benchmarks, and any advertiser able to pass post-click revenue data back to the platform. ### 3. Maximize Conversions Maximize Conversions is the volume-first strategy. You set a budget, and the algorithm spends it as efficiently as possible to maximize conversions. There’s no CPA guardrail as the goal is pure volume within the budget constraint. This makes it a useful launch strategy. When you’re starting a new campaign and don’t have enough historical data to calibrate a tCPA target reliably, Max Conversions gives the model room to explore. It bids broadly, gathers conversion data across a range of placements and audiences, and surfaces the patterns that tCPA can then exploit. As Nadim Batista-Kuttab of Xevio, one of the largest native advertisers in the world, explained in Taboola’s comparison of max conversions vs. target CPA: Max Conversions is the strategy that unlocks growth and fuels platform learning, while tCPA is the precision tool to use once campaigns have matured or hit a plateau. The practical approach is to run Max Conversions until you have a stable conversion baseline, then transition to tCPA once the algorithm has enough data to optimize toward a specific cost target. Max Conversions works best for campaign launches, budget-flush periods where volume matters more than efficiency, and any situation where you need to generate conversion data before tightening toward a CPA goal. ### 4. Viewable CPM (vCPM) for Awareness VCPM is the odd one out in this group since it’s not a conversion-focused strategy. While tCPA, tROAS, and Max Conversions are lower-funnel tools, vCPM is an awareness-and-consideration play. The difference from standard CPM is that you’re not paying for impressions that nobody saw. VCPM charges only when an ad meets a viewability threshold — typically 50% of the ad is visible for at least 1 second for display, 2 seconds for video. This filters out the buried placements that inflate impression counts without generating any actual exposure. For brands running upper-funnel campaigns alongside lower-funnel performance activity, vCPM can be a cost-effective way to build reach among audiences not yet ready to convert. The goal is to move users through the consideration phase so that later retargeting and performance campaigns find a warmer audience. VCPM works best for brand awareness campaigns, new product launches, reaching audiences early in a longer consideration cycle, and situations where visibility confirmation matters more than click volume. ## How to Choose the Right Bidding Strategy for Your Campaign Goals The choice usually comes down to three questions: - Where are you in the campaign lifecycle? - Do you have enough conversion data? - What’s your business objective? If you’re launching a new campaign and don’t have historical data on your open web audience, Max Conversions is often the right entry point. It generates the data you need to make better decisions later. Jumping straight to tCPA without adequate data leads to erratic performance as the model guesses rather than learns. Once you’ve accumulated enough conversions to establish a reliable cost baseline, tCPA becomes the right tool. You have a cost constraint, the data to calibrate it, and the algorithm can now optimize specifically toward that number. If you’re operating an e-commerce business and can pass revenue data back to your platform, tROAS is worth the additional setup complexity for bid optimization. The efficiency gains from value-based bidding at scale routinely outpace what tCPA can achieve on its own. For the best performance marketing channels question, performance advertisers should default to outcome-based strategies — tCPA, tROAS, or Max Conversions — rather than awareness-based models like vCPM, unless you’re running a distinct upper-funnel campaign with brand reach as the stated objective. Mixing optimization goals within the same campaign is a common source of performance confusion. Here’s a practical framework to help you choose the best bidding strategy for your campaign. If: - the campaign is new with limited conversion history, start with Max Conversions to build data. - your campaign has 30-plus monthly conversions and a defined CPA target, move to tCPA. - it’s e-commerce with variable order values and revenue tracking in place, test tROAS. - your goal is upper-funnel brand awareness with confirmed viewability as a priority, use vCPM as a distinct campaign. ## Manual vs. Automated: The Evolution of Bidding Strategies Search and social built the performance advertising playbook. But rising costs per mille (CPMs), shrinking audience pools, and creative fatigue have made those channels increasingly expensive to scale. Advertisers competing in the same auctions, for the same users, with the same formats are finding that efficiency gains are harder to come by. The open web is where that growth ceiling doesn’t exist yet. Billions of daily impressions across premium publisher environments, reaching audiences that search and social simply can’t access. The inventory is there. The challenge is using it effectively — and that’s where manual bidding falls apart. Manual bidding made sense when digital advertising was simpler. A handful of keywords, a clear audience, and a maximum cost per click (CPC) set at the ad group level. That felt like control. The open web broke that model. Millions of publisher properties, each with distinct audiences, content environments, and conversion probabilities. No human team can evaluate every impression at that scale. By the time someone identifies that certain placements are underperforming and adjusts accordingly, a machine learning model has already made that adjustment thousands of times over. Automated bidding removes that friction. Instead of setting a static maximum bid, you define an outcome — a target cost per acquisition (CPA), a return on ad spend (ROAS) goal, a conversion volume objective — and the algorithm works backward from that goal to determine the optimal bid for every auction it enters. It also filters automatically, deprioritizing low-probability impressions and flagging fraudulent traffic before budget gets wasted on placements that were never going to convert. The result is bidding that’s faster, more responsive, and — once it’s had enough data to learn from — often more efficient. ## The Key Data Signals That Power AI Bidding Models The reason AI bidding outperforms manual bidding lies in its ability to process data at a scale humans can’t replicate. When an AI bidding model evaluates an impression, it ingests and weighs dozens of contextual signals in real time. Some of the most significant ones: - Device type and browser: A user on a desktop Chrome browser completing a purchase-intent search behaves differently than a user casually scrolling on mobile. AI models learn these patterns at the publisher and placement level, not just the audience level. - Geolocation: Conversion rates vary by region, city, and even neighborhood. A bidding model that accounts for location can reduce wasted spend on geographies that consistently underperform. - Time of day and day of week: User intent and purchase behavior shift throughout the day. AI bidding adjusts in real time rather than waiting for a scheduled campaign review. - Publisher context: The content a user is reading when they encounter your ad matters. A reader on a financial news site is in a different mindset than a reader on a movie review site. Contextual signals help real-time bidding models weigh the value of each impression more accurately. - Historical performance data: Every conversion or non-conversion feeds back into the model. Over time, it builds a detailed picture of which impression characteristics predict success for your specific campaign. - First-party behavioral signals: Platforms like Realize use proprietary data built over years of network activity to identify patterns that go well beyond what any individual advertiser could observe from their own campaigns alone. These signals combine in real time to produce a single bid figure. A good impression for a high-intent user in a relevant context on a proven publisher gets a higher bid. A marginal impression gets a lower one. A consistently underperforming placement might get no bid at all. ## Navigating the AI Learning Phase and Data Requirements Every automated bidding system requires a period of learning before it performs reliably. This isn’t a flaw in the technology — it’s how machine learning works. The algorithm needs enough conversion data to identify patterns, and gathering that data costs time and money. The learning phase typically lasts one to three weeks for most campaigns. During this time, machine learning optimization and performance will often look worse than expected. CPAs may run higher, ROAS may be lower, and the temptation to intervene is strong. Resisting that temptation is one of the harder disciplines in automated bidding. A few things to avoid during the learning phase: - Changing the CPA target significantly. When you shift your target, the algorithm has to re-learn what “good” looks like. Frequent target changes reset the learning process and extend the timeline before you see stable performance. - Pausing and restarting the campaign. Every pause interrupts the data accumulation that the algorithm depends on. If the campaign doesn’t have continuity, neither does its learning. - Making large budget changes. A sudden budget increase can force the algorithm to compete for inventory it hasn’t yet learned to evaluate, which often temporarily drives up costs. Automated bidding strategies maintain sufficient daily budget for at least two weeks without major changes and consistently outperform more reactive campaign management. So, commit to the learning period and evaluate performance on a two- to four-week average rather than daily for the best results in smart bidding. ## Best Practices for Optimizing Your Open Web Bidding Strategy Here is how to get the most out of your bidding strategy. ### Set up conversion tracking before the campaign launches The algorithm is only as good as the signal it receives. If your conversion events are misconfigured, delayed, or incomplete, the model is optimizing toward noise. Installing the Taboola Pixel correctly and verifying that events fire reliably is non-negotiable before switching to any automated bidding strategy. ### Use clean, complete first-party data Predictive targeting and audience optimization improve significantly when algorithms have access to high-quality first-party behavioral data. Customer relationship management system (CRM) audiences, pixel audiences, and server-side event data all help the model identify users who resemble your best customers. Feed the algorithm better inputs, and it returns better outputs. ### Calibrate your CPA target carefully Setting a target that’s too aggressive relative to what your campaign can realistically achieve leads to under-delivery. Setting it too loose leads to inefficient spending. The right starting point is usually close to your historical average CPA from other channels, then adjusted based on actual open web performance data as it accumulates. ### Don’t judge performance by day-one numbers Data-driven campaign management means evaluating performance over a meaningful window, not a 24-hour snapshot. Automated bidding strategies typically require two to four weeks of data before performance stabilizes. Building that evaluation period into your reporting cadence prevents premature campaign changes. ### Test creative continuously The bidding strategy is not the only lever. Motion ads, in particular, give AI bidding models richer engagement data to optimize against. Creative fatigue is a real performance drag, and algorithms that work with fresh, varied creative consistently outperform those running on exhausted, static assets. ### Measure true performance with multi-touch attribution Post-click conversions don’t capture the full impact of open web campaigns. Controlled experiments, blended metrics such as the Marketing Efficiency Ratio, and third-party attribution tools provide a more accurate picture of what your campaigns are actually driving. Walled-garden attribution consistently overstates performance — measuring open web campaigns with the same rigor prevents that distortion. ## The Future of AI in Media Buying The shift from manual to automated bidding is the first act. The second act is already underway. AI agents are beginning to handle tasks that still require human input today: budget reallocation across channels, creative generation and testing, audience expansion decisions, and bid strategy selection itself. Platforms like Realize are already moving in this direction, using predictive models trained on years of proprietary first-party data to surface recommendations and automate decisions that previously required an analyst. Predictive targeting represents one of the clearest near-term applications. Rather than building audiences based on demographic similarity to existing customers, predictive models identify users who are actively demonstrating intent-consistent behavior — reaching high-value prospects earlier in their decision cycle, before they’ve entered the saturated search and social auctions where every competitor is vying for the same clicks. The open web’s interoperability challenge is real, but platforms investing in unified data layers and privacy-compliant signal management are building infrastructure that makes advanced AI optimization increasingly viable outside the walled gardens. As that infrastructure matures, the performance gap between open web campaigns and search/social will continue to narrow for advertisers willing to invest in the technology stack. The long-term trajectory looks like automated bidding becomes table stakes, and the competitive advantage shifts to the quality of the data feeding those models and the sophistication of the AI systems interpreting it. ## Key Takeaways The core lesson from the shift to AI bidding strategies is simple: human decision-making doesn’t scale to the volume and complexity of modern open web advertising. Automated and AI-powered bidding strategies — tCPA, tROAS, Max Conversions — give performance advertisers a way to compete at scale across the open web without requiring a team of analysts to manage every placement and adjustment manually. The practical path forward: - Match your bidding strategy to your campaign stage and objective. - Commit to the learning phase without premature intervention. - Feed the algorithm clean, complete conversion data. - Evaluate performance over meaningful time windows, not daily snapshots. - Layer in value-based optimization as your data matures. Stop treating the web as a secondary channel. You already have the tools needed to compete at scale, so use them. ## Frequently Asked Questions (FAQs) ### What is a bidding strategy? A bidding strategy is a specific set of rules or algorithms that determines how much an advertiser is willing to pay for an ad placement, click, or conversion. It governs how your budget is allocated across real-time ad auctions, and whether those decisions are made manually by a human or automatically by a machine learning model. ### How does an AI bidding strategy differ from manual bidding? Manual bidding requires an advertiser to set and periodically adjust a maximum bid at the campaign or ad group level. An AI bidding strategy uses machine learning to analyze signals — device type, geolocation, browser, time of day, publisher context, and historical conversion data — in real time, adjusting each bid for bid optimization based on the predicted likelihood of a conversion. The result is faster, more granular optimization than any manual process can achieve at scale. ### Why should performance advertisers expand to the open web? Search and social audiences are finite and increasingly expensive. The open web offers access to a vastly larger audience across premium publisher environments, at CPMs that are often more efficient than those in saturated walled-garden auctions. Paired with AI bidding technology like Realize, the open web can deliver comparable or better customer acquisition costs — with the added benefit of reaching users in content-engaged, high-intent contexts that search and social can’t replicate. ### What is the difference between target CPA and target ROAS? Target CPA optimizes your campaign to acquire conversions at a specific average cost. The algorithm bids based on conversion probability without differentiating between the value of different conversions. Target ROAS is a value-based strategy that optimizes toward the predicted revenue a given user will generate — prioritizing high-value transactions over raw conversion volume. TROAS requires that revenue data be passed back to the platform and generally performs best in e-commerce and subscription contexts where conversion values vary meaningfully. ### How long does the AI learning phase take? Most AI bidding algorithms require one to three weeks to accumulate enough conversion data to optimize reliably. The exact duration depends on campaign budget and conversion volume — higher-spend campaigns with more frequent conversions learn faster. During this period, performance will often look inconsistent. Avoiding major changes to bids, budgets, or targeting during the learning phase is important for giving the model enough continuity to stabilize. --- ### Campaign Reporting: Winning Open Web Advertising With AI URL: https://www.taboola.com/marketing-hub/campaign-reporting/ Last Modified: 2026-06-30 08:37:00 The open web has emerged as the premier destination for brands seeking authentic scale and diverse audiences. With this expansion comes a new layer of complexity: data fragmentation. In an era where consumer journeys are various and quick-moving, traditional look-back reporting is no longer sufficient. Marketers must transition from passive data collection to active, AI-driven campaign intelligence that turns disparate signals into a unified competitive advantage. ## What Is Campaign Reporting and Why Is it Critical for Open Web Expansion? Campaign reporting is the collection and analysis of data across all marketing channels and initiatives. It’s a critical step in campaigns to evaluate what took place, how it performed, and ways to iterate to improve or repeat what worked well. In walled gardens (closed ecosystems where data is controlled by one entity, e.g., Google or Meta), campaign reporting performance metrics and analytics are more self-contained. In open web advertising, you may be collecting data and inputs from multiple systems and platforms, measuring various ad types, and gaining access to disparate data. With the expansion to the open web, marketing campaign reporting success requires a modernized, unified approach to tracking data. ## The Data Fragmentation Challenge: Life Beyond Walled Gardens With these disjointed platforms come challenges to collecting, interpreting, and acting upon data. Privacy changes, fragmented proprietary platforms, and inconsistent cross-channel approaches limit the ability to connect media activities to business outcomes, according to the IAB State of Data 2026 report. Each service or platform reports and tracks data differently — events, engagement, views, all different and housed in different spots. Without tech intervention, manual reporting can feel impossible to stay on top of, let alone make meaningful insights with. ## Key Performance Metrics to Track on the Open Web Performance marketing with open web advertising requires more sophisticated measurement than traditional metrics provide. Beyond clicks, track AI-driven marketing analytics such as: - Viewability. - Attention. - Conversion. - Cost per acquisition (CPA). - Return on ad spend (ROAS). Be wary of giving each equal weight and importance, as different campaigns or business models will have different emphasis of importance. Look for an AI-driven marketing analytics platform that will weigh metrics based on your campaign and marketing goals. ### Viewability and Attention Metrics Viewability and attention metrics are important; to get real results, ads need to be viewed by humans where they are — prime real estate on independent domains, rather than out of sight as non-viewable impressions. Alongside viewability metrics, attention metrics are measurements of what a user is actually doing with and around the ad. Think dwell time, mouse movement, video completion. These show actual engagement and intent. On the open web, where context and placement can be left to bidding platforms or automatic ad buys, attention metrics will help show that the ad is being served where people can see, in a way that creates interaction and engagement. ### Conversion, CPA, and Predictive ROAS Revenue is the end goal, which is tracked by actual conversions attributed to an ad or campaign. CPA tracks outcomes like signups, purchases, and qualified leads — the actions that move something from marketing into sales. ROAS tracks revenue earned against marketing dollars spent. AI analytics platforms can support better outcomes and predictions based on early engagement signals, and serve ads that include visuals, wording, placement, and timing that are more likely to convert, meaning ad spend is used more strategically. ## How AI Is Transforming Campaign Reporting AI is an operational necessity for campaign reporting, if you want to process open web data in a meaningful and timely way. With the scale and speed at which information is collected in open web advertising, AI reporting is key for optimization. AI-powered analytics platforms automate aggregation, data cleaning, and data schemas for cross-channel attribution. AI campaign reporting platforms can ingest fragmented data to surface insights, leaving marketers time to make business-critical marketing decisions. ## Moving From Static Dashboards to AI-Driven Insights Traditional, look-back reporting models operate on data they’ve collected during a prior window of time. AI-driven marketing analytics proactively anticipate actions and behaviors ahead of spend or activity. They then provide a briefing in natural language that summarizes what changed, why, and what’s next, along with approachable and actionable recommendations. ## Real-Time Monitoring and Anomaly Detection With status dashboards and traditional campaign reporting, it could be days between a report being pulled and an action being taken, leaving campaigns running on stale data. With real-time campaign monitoring and open web updates, issues can be flagged instantly so your team can respond right away, preserving budget. ## Cross-Channel Attribution: Connecting the Dots The Rule of Seven in marketing says that the average customer needs to come in contact with a brand seven times before converting. While that number can be lower for low-cost B2C items, it can easily push 20 contact points in B2B. On the open web, people are coming into contact with your messaging from everywhere, sometimes on multiple devices in quick succession. The challenge is how to attribute an ad to a conversion. Enter cross-channel attribution: Rather than the last touchpoint getting all the credit, AI and machine learning models connect the dots more accurately, weighing all touchpoints and how a customer interacted with them, for more influenced and informed reporting. ## Navigating Signal Loss With Privacy-First Measurement As privacy constraints grow and pixel or third-party cookie methods depreciate, more sophisticated marketing reporting is critical. Through the use of contextual signals and predictive advertising analytics, models can now recognize patterns and apply first-party data to predict behavior, resulting in greater success and more powerful reporting. ## Actioning Your Data: Automated Budget and Creative Optimization With automated budget and creative optimization, you can boost how you use your ad spend. An AI workstream can automatically reallocate ad spend to higher-performing placements, without the need for manual intervention. This approach allows for quicker and more data-focused decision-making and allocation of ad spend. ## Building Your AI-Powered Reporting Tech Stack When selecting the right tools for open web expansion, keep the following in mind: - As platforms and touchpoints increase, consider AI agents that can execute across platforms and self-correct as they go. - When data is spread across multiple systems, enable integrations so everything works together cohesively. - If gathering data to make decisions is bogging you down, try forward-looking dashboards and reports that offer real insights into what the information means, as well as predictions for outcomes. - When ad spend seems inflated for your return, utilize AI agents to monitor anomalies or potential bot or fraudulent traffic, then turn off campaigns that could be bleeding spend. - Address creative refreshes before they become a financial boondoggle. AI creation learns and adjusts what’s being served before it becomes stale. ## Best Practices for Future-Proofing Your Campaign Reports When bringing on automated marketing reporting, start with a single channel or campaign type, unifying your data sources. Define clear business goals so the model understands expectations and what success looks like. Then, AI models can work while you focus on strategy. Fold in other channels or campaigns once you have confidence the first is in a good spot. ## Key Takeaways To succeed on the changing open web, marketers must transition from labor-intensive manual reporting to AI-driven campaign intelligence. Prioritize attention metrics and ROAS for a predictive and contextual framework. Through real-time automation, you’ll be able to proactively optimize budgets and creative to boost ad spend on high-performing placements. ## Frequently Asked Questions (FAQs) ### What is the difference between campaign reporting for walled gardens vs. the open web? Walled gardens, such as search and social platforms, offer self-contained metrics that those platforms control within their ecosystem. The open web requires the capability to bring together information from a multitude of platforms and reporting systems to unify fragmented data. ### How does AI improve marketing campaign reporting? AI improves marketing campaign reporting by automating tedious data collection, pulling together performance briefs in natural language, and detecting performance anomalies in real time. Instead of simply compiling what has occurred, AI recommends the best action to optimize your budget. ### Which metrics matter most for open web performance advertising? Beyond clicks, focus on viewability, attention metrics, conversion, cost per acquisition (CPA), and return on ad spend (ROAS). Based on early engagement signals, AI can predict these KPIs. ### How do privacy changes impact campaign reporting on the open web? With privacy changes such as cookie depreciation and signal loss, campaign reporting has had to shift. Traditional methods and metrics that focus on identity are replaced by predictive modeling, behavioral signals, and first-party data to better deliver ads people want, where they are. --- ### Ad Spend Optimization: AI-Powered Strategies for Better ROI URL: https://www.taboola.com/marketing-hub/ad-spend-optimization/ Last Modified: 2026-06-30 08:29:32 Digital advertising used to be predictable: scale your budget, monitor results, repeat. Not anymore. As competition intensifies across search and social, costs are rising, and performance is getting harder to sustain. What once worked reliably is now delivering diminishing returns, forcing advertisers to rethink where and how they invest. The open web offers a compelling alternative. It unlocks access to new audiences, lower cost per mille (CPM), and nearly limitless inventory. But, it also introduces a level of complexity that manual optimization can’t easily handle. That’s where artificial intelligence (AI)-driven marketing changes the equation. In the agentic era, intelligent systems don’t just assist, they actively manage campaigns and improve performance in real time. Instead of reacting after performance shifts, they anticipate changes before they affect results. In this guide, I’ll explore how AI ad optimization is reshaping open web advertising and what it takes to scale efficiently in this new environment. ## Ad Spend Optimization in the Agentic Era Ad spend optimization in the agentic era means allocating budget dynamically to maximize business outcomes, not merely reduce media costs. At its core, ad spend optimization means allocating budget in a way that drives measurable business results, such as revenue, conversions, or profit. While traditional approaches focused heavily on cost per click or CPM, modern strategies prioritize efficiency across the full funnel. The agentic era introduces a major shift in how this optimization happens. Instead of relying on manual adjustments, agentic AI systems analyze massive datasets, including user behavior, contextual signals, and historical performance. They then use the information to make autonomous decisions. These systems can: - Continuously evaluate performance across channels. - Reallocate budgets dynamically based on predicted outcomes. - Execute optimization decisions in real time without human delay. This evolution makes optimization predictive, rather than reactive. Instead of fixing inefficiencies after they occur, AI-driven systems anticipate performance trends and act before budgets are wasted. ## Why Performance Advertisers Must Expand Beyond Search and Social Search and social platforms still play a critical role in performance marketing, but relying on them exclusively is becoming increasingly difficult. This shift is driving more marketers to rethink walled gardens vs. open web strategies as they look for new ways to scale efficiently. As competition grows, advertisers face: - Higher costs per acquisition (CPA) driven by auction pressure. - Limited scalability due to finite inventory. - Overlapping audiences that reduce incremental reach. - Less control over data visibility and measurement. These constraints make it harder to maintain efficiency while scaling spend. The open web gives advertisers more room to scale, test, and reach audiences beyond closed platform ecosystems. Its advantages include: - Lower CPMs that improve cost efficiency at scale. - Access to incremental audiences beyond platform ecosystems. - A wider range of publishers, from premium media to niche communities. - Greater flexibility in ad formats and creative execution. - Opportunities for more contextually relevant placements. - Increased control over how inventory is sourced and optimized. That said, expanding into the open web isn’t as simple as shifting budget. It requires a different operational approach — one that can manage complexity, unify data, and optimize performance in real time. ## The Open Web Challenge: Why Manual Optimization No Longer Works The open web is vast, and that’s both its strength and its biggest challenge. Programmatic ad spend requires marketers to navigate multiple demand-side platforms, exchanges, and supply paths. Each impression is influenced by countless variables, including user behavior, timing, device, and context. Manual optimization struggles in this environment for several reasons: - Data latency means decisions are based on outdated information. - Human analysis cannot process the scale of available signals. - Budget adjustments often happen after inefficiencies have already occurred. By the time a marketer reviews performance reports, underperforming placements may have consumed a significant portion of the budget. This reactive approach limits growth. To scale effectively on the open web, advertisers need systems that can operate at the same speed and scale as the ecosystem itself. ## AI-Driven Technology: The Key to Open Web Advertising AI-driven marketing changes how advertisers interact with the open web. Instead of relying on periodic reporting and manual adjustments, AI systems continuously ingest and analyze massive volumes of data. That includes everything from user engagement signals to contextual placement data and historical performance trends. These systems don’t just identify what’s working, they also anticipate what is likely to work next. That shift is what enables predictive optimization. Rather than reacting after performance declines, AI can forecast when a placement is likely to fatigue, when a new audience segment is emerging, or when pricing inefficiencies are about to appear. The result is a system that can automatically shift budget toward high-performing inventory and away from underperforming placements as conditions change. This is what makes scaling on the open web not just possible, but efficient. ## Eliminating Wasted Spend with Supply Path Optimization (SPO) One of the most important concepts in programmatic advertising is supply path optimization (SPO). In the open web ecosystem, the same impression is often available through multiple intermediaries. Each additional hop introduces cost, reduces transparency, and increases the risk of inefficiency. Without SPO, advertisers may be competing against themselves or paying unnecessary fees without realizing it. AI-driven SPO helps streamline this process by identifying the most efficient routes to inventory. This leads to: - Reduced exposure to duplicate auctions and redundant bidding. - Lower effective CPMs by eliminating unnecessary intermediaries. - Greater transparency into where impressions are actually sourced. - Stronger relationships with high-quality, direct publishers. - Improved consistency in performance due to cleaner supply paths. By prioritizing efficiency at the supply level, advertisers can ensure their budgets are working harder, not just spending more. ## Protecting Your Budget: Ad Fraud on the Open Web Ad fraud is a persistent challenge in open web advertising. Unlike closed platforms, the open web includes a wide range of publishers and exchanges, not all of which meet the same quality standards. Fraudulent activity can take many forms, from bot-driven impressions to domain spoofing, and it often goes undetected in manual workflows. AI-driven systems address this challenge by continuously analyzing traffic patterns and identifying anomalies that signal invalid activity. This allows advertisers to detect and block suspicious impressions before budget is spent. It also lets them continuously refine fraud detection models as new threats emerge. The key advantage for ad fraud prevention is speed. Instead of identifying fraud after the fact, detection now happens in real time, helping safeguard both spend and data quality. ## How to Set Up an AI-Powered Optimization Engine Transitioning to AI-driven ad spend optimization requires a strong foundation. Without the right infrastructure, even the most advanced systems can’t perform effectively. ### Audit Infrastructure and Implement Server-Side Tracking Traditional browser-based tracking is becoming less reliable due to privacy changes and cookie deprecation. Server-side tracking provides a more accurate, privacy-first alternative. It allows advertisers to: - Capture first-party data directly from their own systems. - Maintain consistent tracking across devices and browsers. - Feed reliable signals into AI models. Without accurate data, AI optimization becomes guesswork. Server-side tracking ensures that decisions are based on complete and trustworthy information. ### Establish Cross-Channel Attribution Models Understanding performance across channels has become complex. Each platform reports its own version of success, often taking credit for conversions that were influenced by multiple touchpoints. This creates a fragmented view of the customer journey and makes it difficult to determine where budget is actually driving value. Cross-channel attribution solves this by consolidating data across environments, bringing together signals from both open web advertising and traditional platforms. When this unified dataset is fed into AI systems, optimization decisions become far more accurate. Instead of overinvesting in channels that appear to perform well in isolation, advertisers can allocate budgets based on true contribution to conversions. This is especially important when scaling beyond walled gardens, where the incremental impact of new channels may otherwise be undervalued. ### Define Optimization Rules and Thresholds Even in the agentic era, human oversight remains essential. Marketers must define the parameters within which AI operates, including: - Target efficiency metrics such as CPA, return on ad spend (ROAS), or lifetime value (LTV) thresholds. - Budget pacing rules to control spend velocity and allocation. - Brand safety guidelines to avoid unsuitable placements. - Business-specific constraints, such as geographic or audience priorities. These guardrails ensure that AI systems align with business objectives while maintaining control over risk. The combination of human strategy and machine execution creates a powerful optimization engine. ## Measuring True Lift with Incrementality Testing Not all conversions are created equal. Some would have occurred regardless of advertising, while others are directly influenced by campaigns. Incrementality measurement helps distinguish between the two. This distinction highlights a core challenge in ad spend optimization: measuring causation, not just correlation. A reported conversion doesn’t always mean an ad drove the outcome. In many cases, users were already likely to convert, and advertising simply captured existing demand. Without incrementality measurement, this can lead to overestimating performance and misallocating budget. AI simplifies what was once a complex and resource-intensive process. It can dynamically create and manage control groups, compare exposed versus unexposed audiences, and continuously refine models based on observed outcomes. The result is a clearer view of true business impact, ensuring that spend is driving incremental growth, not just attributed conversions. ## Human-AI Collaboration: Elevating the Marketer’s Role AI isn’t replacing marketers. It’s redefining their role. By automating execution, AI frees marketers to focus on areas that require human judgment and creativity. This includes: - Defining overall campaign strategy and performance goals. - Shaping brand voice and messaging across channels. - Developing creative that aligns with audience intent. - Interpreting performance trends in a broader business context. - Identifying new growth opportunities beyond existing campaigns. This shift toward agentic workflows creates a more effective division of labor. AI manages real-time optimization, processing massive datasets and executing decisions at speed. Marketers, in turn, guide direction, refine inputs, and ensure that campaigns align with broader business objectives. Together, this collaboration leads to stronger performance, not just through better execution, but through better strategy. ## Key Takeaways Ad spend optimization now depends on faster, more predictive decision-making. As advertisers expand beyond search and social, the open web offers scale, but only if they can manage its complexity efficiently. AI helps with automated budget allocation, improving supply-path efficiency, detecting fraud, and acting on performance signals quicker than manual workflows can. To work well, these systems still need strong inputs, including server-side tracking, cross-channel attribution, and clear human-defined guardrails. ## Frequently Asked Questions (FAQs) ### What is ad spend optimization? Ad spend optimization is the strategic allocation of advertising budgets to drive the strongest possible business outcomes, focusing on revenue, conversions, and long-term profitability rather than just lower costs. ### How does AI improve ad spend ROI on the open web? AI improves ROI by analyzing large volumes of data in real time, predicting performance trends, detecting fraud, and automatically shifting budgets toward the most effective placements before performance declines. ### What is supply path optimization (SPO)? Supply path optimization (SPO) is the process of streamlining the programmatic supply chain to reduce inefficiencies. It eliminates duplicate auctions and unnecessary intermediaries, helping advertisers access inventory more directly and cost-effectively. ### Why is server-side tracking important for automated ad spend? Server-side tracking provides the reliable, privacy-compliant data that is essential for accurate attribution and optimization. Without it, AI systems lack the quality signals needed to make effective decisions, reducing the effectiveness of automated campaigns. --- ### Automated Bidding: Using AI-Powered Bidding on the Open Web URL: https://www.taboola.com/marketing-hub/automated-bidding/ Last Modified: 2026-07-02 12:22:17 Search and social have long been the go-to channels for performance advertising, but that comfort zone is expensive. CPCs are climbing, audiences are saturated, and diminishing returns are hitting harder than ever. The open web is where the scale is, and for advertisers who know how to buy it efficiently, the opportunity is enormous. That’s where AI-powered automated bidding changes the game. Instead of manually adjusting bids and hoping for the best, machine learning now processes millions of real-time signals across the open internet and makes smart buying decisions faster than any human team can. The result is fewer wasted impressions, lower CPA, and profitable conversions at scale. This guide breaks down exactly how it works, what makes it different from legacy bidding approaches, and how to use it to unlock performance across the open web. ## What Is Automated Bidding on the Open Web? Automated bidding means that, instead of manually setting bids for every placement and audience segment, machine learning algorithms do it for you in real time, at a scale no human team can match. You define the goal (whether it’s a target CPA, a target ROAS, or a conversion volume) and the system finds the optimal bid for every auction. It’s constantly reading signals like user behavior, content context, device type, and historical conversion data, to determine what each impression is worth. It then bids accordingly. On the open web, this matters more than it does on search or social. You’re operating across thousands of publishers and placements simultaneously, and manual bidding at that scale is too inefficient. What automated bidding really does is change your role as an advertiser: less time tweaking bids and second-guessing placements, more time focused on strategy — setting the right goals, defining the right parameters, and letting the AI handle execution. ## Manual Bidding vs. AI-Powered Automated Bidding Manual bidding made sense when digital advertising was simpler. Fewer channels, fewer placements, more predictable auctions. You could reasonably monitor performance, adjust bids by hand, and stay competitive. That world doesn’t exist anymore. Search and social have become crowded, expensive environments where CPCs keep climbing and audience saturation is a real ceiling on growth. As advertisers expand onto the open web — with its thousands of publishers, formats, and audience segments running simultaneously — manual bidding doesn’t just become harder, it becomes a liability. A human team can monitor and adjust bids periodically — maybe a few times a day, if you’re well resourced. An AI-powered bidding system is doing it every millisecond, processing millions of data points across historical performance, real-time market conditions, user behavior, and contextual signals all at once, and placing the optimal bid before a human could even open the dashboard. ## Core Mechanisms: How Automated Bidding Algorithms Work Early bidding systems worked on simple rules. If CPA exceeds X, lower the bid. If CTR drops below Y, pause the placement. Useful, but rigid, and only as smart as the rules you wrote. Modern automated bidding is a different animal entirely. Today’s systems continuously learn, adapt, and optimize based on a growing pool of data. The more impressions, clicks, and conversions the algorithm sees, the sharper its decision-making becomes. Two capabilities sit at the heart of how it works: ### Analyzing Real-Time Contextual Signals Every auction on the open web is unique. The same user on a different device, in a different location, at a different time of day can have a completely different conversion probability. Manual bidding treats these as rough averages, while automated bidding treats each one as a distinct opportunity. In the milliseconds before placing a bid, the algorithm evaluates: - Device type: Mobile vs. desktop behavior patterns differ significantly, and so does conversion intent. - Geographic location: Market conditions, purchasing power, and audience intent vary by region. - Operating system: A meaningful proxy for user demographics and engagement patterns. - Time of day: Conversion rates shift based on audience behavior and content consumption habits. - Content context: The editorial environment surrounding the ad influences how receptive a user is likely to be. No human team can weigh all of those variables simultaneously across thousands of auctions per second. The algorithm does it as standard. ### Predictive Modeling Real-time signals tell the system who’s in front of the ad right now. Predictive modeling tells it what that person is likely to do next. The algorithm draws on historical conversion data — which audience segments converted, under what conditions, at what frequency — and builds a probability model for every new impression. It’s answering one question at speed: How likely is this specific user, in this specific context, to take the desired action? If there’s a high probability, the answer is to bid aggressively. If the probability is low, it will bid conservatively or pass entirely. Each conversion (or non-conversion) feeds back into the system, making predictions increasingly accurate over time. It’s the difference between guessing what an impression is worth, and knowing it. ## Top Automated Bidding Strategies for Performance Marketers Not all campaign goals are the same, and your bidding strategy shouldn’t be, either. The right model depends on your growth stage, what you’re optimizing for, and how much conversion data you have to work with. Here are the three core smart bidding strategies performance advertisers use on the open web: ### 1. Target CPA (Cost-Per-Acquisition) Target CPA is the go-to for advertisers who need to stay profitable at scale. You set the cost you’re willing to pay for each conversion, and the algorithm works to hit that number consistently, bidding aggressively when a high-probability opportunity appears, and pulling back when it doesn’t. If you know your margin and what a customer is worth, target CPA gives the AI a clear guardrail to optimize within. Just give it enough conversion data to model accurately before drawing conclusions. Best for: Lead generation, direct response, subscription models, and any campaign where cost control is the primary constraint. ### 2. Target ROAS (Return on Ad Spend) Target ROAS shifts the focus from acquisition cost to revenue value. The algorithm prioritizes conversion quality, bidding more for users predicted to spend more, less for those likely to convert at lower values. A customer who spends $200 is worth more than one who spends $20, even at the same acquisition cost. Target ROAS accounts for that and allocates budget accordingly. It requires passing revenue data back to the platform so the algorithm can learn what high-value actions look like for your business. Best for: E-commerce, retail, and any advertiser optimizing for revenue rather than volume. ### 3. Maximize Conversions Maximize conversions drives the highest possible volume of actions within your set budget. The algorithm has more flexibility to chase volume, prioritizing scale over strict cost efficiency. This works well for market share goals or rapid growth, and it’s useful early in a campaign when you need to generate enough conversion data to eventually shift to target CPA. Monitor CPA closely and set budget guardrails before scaling. Best for: New campaigns building conversion volume, brand expansion plays, and advertisers prioritizing scale over margin in the short term. ## The Major Benefits of Adopting AI Bidding Technologies Switching to automated bidding isn’t just about keeping up with technology. The performance case is concrete, and the advantages compound quickly at scale. ### Time Savings That Move the Needle Manual bid management is relentless. Pulling reports, adjusting bids, monitoring placements — it adds up fast and pulls focus away from higher-value work. AI-powered bidding hands that operational load back to the algorithm, freeing your team to focus on creative strategy, audience development, and campaign architecture. ### Better ROI and Scale, Without the Headcount Human bidding is limited by what we can process and when we can act. Automated bidding doesn’t have those constraints. It evaluates every impression against real conversion probability, adjusts in real time, and allocates budget toward opportunities most likely to deliver a return. It does this across thousands of publishers and placements simultaneously, without a proportional increase in resource or overhead. ### Removing Human Bias From the Equation Even experienced media buyers carry assumptions. They may follow gut feelings about placements, have preferences built on past campaigns, and may tend to over-index on familiar channels. AI-powered bidding removes that bias. Decisions are made purely on data: what’s actually converting, in what context, for which audiences. That objectivity opens up a broader mix of placements and formats than most human teams would explore on instinct alone. ## Essential Best Practices for Implementing Automated Bidding Automated bidding is only as powerful as the foundation it runs on. Getting implementation right comes down to two things: data quality and patience. ### Flawless Conversion Tracking This is non-negotiable. If your tracking is misconfigured, you’re flying blind and actively misleading the algorithm. It will optimize toward whatever signal you feed it, whether that reflects real business outcomes or not. Garbage in, garbage out — at machine speed. Before launching, make sure: - Every conversion action is correctly tagged: Purchases, sign-ups, form fills, whatever constitutes a meaningful action. - Deduplication is in place: So the same conversion isn’t counted multiple times. - Your pixel is verified: Test it before you scale, not after. - Conversion values are passed accurately: Especially critical for target ROAS campaigns. Think of conversion tracking not as a setup task, but as ongoing infrastructure. ### Navigating the Learning Phase Every automated bidding system goes through a learning phase where performance can look inconsistent — CPAs might swing, volume might feel unpredictable. This is normal, and it requires patience. The instinct is to intervene. Resist it! Every significant change resets the learning phase from scratch. Give the system enough conversion volume — most platforms need 30-50 conversions per month, minimum — and avoid major structural changes until it stabilizes. The advertisers who see the best results are usually the ones who trust the process long enough for it to work. ## Common Pitfalls and How to Avoid Them Most mistakes that undermine automated bidding come down to the same root causes: impatience, unrealistic targets, and a misunderstanding of how the algorithm operates. ### Setting targets too aggressive too soon Launching with a target CPA or ROAS far below realistic market rates starves the algorithm of viable auctions. Start with targets based on actual historical data, then tighten gradually once performance stabilizes. ### Making sudden budget changes Cut spend by 50% overnight or double it in a day, and the algorithm recalibrates from scratch. Scale up or pull back incrementally, no more than 15-20% at a time. ### Over-optimizing during the learning phase Every structural change, whether it’s adjusting bid targets, swapping audiences, or pausing placements, forces the algorithm to relearn. Set a weekly review cadence and resist intervening before the data is statistically meaningful. Insufficient conversion data and low-volume campaigns can’t build an accurate predictive model. Consolidate campaigns rather than fragmenting the budget across multiple small ones. ### Letting data integrity slip The algorithm optimizes toward whatever signal you feed it, flawed or not. Audit your conversion tracking regularly. ## The Future of AI-Driven Performance Advertising AI-powered bidding is already transforming open web advertising, and the gap between early adopters and everyone else is only going to widen. A few trends worth watching: ### Deeper contextual and audience intelligence Tomorrow’s algorithms will evaluate the semantic content of a page, the emotional tone of surrounding editorial, and the nuanced intent behind browsing behavior. Combined with smarter audience discovery, this means targeting precision well beyond demographic or behavioral categories, surfacing high-value segments that human media buyers would never think to find. ### Cross-channel predictive modeling AI systems are increasingly able to model conversion probability across the full customer journey, instead of only the last click. Expect real-time bidding algorithms to factor in upper-funnel signals and cross-device behavior to make smarter decisions at every funnel stage. ### Bidding and creative converging Future systems won’t just decide how much to bid — they’ll factor in which creative is most likely to convert for a specific user in a specific context, and adjust both simultaneously. ## Key Takeaways Automated bidding isn’t just a nice-to-have anymore. For performance advertisers serious about scaling on the open web, it’s the engine that makes it possible. The advertiser’s role doesn’t disappear — it evolves. Less time managing bids, more time directing strategy. Remember the following for the best results: - Match your strategy to your goal: Target CPA for cost control, target ROAS for revenue, maximize conversions for scale. - Invest in clean data: The algorithm learns fast when the inputs are right, and stalls when they’re not. - Respect the learning phase: Campaigns given room to stabilize consistently outperform those that are over-managed. - Think long term: Every conversion sharpens the model, and that compounding effect is the strongest argument for starting sooner rather than later. The open web is a massive opportunity. AI-powered bidding is what makes it scalable. The foundations you build today are the performance advantage you’ll have tomorrow. ## Frequently Asked Questions (FAQs) ### What is the difference between automated bidding and manual bidding? Manual bidding means you set and adjust your bids yourself. It’s slow, labor-intensive, and hard to scale. Automated bidding lets AI evaluate real-time signals and place the optimal bid in milliseconds. Manual bidding is reactive, while automated bidding is predictive. ### How long does it take for AI bidding algorithms to learn? Most systems need 1-2 weeks to learn and a minimum of 30-50 conversions within a 30-day period to move through the learning phase and start optimizing accurately. During that window, performance can look inconsistent. That’s normal. Making structural changes resets the process, so set it up correctly, give it time, and let it learn. ### Can automated bidding work for open web campaigns as well as it does on search? For many advertisers, automated bidding works better. Search operates in a contained environment with predictable signals. The open web is far more complex, with thousands of publishers, formats, and audience segments running simultaneously. That complexity is exactly where automated bidding thrives, and where the gap between human and algorithm widens most. ### What data is needed for automated bidding to be effective? You need three different data points: - Accurate conversion tracking. - Historical performance data. - A clear goal. Conversion tracking tells the algorithm what success looks like. Historical data gives it a starting point. A defined target — target CPA or target ROAS — tells it what it’s optimizing toward. Get those three right and the algorithm has everything it needs. Cut corners on any one and you’ll feel it in your results. --- ### 4 Troubleshooting Steps if Your Performance Campaign Is Failing URL: https://www.taboola.com/marketing-hub/troubleshooting-steps/ Last Modified: 2026-06-30 07:55:48 After you’ve put hours of time, research, planning, and execution into a performance marketing campaign, it’s frustrating to watch it fail. Failure can look like draining your budget without generating sales or, worse, failing to serve any impressions. When a campaign isn’t working as planned, you might be facing several different scenarios: stalled ads, your budget going toward unsatisfactory placements, or a skyrocketing cost per acquisition (CPA). When you’re spending but not converting, the problem is usually a structural disconnect in tracking, bidding, or in the funnel. The good news is that there are tools available to navigate and fix these problems so you can get your performance marketing campaigns back on track. This guide gets to the root causes of failing ad campaigns, and can help you as you start to troubleshoot your own campaign. Read on for technical troubleshooting tips for plugging any leaks in your campaign. ## Steps for Troubleshooting Your Performance Campaign ### 1. Diagnosing a Campaign That Isn’t Serving If your campaign isn’t delivering the results you expected, check these common issues while you triage the underperformance. (I’ll use Realize troubleshooting as an example for many of these familiar problems.) - Account-level Hurdles: Check for “red light” issues that might be causing a hard stop in delivery. In the Realize platform, a financial or account hold could cause a frozen account status. Check for expired credit cards or rejected recharges. - Over-targeting: See if you’re layering too many exclusions or small audiences, such as targeting a tiny zip code, or interests that are too specific. This may also show up as the potential reach indicator in your ad platform, leading to a pool too small for the algorithm to bid effectively. - Creative Policy: Take a look at the “edit ad” section of your platform for any rejection flags, such as a misleading headline or low image quality. Realize has strict content policies; users sometimes run into flags around quality or prohibited content that are preventing better campaign performance. - Low Bids: If the cost per click (CPC) number you set is significantly below the network average, you’ll continuously lose at auction. See if your bid is competitive, particularly for various platforms (for example, desktop versus mobile). ### 2. What to Do When Your Campaign Is Spending But Not Converting Low spending can be an easy fix — as long as you have the budget to spare, of course. But, if you’re seeing specific issues after you’ve made sure the spending is appropriate, assess these areas: Broken Tracking and Post-Click Friction: Check your tracking and click actions before assuming the creative is bad. In the Realize dashboard, this feature is in the Tracking Test Tool. This can simulate the user’s click and make sure the pixel is actually firing on the “thank you” page. Ensure Full Alignment: These three features are the holy trinity of campaign alignment: ad creative, landing page, and conversion goal. If these aren’t perfectly aligned, you’ll get clicks but no sales. For example, if you’re running a purchase campaign, but the ad looks like a news article, users will quickly bounce, which will waste your spending. Landing Page Mismatch: Check the site speed and mobile responsiveness. A slow-loading page is the number one cause of spending but not converting. In ad platforms like Realize, make sure the pixel is active. This will ensure the algorithm has the information it needs to optimize for conversions. ### 3. How to Troubleshoot Unprofitable or High CPAs After you’ve done a conversion tracking audit and addressed those issues, your next step is to get into the details of unprofitable or too-high CPA numbers. Bidding Strategy Misalignment: To troubleshoot high CPA, look at what your platform can tell you. Using AI-driven, automated tools, such as Maximize Conversions, can be incredibly helpful for performance marketing teams. However, make sure you’re auditing how they’re working. In Realize, the pacing health score shows a percentage score; if it’s red or yellow — over 110% — you’re spending too fast. This leads to higher CPAs. Your platform can give you feedback about the pacing of spending. Limited Learning Phase: These same AI-driven tools rely on algorithmic learning, which requires a little time to calibrate (and then optimize) bidding. If you’re changing bids or budgets too frequently — more than once every 48 to 72 hours — the algorithm has to reset. Too much tinkering, and your campaign will end up stuck in a permanent learning phase, preventing the algorithm from finding the most cost-effective placements. The 10x Budget Rule: This rule applies to CPA-based campaigns, which need a certain amount of liquidity to stabilize. Your daily budget should ideally be 10x the target CPA, which lowers the average cost over time. So, if your target CPA is set at $20, but your daily budget is $50, the algorithm doesn’t have enough liquidity to learn what works and put it into practice. ### 4. Eliminating Unsatisfactory Site Placements and Wasted Spend Finally, try to uncover placements that aren’t working for your brand and wasting precious budget. Take a look at: Default Settings: These can be a hidden source of budget drain. Review your reporting by both site and platform to see if any settings don’t make sense for your campaign. For example, the high CPA might be coming from a specific device. If mobile is taking up 90% of the budget, but zero conversions are happening there, use the Platform Targeting feature in Realize to shift spending toward desktop instead. Proactive Waste Reduction: Identify sites with high spend and zero conversions, confirm that those sites are not a fit for your ads, and then use the Block Site feature to exclude them immediately. Within Realize, for example, you can proactively block specific publishers or categories that aren’t a fit, such as Gaming or Hard News, to stop spending budget there without any ROI. Search Terms and Macros: Check if macros (like {site}) are showing traffic from irrelevant apps or low-quality networks, and regularly review the search terms report. Apply category exclusions (e.g., excluding “Gaming” or “Hard News”) if those placements aren’t driving ROI. ## Key Takeaways Campaign failure issues are usually technical or structural, not creative, which means they’re fixable. It’s important to ensure that everything is aligned for the algorithm to work correctly: diagnose ad delivery issues or blocks (such as account freezes), fix broken tracking via a tool like the Pixel Test, remove wasted placements, and respect the SmartBid optimization learning phase. Diagnosing and fixing issues — and then optimizing over time — is a process of elimination. ## Frequently Asked Questions (FAQs) ### Why is my PMax/Realize campaign spending budget but not getting any conversions? Common root causes of this issue are broken conversion tracking, optimizing for the wrong conversion action, or having Final URL Expansion enabled. This allows the possibility of the algorithm sending paid traffic to low-intent pages (such as a blog or careers page) instead of your product pages. Run a Tracking Test to make sure your pixel is active. ### How long should I wait before changing my target CPA? Give the algorithm enough time to work. Wait until it has enough data, which is typically 30 to 50 conversions over a seven-day period. Setting a target CPA goal too early, or setting it too low, restricts the algorithm during its learning phase and can completely halt ad delivery. ### How long does it take for a campaign to exit the learning phase? As a rule of thumb, the algorithm needs 30 to 50 conversions over a seven-day period to fully optimize. During this period, try not to make major changes to your budget or CPC number, as this will reset the learning process. ### How can I tell if my bid is too low? If your campaign is set to Active, but has zero impressions, your bid is likely below the competitive threshold for the sites you’re targeting. Try increasing your CPC by 20% to 30% to start seeing initial traffic. ### Why are my ads getting high click-through rates but no sales? The problem of high click-through rates but zero conversions usually indicates a post-click funnel issue. Causes can include a disconnect between the ad’s promise and the landing page, a friction-heavy checkout process, or a tracking pixel misfire that isn’t recording the sales that are actually happening. Within Realize, check pixel health and use the Tracking Test Tool to ensure the conversion event — Lead or Purchase — is mapped accurately to the landing page. If the pixel isn’t firing, Realize won’t report conversions, even if they are actually happening. ### What does it mean if my account status is “Frozen”? A “Frozen” status in your ad platform likely reflects a billing issue, such as an expired credit card or a maxed-out account credit limit. Updating the card or adding account credit will fix the issue — in Realize, campaigns typically resume within a few hours. --- ### App Promotion: Scaling on the Open Web with AI URL: https://www.taboola.com/marketing-hub/app-promotion/ Last Modified: 2026-06-30 07:38:50 An average of 3,205 apps are released on Apple’s App Store daily. As the competition for people’s attention intensifies, relying solely on saturated search and social channels is no longer enough to drive cost-effective growth. For performance advertisers, the open web presents a massive, untapped opportunity to acquire high-intent users at scale. By using artificial intelligence (AI) to predict user behavior, optimize cross-channel placements, and dynamically personalize creatives, marketers can build sustainable, long-term return on investment (ROI) using the latest AI-driven web technologies. This guide breaks down exactly how to do that. ## What Is App Promotion in Today’s Digital Landscape? App promotion is the practice of driving installs, engagement, and revenue from your mobile app through paid and organic channels. It sounds simple enough. The execution, though, has gotten considerably more complex over the last several years. A few years ago, getting your app in front of users mostly meant showing up in App Store search results and running a few ads on Google or Facebook. That was enough. Today, those channels are crowded, expensive, and increasingly restrictive in how they let you track and attribute results. Privacy changes like iOS 14.5 reshaped mobile measurement overnight. Auction inflation has pushed up costs per install (CPIs) on major platforms. And audiences on the big walled gardens, like Facebook, LinkedIn, and Google, have, in many categories, gotten fatigued. Modern app promotion demands a more distributed approach. One that spans multiple channels, uses smarter targeting, and treats a download not as the finish line but as the beginning of a longer user relationship. ## Unlocking the Open Web for Scalable App Growth The open web is everything outside the closed ecosystems of Google, Meta, Apple, and Amazon. It includes premium publisher websites, news outlets, niche blogs, content networks, and millions of independent sites where people spend a significant chunk of their online time. For app marketers, this is largely untapped territory. Audiences on the open web are actively engaged with content they chose to read. That creates a different, and often more receptive, mindset than the interruption model of social feeds. There’s also significantly less auction competition than on the major walled gardens, which translates directly into lower acquisition costs for advertisers willing to go where others haven’t. While platforms like Google, Facebook, and Instagram offer scale and familiarity, they also have shortcomings. Cross-platform attribution is harder. Signal loss from privacy restrictions is ongoing. And audience pools in saturated categories don’t refresh the way open-web audiences do. A user acquisition strategy that depends entirely on two or three major platforms is fragile by design. The open web solves for all three. It adds fresh audience supply, contextual relevance, and a diversification layer that stabilizes performance when any single platform has a bad week. Platforms like Realize by Taboola give you access to premium publisher inventory across thousands of sites, with sophisticated targeting options and app promotion campaigns built specifically to drive installs at scale. ## How AI Is Revolutionizing App Promotion on the Open Web Running app install campaigns across the open web used to require a lot of manual work. You’d pick placements, set bids, test creatives by hand, and wait days for meaningful data. AI has replaced most of that guesswork with automated, real-time decision-making that improves as it learns. Here’s what app promotion looks like in practice: ### Predictive Audience Targeting AI-powered ad targeting goes deeper than traditional targeting, which mainly relies on demographic and interest categories. It analyzes behavioral signals across the web, looking at what someone reads, how long they spend on a page, what they’ve engaged with recently, and builds predictive models that identify users with the highest probability of not just installing your app, but sticking around and converting. An install from someone who churns in 48 hours costs the same as an install from someone who becomes a paying subscriber. But they’re worth very different amounts to your business. AI systems trained on post-install event data learn to distinguish between those two users before they click, so your budget flows toward the installs that actually generate revenue. Realize’s contextual targeting does this at the publisher level too, matching your app to articles and content environments where readers are already in the right mindset to convert. ### Real-Time Creative and Bid Optimization AI also handles the moment-to-moment decisions that would be impossible to make manually at scale. When a user matches a high-intent profile, the system adjusts the bid in milliseconds to win that impression. When one creative variant is outperforming another, it reallocates spend accordingly without waiting for a human to run a report. Tools like GenAI AdMaker, built directly into Realize, let you generate and test high-quality creatives quickly, without a design team or production bottleneck. This continuous optimization loop means your mobile app growth campaigns get more efficient over time. The system gets smarter the more data it processes, which creates a compounding advantage for advertisers who start early. ## Getting Recommended by AI: SEO and Discovery Outside the App Store A growing number of users don’t start their search in the App Store. They ask ChatGPT, Perplexity, or an AI-powered search feature something like “what’s the best budgeting app for freelancers,” and they act on whatever gets recommended. Those recommendations typically don’t come from App Store listings. Instead, they come from the open-web content that those AI systems have indexed and learned from. If your app doesn’t have a presence in comparison guides, review articles, and landing pages that clearly explain who your app is for and what problem it solves, you’re invisible to this growing discovery channel. This means content on the open web is now part of your user acquisition strategy, whether you treat it that way or not. Apps that build indexed, well-structured web content — written around the specific questions their target users ask — show up in AI recommendations. Apps without that content don’t get cited. Of course, App Store Optimization (ASO) still matters, but it’s no longer sufficient on its own. ## Core Foundations: App Store Optimization (ASO) and Landing Pages Before you scale web-to-app conversion, the destination needs to convert. Two foundations matter here: your App Store presence and your web landing page. On the ASO side, the basics still do a lot of work. A clear, compelling app icon. A description that speaks to a specific user problem in plain language. Screenshots that show the app in actual use rather than abstract brand visuals. And a steady stream of genuine user reviews that build social proof for users who are on the fence. Your web landing page serves a slightly different function. It’s where users land before they hit the App Store, and it needs to do the trust-building work that a short app description can’t. A good landing page explains who the app is for, shows real use cases, surfaces reviews and ratings, and makes the download step feel like the obvious next move. Think of it as a conversion layer between your ads and the App Store listing. Taboola’s own testing found that driving users directly to the App Store via the App Install format reduced cost per acquisition (CPA) by 29% compared to routing users through a landing page first. Both approaches have their place depending on your goals, but for pure install volume, cutting out the extra step makes a real difference. ## Crafting High-Converting Ad Creatives for Open Web Audiences Open-web ad creatives work differently from social media ads. Users on publisher sites are in a content consumption mindset, not a scrolling-and-reacting mindset. Ads that feel disruptive or overly promotional perform poorly, while the ones that feel like a natural extension of the content perform much better. Native-style creatives tend to win here. Headlines that frame a real problem rather than push a feature. Visuals that look editorially relevant. Copy that speaks directly to the user’s situation rather than listing what the app does. Creator-generated content (CGC) is also worth the investment. Ads built around real users explaining how they use the app in their own words carry a credibility that brand-produced creative can’t replicate. They don’t feel like advertising, which is exactly why they work. Taboola Trends is a useful starting point here. It shows you which creative formats, titles, and visual styles drive the highest click-through rate (CTR) by vertical and region, so you’re not guessing what resonates with open-web audiences in your category. As a rule of thumb, keep motion subtle, include real people, front-load the key message in your headline, and match the visual style to the editorial environment in which your ad appears. ## Deep Linking and Seamless Web-to-App Journeys Technical friction kills conversions. A user who clicks an ad for a specific feature and lands on a generic app store page has already lost context, and many of them don’t navigate to find what they were originally interested in. Deep linking solves this. Tools like Branch let you route users from a web click directly to a specific screen or piece of content in the app, whether they already have the app installed or are going through the install process for the first time. The experience remains consistent across ad and app, and conversion rates reflect that. Realize supports third-party tracking partner URLs natively in its App Promotion campaign setup, so you can connect your deep linking provider without a complicated workaround. For retargeting campaigns in particular, this kind of seamless routing is nearly essential. Users who’ve interacted with your app before respond much better to ads that take them back to something specific rather than ads that dump them at a generic home screen. ## Tracking, Attribution, and Measuring Post-Install KPIs Downloads are a vanity metric if you’re not measuring what happens after the install. Revenue comes from registrations, subscriptions, purchases, and repeat engagement. Your campaigns need to be optimized against those events, not just installs. This is where a Mobile Measurement Partner (MMP) like Adjust, AppsFlyer, and Kochava becomes critical infrastructure. Realize integrates directly with all of these, so post-install events get attributed back to the specific ad, publisher, or channel that drove them. That data closes the loop and lets you see which campaigns are actually generating revenue, not just volume. Without MMP integration, you’re flying blind. You might know that a campaign drove 5,000 installs. You won’t know that 4,200 of them churned in three days, while a smaller campaign on a different channel produced 800 installs with three times the 30-day retention rate. Setting up post-install event tracking before you scale is the difference between optimizing for installs and optimizing for business results. ## Retargeting and Re-Engaging Lapsed App Users Not every lost user is gone for good. People uninstall apps for all sorts of reasons: storage pressure, a busy period, a frustrating experience they’d be willing to try again if something changed. Open-web retargeting lets you reach those users at the right moment, with messaging tailored to where they dropped off. AI makes this more precise. Instead of blanketing all lapsed users with the same win-back message, AI-powered ad targeting can segment by behavior. Users who installed but never completed onboarding get a different message than users who used the app regularly for two months and then stopped. The more relevant the message, the better the reactivation rate. The open web is particularly effective for re-engagement because you’re reaching users in a context outside the app ecosystem, where push notifications are not already bombarding them. A well-placed article or native ad on a relevant publisher site can prompt a reconsideration that in-app messaging never would. ## Key Takeaways App promotion has moved well past App Store search and social feeds. The open web is where sustainable, cost-effective growth lives for performance advertisers willing to build there. AI is what makes open-web advertising viable at scale. Predictive targeting, real-time bid and creative optimization, and post-install event learning all work together to improve efficiency over time and direct spend toward the users who actually generate revenue. AI-powered search and chatbots are also changing how apps get discovered. Web presence — comparison pages, landing pages, indexed content — is now part of the user acquisition strategy, not just a nice-to-have. And diversification matters. Over-relying on a handful of walled gardens creates fragility. A channel mix that includes the open web gives you access to fresh audiences, more competitive CPIs, and a more stable overall acquisition engine. If you’re ready to put AI app marketing into practice, Realize by Taboola is built specifically for performance advertisers running app install campaigns on the open web. The technology is here, the audiences are there, and the opportunity is in being early. ## Frequently Asked Questions (FAQs) ### What is the open web in app promotion? The open web refers to the vast network of independent websites, premium publisher blogs, and news platforms that exist outside of closed walled gardens like search engines and social media platforms. Promoting your app on the open web offers substantial reach, access to diverse audiences, and often much lower acquisition costs compared to major platforms. ### How does AI improve app user acquisition? AI enhances campaigns by analyzing large volumes of real-time data to predict user behavior, automate bidding, and dynamically adjust ad creatives. The result is that your budget flows toward users who are statistically most likely to install the app and complete valuable post-install events, not just users who click. ### Why should app marketers diversify beyond search and social media? Over-reliance on a few major platforms tends to push up user acquisition costs, create audience fatigue, and leave campaigns vulnerable to policy changes. Diversifying to the open web provides access to fresh, contextually relevant audiences while stabilizing overall campaign performance and ROI. ### How is AI changing app discovery outside the App Store? Users are increasingly turning to AI search engines and conversational chatbots for app recommendations. Because these AI systems draw on indexed web content, apps that build a strong open-web presence through comparison guides, landing pages, and review content are more likely to be cited than apps that rely solely on App Store Optimization. --- ### Made-for-Advertising (MFA) Sites Explained URL: https://www.taboola.com/marketing-hub/made-for-advertising/ Last Modified: 2026-06-30 07:35:31 Every performance advertiser eventually reaches a tipping point where the walled gardens of search and social start feeling less like a breath of fresh air, and more like a crowded, high-priced elevator. Open web advertising offers somewhat of an escape hatch for this situation, as it brings massive scale, diverse audiences, and distance from the latest algorithm “mood swings.” The trouble is, once you begin exploring the open web, you’ll notice the terrain is littered with hidden economic landmines, particularly made-for-advertising (MFA) sites. These digital ghost towns exist for a single purpose — ad arbitrage. They use shocking clickbait and generative AI to trick programmatic systems into buying worthless impressions, draining budgets without delivering a single real customer. The good news is that the tide is turning (or, at least, starting to turn). By weaponizing AI-driven ad tech, smart marketers are actively filtering out this digital pollution to secure high-converting, genuine human engagement across the open web. ## What Are Made-for-Advertising (MFA) Sites? Think of MFA sites as the digital equivalent of a tourist-trap souvenir shop. They’re flashy on the outside and completely hollow on the inside, engineered solely to separate you from your money. An MFA site is a web property built from scratch for ad arbitrage. The business model is incredibly simple but highly cynical: The operator buys traffic from social media feeds or discovery widgets using sensational headlines (like “You won’t believe what this 90s child star looks like now!”). Once a real human clicks that link and lands on the page, they are immediately ambushed by an overwhelming wall of ads. If the operator buys the initial click for $0.05 and serves enough stacked ads on the page to make $0.08 off programmatic buyers, they pocket a clean profit. The user gets zero value, the publisher gets a micro-payout, and the advertiser gets an impression that’s completely worthless. ## Anatomy of an MFA Site: Recognizing the Red Flags If you ever stumble onto an MFA page accidentally, your browser will likely lag instantly. That’s because the layout is explicitly designed to maximize ad density at the expense of user experience. Here are the classic hallmarks of an arbitrage factory to watch out for: - Extreme Ad Stacking: Ads are shoved into every available pixel — top, bottom, sides, and directly on top of the text. - The Infinite Slideshow: A 200-word article is chopped up into a 50-page slideshow, forcing the user to click “Next” repeatedly just to read a single sentence, refreshing a new batch of ads with every click. - Autoplay Video Overload: Multiple sticky video players pop up simultaneously, floating along as you scroll and playing muted ads. - Accidental Click-Baiting: Buttons are positioned right next to close markers, making it easy for the user’s thumb to misclick and the site to log a false engagement metric. ## The Double-Edged Sword of AI: How Generative AI Fuels the MFA Boom In the past, running an MFA network required a modest amount of human effort. Operators had to hire cheap content writers or use basic RSS scrapers to build their thin articles. Not anymore: Generative AI content has supercharged the dark side of programmatic advertising. Bad actors now use automated large language model (LLM) scripts to scrape mainstream news, rewrite it slightly to bypass plagiarism filters, and publish tens of thousands of low-quality pages a day. This automated explosion makes manual blocklists completely obsolete. By the time an agency manually flags and blocks the URL, the operator has generated a hundred new domains with entirely different names. ## The Gray Area: Is MFA Considered Ad Fraud or Invalid Traffic (IVT)? This might be the most frustrating thing for performance marketers: technically, most MFA sites are completely legal, and they don’t trigger traditional anti-fraud alarms. Here’s a quick breakdown: Traditional Ad Fraud (IVT): - Uses non-human bots. - Domain spoofing tactics. - Blatantly illegal. Made-for-Advertising (MFA): - Uses real human traffic. - Attracted via clickbait. - Technically legitimate. Because MFA networks use paid ads on major platforms to lure real humans to their pages, their traffic doesn’t register as Invalid Traffic (IVT). There aren’t any botnets clicking the ads, just confused people trying to read a slideshow. But, since they are humans, legacy cybersecurity filters give it a green light. This structural loophole is why you can’t rely on basic anti-fraud tools to protect your media spend; you need an entirely different layer of defense. ## The Hidden Costs: How MFAs Drain Performance Budgets When 15% of your total programmatic ad spend is vanishing into a black hole of empty impressions, your down-funnel performance takes a severe hit. This traffic possesses zero commercial intent. Users didn’t visit this site because they were researching a product. Instead, they were tricked into clicking a sensationalized link. Buying these placements drives up your customer acquisition costs (CAC) and completely tanks your true return on ad spend (ROAS). ## Why Traditional Programmatic Metrics Fail to Catch MFAs At this point, you may be asking, “If MFA sites are so bad, why do automated media buying algorithms keep buying them?” The simple answer is: MFA sites are explicitly built to look like star performers on a standard media dashboard. Traditional viewability metrics are flawed because they only track if an ad was technically rendered on a screen for a couple of seconds. MFA sites game this system perfectly by using sticky, floating ad units that follow the user down the page, scoring near-perfect 98% viewability ratings. Plus, because their traffic is cheap, their CPMs are incredibly low. If your automated campaign settings are programmed to blindly hunt for the lowest cost and the highest viewability, your budget will automatically default to these low-value domains, mistaking accidental screen placement for high consumer engagement. Leveraging AI-Driven Technology for Real-Time MFA Detection If AI created the content scaling problem, then a smarter version of AI is required to fix it. Modern programmatic platforms use machine learning to execute advanced, pre-bid filtering before an ad dollar is ever traded. Instead of relying on an outdated domain checklist, predictive AI models evaluate thousands of page-level attributes in a fraction of a second. The algorithm analyzes the site’s historical traffic sources (checking for abnormal spikes in paid social traffic), calculates the exact ad density relative to the text, detects repetitive generative AI text structures, and checks the age of the domain. If these parameters scream “arbitrage factory,” the AI automatically drops out of the auction, which saves your budget before the bid is made. ## Shifting the Focus: Using AI-Powered Attention Metrics To truly insulate your campaigns from low-value inventory, you need to change what you optimize for. Leading performance brands are moving away from basic viewability and adopting AI-powered attention metrics. Attention models go incredibly deep, tracking active human engagement signals, such as scroll velocity, device orientation, and precise dwell time. An MFA site might keep an ad on a screen for ten seconds, but attention data will reveal that the user was scrolling at lightning speed to escape the layout, resulting in zero human eyes on your creative. By training your optimization algorithms to buy placements that command real human focus, your ads naturally move away from clickbait farms and toward high-quality, reputable publisher environments where users are genuinely reading the content. ## Supply Path Optimization (SPO): Curating Premium Open Web Inventory Protecting your media spend also requires cleaning up the pipeline between your brand and the publisher. This practice, known as supply path optimization (SPO), uses machine learning to map out the most direct, transparent, and cost-effective pathways to programmatic inventory. Instead of navigating a maze of confusing ad tech middlemen — many will bundle MFA inventory into their standard packages — SPO cuts out the clutter. By using dynamic inclusion lists and attentive private marketplaces (aPMPs) curated by smart traffic intelligence, you ensure your media dollars skip the arbitrage toll booths entirely and land exclusively on premium publisher networks. ## Expanding Beyond Walled Gardens: The Open Web Opportunity Full transparency here: Relying entirely on Meta and Google is somewhat of a dangerous game. Media buyers are tired of watching their acquisition costs creep upward while having zero control over platform rules. That’s where the open web represents a land of opportunity, one where you can diversify your media mix and scale campaigns efficiently. However, buying programmatic inventory outside the walled gardens requires you to be sharper and savvier. The open web is a decentralized ecosystem, meaning it doesn’t have built-in quality control. If you blast your budget out via traditional programmatic channels without a solid filter, you’ll inevitably run into low-quality supply traps designed to siphon off your cash before you even know what hit you. ## Best Practices for Scaling Open Web Performance Safely Want to scale your campaigns across the open web without getting caught in the arbitrage trap? Implement these guardrails: - Turn on Pre-Bid AI Filters: Ensure your DSP has live, predictive AI filters enabled to identify new MFA domains in real time. - Ditch Top-of-Funnel Optimization: Stop optimization algorithms from blindly chasing ultra-low CPMs and superficial click volume; this can inadvertently funnel spend directly to ad-heavy layouts. - Focus on Business Outcomes: Tie your campaign performance tracking to down-funnel milestones, including verified leads, add-to-carts, or direct purchases. - Demand Supply Chain Transparency: Partner exclusively with clean, trusted networks that offer direct supply paths to premium, authenticated publishers. ## Key Takeaways The open web remains one of the most powerful places to scale your digital marketing and discover new customer segments, but you can’t navigate the open web using old, easily-gamed metrics, such as simple viewability or low cost. By upgrading your toolkit with AI-driven, real-time pre-bid detection, shifting your focus toward true attention metrics, and utilizing direct supply channels to premium publishers, you can effectively starve MFA networks of your capital while maximizing your true ROAS on high-quality web real estate. ## Frequently Asked Questions (FAQs) ### What is the difference between an MFA site and ad fraud? Traditional ad fraud relies on illegal methods, such as non-human bot traffic or domain spoofing, to create invalid traffic (IVT). MFA sites operate in a legal gray area because they bring real human eyes to their pages via cheap, deceptive clickbait links. Because the traffic is from humans, it slips past standard fraud detection software, even though the layout is engineered exclusively to drain ad spend, rather than providing editorial value. ### How much ad spend is currently wasted on MFA sites? A landmark programmatic media supply chain study by the Association of National Advertisers (ANA) revealed that MFA properties account for an alarming 21% of all digital ad impressions and capture roughly 15% of total programmatic ad spend. ### Why do MFA sites often have such high viewability scores? MFA sites are specifically built to exploit programmatic bidding algorithms that reward cheap, viewable placements. They use tactics like sticky floating video players, hidden ad stacking, and infinite layouts to keep ads visible in a browser window for long periods, even if the user is completely ignoring them. ### How does AI help advertisers avoid MFA inventory? Bad actors can generate new MFA sites faster than any human team can update a blocklist. AI-backed programmatic tools solve this by evaluating thousands of live, pre-bid indicators — lopsided ad-to-content distribution ratios, high percentages of paid incoming traffic, and repetitive, generative AI filler text — to identify and block these properties before a bid is ever submitted. --- ### Ad Management with AI Agents: How to Scale Without the Grind URL: https://www.taboola.com/marketing-hub/ad-management-with-ai-agents/ Last Modified: 2026-06-30 08:04:48 Managing modern ad campaigns feels like trying to play a hundred games of chess simultaneously. It’s a complexity trap where the sheer volume of assets and channels eventually breaks even the best player. To stay competitive, you need to test thousands of creative pieces, but doing that manually is a path to guaranteed burnout for you and your team. This is exactly why modern agencies are aggressively adopting marketing workflow automation, to take the administrative weight off their media buyers’ shoulders. AI advertising agents change the game. They aren’t just fancy dashboards — they’re autonomous partners that build, tweak, and scale campaigns in real time. Here’s how to use agentic marketing to juggle 1,000+ variations without losing your mind (or your ROI). ## Ad Management in the Era of AI In the old days (i.e., about three years ago), ad management meant manual labor: staring at spreadsheets, toggling budgets, and hoping your manual adjustments didn’t break the algorithm. It was a reactive, and tedious, process. Today we’ve moved more into the era of autonomous execution. By deploying advanced AI ad optimization techniques, the system learns and reacts to market changes continuously. AI advertising agents leverage large language models (LLMs) to do more than just report data — they interact with it. These agents function as 24/7 media buyers that can reason through a problem (like a sudden spike in CPC) and execute a fix via API commands instantly, shifting the focus from manual clicking to high-level supervision. ## The Advertiser’s Complexity Trap: Why Manual Management Fails The demand for hyper-personalized content has created a massive bottleneck. If you’re trying to manually oversee 1,000+ ad variations across Meta, Google, and TikTok, you aren’t really optimizing, you’re just surviving (and pretty much drowning). The standard approaches to digital campaign management fail the moment a business attempts to scale their operation across multiple fast-moving online channels simultaneously. When a manual, human team is overwhelmed, optimization cycles stall. Underperforming ads stay live too long, and winning creatives aren’t scaled fast enough. This manual lag is where most ad budgets go to die, resulting in wasted spend and a direct hit to your ROI. ## Traditional Ad Campaign Management Tools vs. AI Advertising Agents It’s easy to confuse these, but the difference is fundamental: - Traditional tools: These are rules-based systems. They follow strict “if/then” logic (for example, “If the click rate drops, pause the ad”) and require a human to set every single parameter. - AI agents: These are perception-based systems. They use APIs to ingest live data, reason through the current market context, and make decisions without needing a human to pre-write a rule for every possible scenario. Don’t think of AI agents as just faster versions of old automated rules, as that’s just not accurate. Unlike legacy automation, agents possess autonomous reasoning. They can interpret why a campaign is failing and pivot strategies across different platforms without a human needing to click a single button. ## How AI Agents Solve the 1,000+ Ad Variations Problem Scaling isn’t just about quantity, it’s about maintaining quality without burning out your staff. When you use smart tech to scale ad variations dynamically, you multiply your market touchpoints without multiplying your team’s workload. ### Automated Ad Creative Generation and Localization AI agents handle the heavy lifting of asset production. They can draft copy, resize images for specific platform requirements, and translate phrasing for international markets. They can do all this while still adhering to your brand guidelines. This ensures your message stays consistent across 1,000+ variations without a designer or copywriter needing to touch every file every time. ### Continuous A/B Testing and Personalization While a human might check a test once a day, an AI agent checks it every second. It identifies creative fatigue (that moment when an ad stops working) and automatically rotates in fresh variations tailored to specific audience segments. ## Core Capabilities of AI-Driven Ad Management Platforms ### Autonomous Bid Adjustments and Budget Reallocation Agents watch your pacing 24/7. If a specific ad group on one platform is performing 20% better than another, the agent can autonomously move the budget to the winner to maximize spend efficiency. ### Unified Cross-Channel Orchestration Instead of logging into separate portals, agents use a single hub to manage Meta, Google, and LinkedIn. This unified style of cross-channel ad management gives the agent a holistic view of your funnel, allowing it to reallocate resources based on true omni-channel impact, rather than siloed platform metrics. It allows for more of a big picture type strategy, where the AI understands how a view on one platform leads to a search on another. ### Real-Time Fraud Prevention Agents use machine learning to spot and block bot traffic as it happens. This keeps your data clean and ensures you aren’t optimizing your campaigns based on fake clicks. ## How Agentic Advertising Transforms Workflows The real shift is moving to conversational commands through model context protocol (MCP). Instead of digging through menus, you can simply tell your agent something like, “Increase spend by 15% on any ad with a ROAS over 3.0,” and it executes the command across your entire stack instantly. By tying your natural language instructions directly to ROAS optimization tools, you ensure the AI focuses entirely on high-intent bottom-of-funnel actions rather than cheap, empty clicks. ## Top AI Ad Management Software and Tools for 2026 - Legacy with AI: Platforms like Skai, AdRoll, and HubSpot have added smart features to their existing dashboards to help with automation. - AI-native agents: Newer platforms like Ryze AI, Adspirer, and Warmly are built as “agents first,” focusing on autonomous execution through conversational interfaces. - Realize+ (Beta): This uses agentic AI to automate ad campaign management, continuously making and executing decisions regarding budget allocation, creative optimization, and targeting without requiring constant human intervention. ## 4 Best Practices for Transitioning to Agentic Ad Management ### 1. Start small Don’t automate everything at once. Pick one channel to really get a feel for what the agent’s logic is before scaling. ### 2. Human-in-the-loop (HITL) Use the agent to generate 1,000 variations, but keep a human in the mix to approve the core strategy and brand voice. ### 3. Establish clear guardrails Define your maximum budget limits, target conversion boundaries, and strict brand exclusion rules inside the platform before launching. These guardrails keep the autonomous system operating within safe parameters. ### 4. Audit down-funnel performance regularly Monitor the data downstream to ensure the conversions the AI is optimizing for are translating into real pipeline value and bottom-line revenue. ## Overcoming Data Privacy and Security Concerns Security is paramount when giving an AI the keys to your ad accounts. While granting access doesn’t mean your data is public, you still need to verify that your vendor is SOC 2 compliant, and has strict data retention policies to ensure your competitive data isn’t used to train models for other users. ## Key Takeaways The transition from button-pusher to system architect is the biggest shift in a media buyer’s career. By using AI agents, you eliminate the repetitive grind, stop wasting budget on human error, and finally have the time to focus on the high-level creative strategy that actually, and actively, gets attention. ## Frequently Asked Questions (FAQs) ### What is the difference between standard ad management software and AI advertising agents? Standard software is a dashboard for manual use, while AI agents are autonomous systems that perceive data and execute changes via APIs, without you needing to click anything. ### How do AI agents manage ad variations without losing brand consistency? You train them on your guidelines. They generate the variations, but you can set up HITL checkpoints to review the work before it goes live. ### Can AI agents manage ads across multiple networks at once? Yes, and pretty easily. They integrate with APIs for Google, Meta, and others to move budgets and ads across platforms from a single interface. ### Will AI agents replace human media buyers? No, and this is a common misconception. They replace the repetitive work that humans are currently doing. Media buyers shift their focus to strategy and audience psychology while the agent handles the manual labor. --- ### Campaign Reporting Best Practices for Performance Advertisers URL: https://www.taboola.com/marketing-hub/campaign-reporting-best-practices/ Last Modified: 2026-06-25 10:44:39 Most marketers have dashboards full of data, yet still struggle to answer the simple question of whether their campaign is working. Numbers alone don’t drive decisions and context is critical. Mastering campaign reporting best practices means transforming raw data into a consistent feedback loop that informs every move you make around your campaign. With platforms like Realize providing granular, real-time data, there’s no longer a reason to wait until a campaign ends to take action on what that data is telling you. ## Why Campaign Reporting Matters Before Your Campaign Even Starts Many people treat campaign reporting as a retroactive exercise, something you do to justify spend once the campaign has already ended. The most effective marketers, however, think of reporting as a proactive framework built into the campaign architecture from the start. Before any impressions are served, your reporting structure should already be defined by what metrics signal success, which audience segments should be compared, and how you’ll evaluate overall creative performance. This is where data storytelling comes in. Think of your report not as a spreadsheet, but as a narrative you’re writing in real time. Building your report framework during the planning phase and defining benchmarks at this point gives you the ability to pivot while the campaign is live, rather than scrambling to explain results once the budget has run out. ## 6 Campaign Reporting Best Practices to Maximize Performance ### 1. Isolate 1-2 Critical Metrics Based on Campaign Goals When every metric matters, pulling too many data points into your report can easily result in analysis paralysis, where insight becomes impossible because the real results are buried under noise. Before launch, decide which one or two metrics will be the primary indicators of campaign success and match them to your goals. For instance, viewable click-through rate (vCTR) for awareness campaigns, and cost per action (CPA) for revenue-focused campaigns. When reviewing your advertising metrics, ask yourself if you would know if your campaign is on track if that was the only number you had access to. Secondary metrics can still be monitored, but they should be informing your primary key performance indicator (KPI), not competing with it. ### 2. Leverage Real-Time Reports for Mid-Flight Action One of the most underused capabilities in modern advertising platforms is real-time campaign reports. In Realize campaign reporting, real-time data highlights underperforming publishers or placements long before a significant portion of your budget has been spent, opening up mid-flight campaign optimization opportunities that would otherwise go unnoticed. Alongside placement performance, monitor pacing health closely. Pacing alerts tell you whether your campaign is spending at the correct rate to fully utilize your budget within the flight window. Catching underpacing or overpacing early can be the difference between a successful campaign and one that falls short. ### 3. Understand and Align Your Conversion Attribution Models Attribution is often one of the most misunderstood elements of campaign analysis. Aligning your conversion attribution models before launch ensures your data reflects how customers actually behave. The key distinction between click-through vs view-through conversions is important to understand. Click-through conversions are recorded when a user clicks and converts within a set attribution window (usually 30 days), while view-through conversions credit a user who saw your ad without clicking and later converted directly on your site, often within 24 hours. Using both models together gives a more complete picture of your advertising ROI measurement, accounting for both direct response behavior and the influence of impression-based brand recall for a later conversion. ### 4. Track “Change History” to Repeat Success Optimization without clear documentation is guesswork. Change History tracking logs every adjustment made to a live campaign, including bid changes, budget reallocations, and site boosts, then correlates those actions to subsequent performance shifts. If you boosted a publisher placement on a Tuesday and your CPA dropped 18% by Thursday, that’s an important and replicable insight. This structured documentation shifts your campaign analytics dashboard from a snapshot tool into an institutional knowledge base, with every campaign building a playbook to make the next one more successful. ### 5. Customize Dashboards to Eliminate “Chart Clutter” A well-designed dashboard and good data storytelling are inseparable. The more columns and macros you add to a reporting view, the harder it becomes to identify what actually needs your attention. Use custom table views to drag-and-drop columns so only the most relevant dimensions and metrics are visible, whether that's for publisher-level analysis using the Site dimension, or device breakdowns using the Platform dimension. Build your views around your primary KPIs and the specific questions you’re trying to answer. The goal isn’t to show everything, but to surface the right data at the right time. ### 6. Establish a Consistent Cadence Before making optimization decisions, ensure your campaign has accumulated sufficient performance data and allowed the algorithm adequate time to learn. Campaigns that are adjusted too early—before enough conversions or meaningful traffic patterns have emerged—often suffer from premature optimization. Use your Budget Pacing Health Score and performance trend reports to confirm that spend and delivery have stabilized, indicating the campaign has moved beyond its initial learning phase and into a state where data-driven adjustments will be effective. Establish a regular review routine, checking real-time pacing and key performance indicators every few days to a week after launch, with deeper analysis reserved for weekly or post-campaign reviews. This keeps you informed, without pushing you into making constant updates. ## Common Campaign Reporting Pitfalls to Avoid Over-optimizing and small data samples are a common issue for marketers. Early data never represents the full campaign distribution. Pausing a creative or reallocating budget after just 24-48 hours of data can eliminate options that would have performed well with more runway. Ignoring pacing alerts is also something that can create problems. These alerts tell you whether your campaign is spending at the correct rate to fully utilize your budget within the flight window. A severely underpaced campaign won’t generate enough data to be meaningful, while an overpaced one will exhaust budget before reaching your most valuable audience window. ## Key Takeaways Effective campaign reporting best practices are not about gathering more data, they’re about gathering the right data, in the right format, at the right time. Start by defining your reporting framework and primary KPIs before launch, then use real-time campaign reports and pacing health to monitor and make informed adjustments during the campaign run time. Align on your conversion attribution models early so that your performance data reflects actual customer behavior across both click-throughs and view-throughs. Document every optimization in a Campaign History log so wins can be easily replicated and mistakes avoided. Keep your campaign analytics dashboard clean and KPI-focused, then allow for enough time for the algorithm to learn before making any significant changes. ## Frequently Asked Questions (FAQs) ### How often should I review my campaign reports to ensure they are on track? Best practice is to check real-time campaign reports within 48 hours to a week after launch, to monitor pacing health and confirm that key KPIs are on track. Deeper strategic review can happen weekly or post-campaign. ### What is the difference between click-through and view-through conversions? Click-through conversions occur when a user clicks an ad and converts within a set window, typically up to 30 days. View-through conversions are credited when a user sees a viewable impression without clicking, then converts on your site, usually within 24 hours. Tracking both gives a fuller picture of your advertising ROI measurement across the entire customer journey. ### Why is tracking “campaign history” important in reporting? Campaign history maps specific optimizations such as bid adjustments or site boosts, which can result in direct shifts up or down in campaign performance. Tracking these changes allows teams to pinpoint which actions drove results, so that they can be replicated or avoided in future campaigns. Over time, this helps individual campaign decisions to be documented as part of the bigger picture playbook that impacts all marketing campaigns across the team. --- ### Campaign Set-Up Best Practices for Advertisers on the Open Web URL: https://www.taboola.com/marketing-hub/campaign-set-up-best-practices/ Last Modified: 2026-06-30 08:47:18 For performance advertisers, a campaign’s success is largely determined before it ever goes live. Fractured ad sets, misaligned bid strategies, and gaps in tracking can drain budget and keep campaigns in the algorithmic learning phase indefinitely. This post covers five essential campaign setup best practices to ensure your campaigns in Realize are structurally built for maximum return on ad spend (ROAS), efficient scaling, and clean attribution as soon as your campaign is published. ## 5 Practices for Your Performance Campaign Set-Up ### 1. Establish Strict Naming Conventions and UTM Parameters Naming conventions may seem like a more administrative task, but they’re the foundation of every optimization decision you make later in your campaign. Consistent campaign naming conventions give you a reliable way to filter, compare, and report across campaigns without relying on memory or manual notes. A strong naming structure follows a logical hierarchy: Platform_Market_Product_Audience_Date. For example: Realize_US_HighHeels_Pros_Nov23. At a glance, anyone on your team can see exactly what this campaign is, who it targets, and when it launched. You also need to add UTM tracking parameters to every destination URL so that sessions map cleanly to your customer relationship management system (CRM) and analytics platform. In Realize, this is where Taboola Macros become essential. Dynamic parameters like {campaign_id} and {site} auto-populate with campaign and publisher data at the impression level, giving you one-to-one attribution back to your source. Without them, you’re making optimization decisions on incomplete data. ### 2. Consolidate Your Ad Sets to Prevent Audience Overlap A common misconception in performance advertising is that running more campaigns gives you more control. In practice, fragmentation is one of the fastest ways to limit your performance. Modern bidding algorithms, including Realize’s Maximize Conversions and Enhanced CPC models, require a sustained volume of conversion events to learn and optimize effectively. When you fracture your budget across a dozen micro-targeted campaigns with overlapping audiences, two things happen. First, each campaign receives too few conversion signals to exit the algorithmic learning phase. Second, your own campaigns begin competing against each other in the same auctions, artificially inflating your costs. Ad set consolidation solves both of these problems. It’s common practice among advertisers on Meta’s systems. On the open web, Realize uses the terms campaign consolidation or campaign grouping to describe this method. By combining related targeting segments into fewer, higher-budget campaigns, you concentrate conversion data where the algorithm can actually use it. Running new campaigns for a minimum of seven to 10 days to establish a reliable baseline is important for data gathering, but that baseline only happens if each campaign has enough daily budget and reach to generate signals. The result is campaigns that exit the learning phase sooner, bid more efficiently, and provide cleaner performance data for your reporting. ### 3. Isolate Funnel Stages Using Strategic Exclusions Running a prospecting campaign and a retargeting campaign simultaneously, without proper boundaries, is one of the most common and costly structural errors in performance advertising. Without funnel segmentation, your cold-traffic acquisition budget starts reaching users who already know your brand, skewing your cost per action (CPA) data and wasting spend that should be driving new customers. There’s an easy solution here. In the Audiences section of your Realize campaign, add exclusions before you launch. Specifically, exclude your “My Customers” audience and any “Past 30-Day Visitors” segments from prospecting campaigns. This creates a boundary between cold and warm traffic. Realize supports My Audiences built from Taboola Pixel data alongside Marketplace and Contextual audiences. Use these segments in retargeting campaigns and as exclusions in prospecting ones. The goal is a funnel where each campaign talks to the right user at the right moment, with no overlapping budget moving between them. This approach also produces more reliable CPA benchmarks. When your prospecting campaigns are reaching cold audiences, the performance data you collect accurately reflects acquisition costs, rather than mixing in warm-traffic conversions that inflate your apparent efficiency. ### 4. Align Daily Budgets with Target Bid Ratios Misaligned budgets are one of the more overlooked structural problems in campaign setup. An advertiser can have the right creative, the right audience, and the right bid strategy, and still see a campaign underperform because the budget doesn’t give the algorithm enough room to move. For target CPA optimization, your daily budget should be set to at least five times your target CPA, with ten times being the recommended standard. When using Maximize Conversions, a daily budget of ten times your CPA goal is recommended to complete the learning phase and reach stable performance. When using Enhanced CPC or Fixed Bid strategies, a five–time daily budget relative to your CPA goal is the minimum viable starting point. This is why the budget-to-bid ratio matters so much. Realize’s auction system runs continuously, evaluating available placements against active bids. If your bid is $2 but your daily budget is $5, the algorithm has almost no room to explore the auction possibilities. It will reach a daily cap before it can accumulate the conversion events needed to learn and further enhance your campaign. The campaign never gets traction and never exits the learning phase. You see poor performance and pause the campaign, assuming the targeting or creative is the problem, when the real issue was always the numbers. Setting your daily budget with enough flexibility for the algorithm to explore the auction isn’t optional. Campaigns that are underfunded relative to their bid or CPA target can’t generate the conversion data required to learn, and performance will stagnate before the campaign has had a real chance to run. ### 5. Pre-Load Proactive Waste Reduction Measures It’s worth taking steps to protect your budget before the first impression is served. Waiting until a campaign has been running for several weeks before applying exclusions means spending money on placements you may already know are unlikely to perform. Realize offers two key tools for this under Advanced Options. Block Sites lets you prevent your campaign from appearing on specific publisher domains, which is useful for applying exclusions carried over from previous campaigns or industry-standard block lists. Brand Safety Pre-Bid filters placements against article-level content categories, keeping your ads away from topics that don’t align with your brand or audience. Negative keyword lists and publisher exclusions work in a similar way. If you have historical data showing that certain content niches, app placements, or site categories consistently underperform, excluding them at setup means your initial budget goes immediately toward higher-quality inventory, rather than subsidizing the discovery of placements you’ve already disqualified. This is especially important for new campaigns. Leaving Brand Safety settings open on truly new accounts to gather baseline data is helpful for advertisers with no prior campaign history. But, for advertisers launching a new campaign with existing performance data, proactively applying known exclusions from day one is one of your best options. The combination of targeted performance advertising strategies and structured waste reduction is what separates campaigns that learn fast from campaigns that simply spend fast. ## Key Takeaways The difference between a campaign that plateaus and one that succeeds comes down to whether the structure is designed to help the algorithm, or only to get some ads live. Consolidation accelerates learning, isolation protects attribution, mathematical budget ratios create the flexibility the algorithm needs to do its job, naming conventions and tracking parameters ensure every decision you make is based on clean data, and proactive exclusions ensure that from the moment a campaign launches, spend is directed toward inventory that can convert. Together, these are the structural prerequisites for any campaign that you’re intending to scale. ## Frequently Asked Questions (FAQs) ### Why should I consolidate my ad sets during campaign setup? Splitting budgets across too many ad sets or campaigns creates audience overlap and self-competition in the auction, which drives up your costs and limits reach. More importantly, it limits each campaign from the conversion volume it needs to learn. Consolidating ad sets into fewer, better-funded campaigns accelerates data aggregation, helping algorithmic bidding models exit the learning phase and optimize delivery faster. ### What is the ideal budget-to-bid ratio when launching a new campaign? Your daily budget should be at least fives times your target CPA, with ten times being the recommended standard for campaigns using conversion-focused bidding. This gives Realize’s algorithm enough room to participate in auctions, gather conversion data, and move out of the learning phase. A budget that is too close to your bid or CPA target caps the campaign before it can build any momentum. ### How should I handle audience exclusions during setup? Exclusions are a key part of funnel structure. Before launching any prospecting campaign, add your warm audiences, including past website visitors, cart abandoners, and existing customers, as exclusions. This keeps your acquisition spend focused on new users, protects the accuracy of your CPA data, and prevents budget overlap between your prospecting and retargeting campaigns. --- ### Best Platforms for Display Ads URL: https://www.taboola.com/marketing-hub/display-ads-platforms/ Last Modified: 2026-06-25 09:42:20 Since the earliest days of the internet, display advertising has been a constant presence. The days of simply throwing a banner on a page and hoping for the best results are over, though. In a digital landscape cluttered with ad-tech taxes and transparency hurdles, performance advertisers need platforms that do more than just generate impressions — you need engines that drive high-intent discovery, with real outcomes you can track and measure. Whether you’re looking for massive reach or just more precision, choosing the right platform is the difference between a wasted budget and a scaling success story. ## Best Platforms for Display Ads Compared Platform Why It’s Essential Core Use Cases and Features Best For (Performance Advertisers) Pricing Model (Indicative) Realize Performance outcome‑focused display activation and optimization. Key performance indicator (KPI)‑tied display campaign setup, automated optimization, real‑time performance alignment. Performance‑driven advertisers seeking direct return on investment (ROI) measurement. Performance-based model; campaigns billed on CPC basis, or CPM for programmatic. Google Display Network (GDN) Largest global display inventory with robust targeting and automation. Responsive display ads, remarketing, audience segmentation, cross‑device reach. Advertisers seeking wide reach with Google’s optimization power. CPC/cost per mille (CPM)/CPA via Google Ads campaigns. Amazon Demand-Side Platform (DSP) Programmatic display with rich shopping data and cross‑channel reach. Display ads on Amazon properties and third‑party sites, behavior‑based targeting. E‑commerce and performance brands seeking high‑intent audiences. CPM/spend‑based via demand-side platform (DSP). Meta Ads (Display and Audience Network) Massive social‑driven display through Facebook/Instagram and Audience Network. Image/video display formats, detailed demographic/interest targeting. Social‑centric performance advertisers with rich creative assets. CPC/CPM/CPA via Meta platform. Microsoft Advertising (Display) Display placements across Microsoft network (MSN, Outlook, Edge). Contextual and audience targeting, budget importing from Google Ads. Advertisers expanding beyond Google with lower CPMs. CPC/CPM spend‑based. The Trade Desk (DSP) Independent programmatic DSP with advanced targeting. Programmatic buying of display, video, and connected TV (CTV) with detailed audience segments. Agencies and large advertisers needing granular control. Custom enterprise eligibility. Adobe Advertising Cloud Full‑stack advertising platform including display and programmatic. AI‑driven audience segmentation, cross‑channel optimization. Enterprise advertisers. Custom enterprise pricing. AdRoll Cross‑channel display and retargeting focused on performance. Dynamic retargeting, multi‑touch attribution, cross‑network display. Brands combining display with retargeting and cross‑channel campaigns. Spend‑based/subscription. Adsterra Independent display ad network with versatile formats. Pop‑unders, banners, native formats, global placements. Advertisers, especially budget-conscious ones. CPM, CPC, and CPA models. Simpli.fi Unmatched local and localized audience precision. Top-tier geofencing, household-level addressable targeting, and foot-traffic attribution. Direct-to-consumer (D2C) brands with physical footprints or highly localized service areas. Custom quote based on campaign needs. ### Realize Why it’s essential: Realize is an AI-powered performance platform that transforms traditional display advertising into a high-intent discovery engine across the open web. It allows advertisers to scale beyond standard banner ads by utilizing a direct-to-publisher network of over 11,000 premium sites, effectively bypassing the ad-tech taxes and transparency issues often associated with legacy programmatic exchanges. In the context of display ads, you would use Realize to bridge the gap between social media-style engagement and large-scale web reach. The platform is used to dynamically place ads in front of predictive audiences who are most likely to convert, turning passive display impressions into measurable business outcomes like leads and sales. Showcased features: - Social Importer: Effortlessly repurposes your existing top-performing Facebook and Instagram posts into optimized display formats for the open web. - Motion Ads Studio: Enhances static display images with subtle, looping animations that drive significantly higher engagement and click-through rates on mobile and desktop. - Maximize Conversions: Uses deep learning to analyze real-time signals and automatically adjust bids for every display impression, to hit your specific CPA targets. - Select: Provides a curated, MFA-free (made for advertising) ecosystem of premium display inventory to ensure your brand appears only in high-quality editorial environments. - SpendGuard: An automated algorithm that continuously monitors display performance and blocks underperforming sites or creatives in real-time to prevent wasted spend. - Auto Resizer: Detects ad dimensions and matches them to the closest IAB standard aspect ratio, meaning no manual selection is required for images, videos, and third party tags. Best for: Realize is best for performance-driven brands in sectors like D2C, financial services, health, and travel, that need to scale their display efforts without a large internal design or data science team. It’s particularly helpful for advertisers who have seen diminishing returns on social platforms and want an automated way to launch high-quality, brand-safe display campaigns on premium publisher sites. Pricing model: Performance-based model, campaigns billed on CPC basis, or CPM for programmatic. Pros: - By placing display ads within premium editorial environments rather than generic exchanges, the platform ensures your brand is associated with authoritative content, driving higher trust and engagement. - Direct integrations with premium publishers eliminate middle-man fees and provide 100% transparency into where your display ads appear. - The predictive targeting engine ensures display ads are shown to users with a high probability of conversion, not just those likely to click. Cons: - The platform’s most advanced AI optimization and simulation features require a baseline of conversion data to function at peak efficiency. - Display assets imported directly from social media are not editable within the dashboard, which may limit minor design tweaks. - Achieving the most granular display attribution for offline or CRM-based conversions requires a more complex server-to-server setup. ### Google Display Network (GDN) Why it’s essential: The Google Display Network is the world’s largest display ecosystem, reaching over 90% of global internet users across more than 3 million websites and apps. It’s a foundational tool for performance marketers who need to combine massive scale with easy-to-use automation, allowing brands to expand their top-of-funnel reach directly from their existing Google Ads dashboard. It’s popular with performance marketers thanks to its seamless integration with Google’s Demand Gen and Search signals, which allow brands to trigger display ads based on what users are actively searching for elsewhere, capturing consumers when buy intent is high. Showcased features: - Optimized targeting: Uses Google AI to find high-value audiences beyond your manual segments, by analyzing landing page conversion patterns. - Demand-gen integration: Combines the visual reach of display with high-intent audience feeds, like YouTube Shorts and Gmail. - Custom intent segments: Allows you to target users who recently searched for your competitors’ keywords or specific product terms on Google. Best for: Advertisers seeking massive scale, rapid A/B testing, and those already heavily invested in the Google Ads ecosystem for search and video. Pricing model: Flexible models including CPC, CPM, and tCPA (target cost per acquisition). Pros:  - Unrivaled reach across almost every niche and geography. - Powerful lookalike audience generation using encrypted first-party data through the Customer Match feature. - Deep integration with Google Analytics 4 (GA4) for closed-loop attribution. Cons: - High risk of junk traffic on mobile games/apps if placement exclusions are not strictly managed. - Limited transparency into exact middleman fees compared to independent DSPs. ### Amazon DSP Why it’s essential:  Amazon DSP is the gold standard for e-commerce performance, since it uses actual purchase history, not just browsing intent, to target users. Instead of optimizing for superficial clicks, it leverages verified purchase and cart-addition history, allowing brands to build precise lower-funnel audiences across the web. The platform lets brands activate this first-party shopper data across owned properties like IMDb, Twitch, and Fire TV, alongside premium third-party sites, creating an efficient acquisition pipeline that dynamically re-engages past shoppers, cross-sells products, and ties ad spend directly to final checkout revenue. Showcased features: - Lifestyle and in-market sequencing: Allows marketers to target users based on life events (e.g., “wedding”) or specific browsing habits within the Amazon store. - Amazon Marketing Cloud (AMC): An advanced clean-room environment to analyze cross-channel journeys and custom attribution. - Shoppable creative templates: Ad units that pull real-time pricing, star ratings, and “Prime” badges directly from product detail pages. Best for: D2C brands, consumer packaged goods (CPG) companies, and even non-endemic advertisers, like insurance or auto, who want to leverage high-intent shopper signals. Pricing model: Primarily CPM-based. Self-service has no strict minimum, but managed-service typically requires a $50,000 monthly commitment. Pros: - Unmatched accuracy in targeting users with a proven ready-to-buy mindset. - Access to exclusive, high-impact placements like the Amazon Homepage and Fire TV. Cons: - The interface is notoriously complex and has a steep learning curve for beginners. - Higher effective costs per mille (eCPMs) compared to general networks due to the premium nature of the data. ### Meta Ads (Display and Audience Network) #### Why it’s essential: Meta Ads remains a cornerstone of performance marketing thanks to its sophisticated predictive machine learning algorithms. Even as privacy regulations shift, Meta’s underlying system processes trillions of behavioral signals across Facebook, Instagram, Messenger, and its sprawling Audience Network daily. This allows the platform to move beyond basic demographic matching and instead capture users based on micro-actions and deep behavioral patterns. Meta’s Audience Network extends beyond its native social feeds, too, allowing you to place high-impact native, banner, and interstitial ads into thousands of premium mobile apps and publisher sites. It’s an essential channel for any brand that needs to rapidly scale conversion volume, while maintaining strict control over efficiency. #### Showcased features: - Advantage+ Shopping Campaigns (ASC): A fully automated campaign type that uses machine learning to streamline your targeting, creative asset variations, and budget allocation all at once. - Dynamic Product Ads (DPA): Automatically showcases products from your catalog to users who have previously expressed interest, or viewed those items on your website or app. - Audience Network Rewarded Video: High-engagement video ad placements inside mobile games, where users willingly watch an ad in exchange for in-app items, resulting in near-perfect completion rates. Best for: Direct-to-consumer (D2C) brands, e-commerce stores, mobile app developers, and business-to-consumer (B2C) services looking to rapidly scale their visual storytelling and acquisition efforts. Pricing model: Primarily auction-based CPM or CPC. There are no strict minimum spend requirements for self-service accounts, making it accessible for budgets of any size. Pros: - Unmatched algorithmic optimization that excels at finding buyers out of massive audience pools. - Exceptionally high engagement rates on visual, video-heavy ad formats like Reels and Stories. Cons: - High volatility in ad costs (CPMs) during peak holiday seasons and competitive quarters. - Audience Network traffic can occasionally skew toward accidental clicks in mobile games if placements aren’t rigorously filtered. Microsoft Advertising (Display) #### Why it’s essential: Microsoft Advertising’s Display Network integrates directly with the Windows operating system, Microsoft Edge, MSN, and Outlook, reaching users while they’re in a focused, high-intent mindset. It’s an older audience that often has more disposable income, and is heavily concentrated in professional office environments where Microsoft services are the default daily standard. What really sets Microsoft’s display ecosystem apart is its exclusive access to LinkedIn professional data. Microsoft owns LinkedIn, so you can overlay B2B targeting parameters (like specific job titles, company sizes, and industries) directly onto native display placements across properties like MSN and the Edge browser. This bypasses the typical black-box display problem by allowing you to serve a standard visual banner exclusively to C-suite executives or specific IT decision-makers while they check their daily news. #### Showcased features: - LinkedIn profile targeting: The exclusive ability to target display ads based on a user’s LinkedIn company, job function, and industry across non-LinkedIn web properties. - Predictive targeting: An AI-driven feature that analyzes your existing conversion data to find new, unexpected high-converting audience pockets across the Microsoft network. - Microsoft Audience Ads: Premium native placements that blend seamlessly into personalized content feeds on MSN, Outlook, and Microsoft Edge tab pages. Best for: B2B software companies, financial services, enterprise solutions, and high-ticket consumer brands looking to reach affluent, desktop-heavy users. Pricing model: CPC and CPM models available via real-time bidding. No platform minimums for self-service setups, though competitive enterprise targeting requires a healthy daily budget to clear bidding auctions. #### Pros: - Access to a highly professional corporate audience with strong purchasing power that’s hard to find on social media. - Direct integration of LinkedIn data onto traditional display inventory provides unparalleled B2B accuracy. #### Cons: - Total search and display volume is lower when compared to Google’s massive global footprint. - Traffic skews heavily toward desktop users, which may not align with mobile-first campaign strategies. ### The Trade Desk (TTD) Why it’s essential: As the leading independent DSP, The Trade Desk provides a walled-garden alternative with total transparency. It’s essential for sophisticated performance teams who want to buy on the open web with granular control over every bid factor and data source. The platform’s true power lies in its cookieless identity infrastructure. By pioneering Unified ID 2.0 (UID2), it enables marketers to securely deploy first-party data across premium connected TV (CTV), digital audio, and retail media networks, aligning ad delivery with real-world consumer behavior safely and precisely. Showcased features: - Koa AI: An engine that evaluates more than 13 million impressions per second to automatically reallocate budget to the highest-performing channels in real time. - UID 2.0: A leading identity solution for the cookieless era that allows for precise retargeting without relying on third-party cookies. - Planner tool: Provides data-driven forecasts on reach and cost before a single dollar is spent. Best for: Agencies and enterprise-level performance brands that require deep transparency, omnichannel reach (CTV, audio, display), and custom data integrations. Pricing model: CPM-based bidding with platform fees typically calculated as a percentage of total media spend. Pros: - Complete transparency into the supply path and bidding auctions. - Best-in-class cross-device attribution and frequency capping to prevent ad fatigue. Cons: - Significant minimum spend requirements usually make it inaccessible for small businesses. - Requires a dedicated trader or highly trained staff to manage the platform’s complexities. ### Adobe Advertising Cloud #### Why it’s essential: Adobe Advertising Cloud is an independent, enterprise-grade DSP built specifically for large brands that require absolute transparency, cross-screen orchestration, and ironclad brand safety. Unlike platforms tied to a specific publisher network (like Meta or Amazon), Adobe doesn’t own any media inventory, and this independence means its optimization algorithms have no bias. The primary reason global enterprises rely on Adobe, though, is its native integration with the broader Adobe Experience Cloud ecosystem. If your company already uses Adobe Analytics, Adobe Audience Manager, or Adobe Real-Time CDP, Advertising Cloud bridges the gap between your data and your media buys. You can take highly complex, multi-layered audience segments built from your first-party website data and activate them across connected TV (CTV), digital audio, video, and programmatic display without risking data leakage or sync errors. #### Showcased features: - Unified cross-screen orchestration: Plan, buy, and measure display, video, CTV, audio, and native campaigns simultaneously from a single user interface. - Adobe Analytics integration: Deep, native data connectivity that allows the DSP to optimize media bids based on actual on-site user behavior data, rather than basic pixel clicks. - Automated budget pacing and optimization: Advanced AI that monitors performance across multiple inventory providers and automatically shifts capital to the most efficient channels in real-time. Best for: Enterprise-level brands, large advertising agencies, and companies with complex first-party data structures that require cross-channel campaign management and deep analytics. Pricing model: Typically operates on a technology fee percentage of total ad spend or via a fixed SaaS subscription. Because it’s a premium enterprise tool, it generally requires a substantial minimum annual spend commitment. #### Pros: - Complete programmatic media independence with zero inventory bias or hidden ad-tech taxes. - World-class integration with first-party enterprise data management systems and analytics. #### Cons: - Prohibitively expensive for small-to-medium businesses or brands with modest ad budgets. - Requires a highly experienced, dedicated programmatic media buyer to properly manage the advanced platform layout. ### AdRoll #### Why it’s essential: AdRoll is a performance marketing platform engineered to democratize high-level retargeting and display prospecting for growing e-commerce brands. While massive enterprise DSPs require immense budgets and specialized data engineers, AdRoll simplifies the programmatic ecosystem into a user-friendly hub. It acts as an equalizer, giving smaller D2C businesses the power to launch complex cross-channel campaigns that look and feel like they were built by a larger agency. The engine under AdRoll’s hood connects directly to over 500 ad exchanges, giving users access to roughly 98% of the secure internet. Its primary strength, though, lies in behavioral retargeting, identifying the users who abandoned a shopping cart or viewed a specific product page on your store, then dynamically serving them tailored product variations as they browse the web, check their emails, or scroll through social media. #### Showcased features: - Dynamic Retargeting banners: Ad layouts that pull products directly from your e-commerce store catalog to display the exact items a user left behind in their cart. - Cross-channel integration: The ability to manage web retargeting, social media placements, and automated email trigger sequences, all within a single dashboard. - Consent management integration: Built-in cookie consent tools that ensure your tracking and retargeting efforts automatically comply with global privacy standards like General Data Protection Regulation (GDPR) and California Consumer Privacy Act (CCPA). Best for: Small-to-medium e-commerce businesses, growth-stage D2C brands, and boutique agencies looking for a streamlined, high-ROI retargeting setup without massive minimums. Pricing model: Flexible hybrid model consisting of a monthly SaaS subscription tier paired with standard media CPM/CPC billing. There are no strict minimum spend requirements to launch a campaign. #### Pros: - Fast, easy setup process that integrates natively with platforms like Shopify, WooCommerce, and Magento. - Automated retargeting algorithms that are highly effective at recovering abandoned carts. #### Cons: - Lacks the hyper-granular control over specific programmatic inventory sources found in enterprise DSPs. - Reporting features can lean toward a first-touch/last-touch bias if not manually configured. ### Adsterra Why it’s essential: Adsterra is a high-performance network that bridges the gap between traditional display and affiliate-style efficiency. It’s built specifically for growth marketers in high-action verticals (utilities, VPNs, e-commerce, gaming) where success depends entirely on down-funnel installations and immediate user sign-ups. Adsterra features high-impact ad formats alongside a CPA-based optimization layer for display traffic, eliminating conversion waste by only charging for verified results. Backed by strict anti-fraud tracking, it allows smaller teams to scale aggressive acquisition campaigns globally. Showcased features: - CPA Goal algorithm: An automated optimization tool that unlinks underperforming placements in real time to keep your campaign within a target acquisition cost. - Social Bar: A high-engagement display format that mimics social media notifications, often yielding higher CTR than standard banners. - SmartCPM: An automated bidding feature utilizing second-price auction logic, ensuring you pay only the minimum required to beat the next highest bidder, rather than your maximum cap. Best for: Performance-first advertisers and affiliate marketers who need high volume, global reach (including Tier-3 GEOs), and automated budget protection. Pricing model: Offers CPM, CPC, and CPA models with a low $100 minimum deposit. Pros: - Includes built-in anti-fraud systems to filter out bot traffic. - Fast account approval and 24/7 personal manager support. Cons: - Inventory is generally more mid-tail than the premium editorial sites found on Realize or The Trade Desk. - High-impact formats like pop-unders require careful creative planning to avoid being overly intrusive. ### Simpli.fi #### Why it’s essential: Simpli.fi is a programmatic advertising giant built specifically around the power of unstructured data and hyper-local geographical targeting. While most display networks force you to buy pre-packaged audience categories (like “Auto Enthusiasts”), Simpli.fi lets you bid on individual data points like specific search terms, physical locations, and exact timestamps. The defining characteristic of Simpli.fi is its geo-fencing technology. The platform can map out precise, custom-drawn polygonal boundaries around real-world locations (like a competitor’s storefront, a convention center, or a specific neighborhood) and track mobile devices that enter that zone. Once a device is flagged, the platform can serve highly relevant display ads to that user for days afterward, making it a powerful tool for tying digital advertising spend to physical foot traffic. #### Showcased features: - Custom polygon geo-fencing: Trace exact spatial boundaries around physical addresses using GPS coordinates, to target ads to real-world visitors with pinpoint accuracy. - Geo-conversion lift metrics: Advanced foot traffic attribution that tracks exactly how many people saw your display ad and subsequently walked into your physical business location. - Keyword match display retargeting: Bids on display impressions based on the exact search queries users typed into search bars across the open web, combining search intent with display scale. Best for: Multi-location franchises, localized service providers (like auto dealers or medical centers), regional brands, and event marketers who rely on driving physical foot traffic. Pricing model: Operating primarily on a programmatic CPM structure. It’s highly accessible for local campaigns, though managed-service setups for franchise scaling usually feature specific monthly spending tiers. #### Pros: - The absolute best-in-class geo-fencing and localized boundary-targeting capabilities in the digital landscape. - Unstructured data usage gives you complete control over targeting variables without relying on generic pre-baked audience bundles. #### Cons: - The granular nature of the targeting requires constant optimization to prevent small campaigns from under-pacing. - B2B intent tracking is less robust compared to platforms with native professional network integrations. ## More About Display Ad Platforms ### How to Choose a Display Ad Platform Choosing a display platform is about more than just checking for the lowest CPM. For D2C brands, the priority is transparency and the ability to track a user from that first discovery click all the way to a final checkout. You should evaluate whether a platform offers direct access to publishers to avoid hidden fees, and whether its AI can actually predict conversion intent, rather than just hunting for accidental clicks on a flashlight app. ### Examples of Different Display Ad Formats Modern display goes far beyond the static 300x250 rectangle. High-performance formats now include motion ads that use subtle animation to catch the eye, and responsive display ads that automatically adjust their size and format to fit the available space on a publisher’s site. Native display formats are also essential for performance brands, as they blend into the editorial content of premium websites, making the ad feel like a natural part of the user’s reading journey. ### Self-serve vs. Managed Service Display Advertising Self-serve platforms give your internal team full control over every bid and creative tweak, which is ideal if you have the bandwidth to manage daily optimizations. However, managed services or highly automated platforms handle the heavy lifting of bidding and creative generation through AI. This allows smaller teams to achieve enterprise-level scale and efficiency without needing to hire a dedicated army of data scientists. ## Key Takeaways These platforms primarily focus on performance rather than just raw impression volume. The right platform for your brand depends on the results you’re looking to achieve, but in all cases, you should be looking for platforms that offer transparency, easy optimization, and access to relevant inventory. ## Frequently Asked Questions (FAQs) ### How should advertisers measure performance on display campaigns? Measuring success generally starts with identifying your primary KPI, whether that’s return on ad spend (ROAS), cost per acquisition (CPA), or lead volume. While traditional metrics like click-through rate (CTR) give you a sense of engagement, performance campaigns on the open web require a deeper look at post-click behavior and conversion probability. Along with that, advanced platforms now use predictive modeling to show your ads to users who are mathematically more likely to convert, meaning you should judge performance based on actual business outcomes rather than top-of-funnel traffic. ### Are display ads still effective for performance campaigns? Absolutely, provided you move away from passive, low-quality inventory and focus on high-intent environments. On the open web, display ads act as a powerful discovery engine that reaches users while they’re actively consuming information, often leading to higher quality conversions than distracted users found on social feeds. Effective performance display ads now rely on dynamic creative optimization and AI-driven bidding to ensure your message is relevant to the user’s current context. ### What’s the difference between a display ad network and a programmatic DSP? A traditional display ad network typically acts as a broker between a group of publishers and advertisers, offering ease of use but sometimes lacking deep technical control. A programmatic demand side platform allows for more complex, automated bidding across a vast array of global exchanges. High-performance platforms offer the massive scale and automation of programmatic bidding, while providing the direct-to-publisher transparency and ad-tech tax avoidance usually only found in direct integrations. --- ### Interest-Based Targeting: How To Drive Engagement on the Open Web URL: https://www.taboola.com/marketing-hub/interest-based-targeting/ Last Modified: 2026-06-24 12:24:40 The era of one-size-fits-all advertising is over, and consumers are the ones who ended it. Today’s audiences expect ads to reflect their interests, habits, and intent, especially in environments where focused attention is limited. Interest-based targeting addresses that challenge by helping you reach people based on what they actively engage with online. Rather than relying on basic demographic assumptions, this approach uses behavioral and content signals to determine relevance. When used correctly, interest-based targeting leads to higher engagement, more efficient spend, and stronger long-term connections between you and your audiences. ## What Is Interest-Based Targeting? Definition and Core Concept Interest-based targeting (IBT) is an advertising approach that delivers content, offers, or messages to users based on their demonstrated interests. That data comes from what people actually do, not what they say they like. Customer interest signals can include: - Browsing behavior. - Content engagement. - App usage. - Search activity. Rather than relying on fixed attributes like age, gender, or location, IBT focuses on long-term patterns. A user who repeatedly engages with fitness content, for example, may be classified into interest categories related to wellness, nutrition, or athletic gear, regardless of demographics. At its core, IBT is about relevance. It enables advertisers to reach people when their mindset aligns with the message, increasing the likelihood of engagement without requiring personal details. ## Why Interest-Based Targeting Matters: Key Benefits for Engagement and ROI People engage with ads that feel like a natural extension of what they’re already interested in. Interest-based targeting makes that possible, improving engagement in a few key ways. ### Better Relevance and Personalization Relevance improves dramatically when advertising reflects both what users consume and what they consistently engage with over time. A user reading about home renovation projects will likely be more receptive to an ad for power tools or flooring materials than to unrelated promotions. Interest-based targeting helps place ads in these moments of intent. Because this personalization is built on aggregated behavior rather than individual profiles, it supports engagement without crossing privacy boundaries or relying on sensitive data. ### Higher Conversion Rates and More Efficient Spending When ads are served to users who are inclined toward that category or topic, conversion paths shorten. Users are less likely to ignore or dismiss messaging, and you benefit from reduced wasted impressions. Interest-based targeting also supports more efficient media spend. It focuses your budgets on users with high-intent probability rather than broad audiences with weak relevance signals. ### Opportunity for Retargeting and Long-Term Engagement Retargeting works best when it reflects genuine interest. IBT strengthens retargeting by grounding follow-up messaging in real engagement signals rather than generic exposure. As a result, you can support longer customer journeys with fewer wasted impressions and more consistent relevance. ### Enhanced Brand Affinity Relevant targeting does more than drive clicks, it shapes perception. When users consistently encounter content that aligns with their interests, they’re more likely to view your brand as helpful, credible, and aligned with their needs. Over time, this relevance builds familiarity and trust, making future engagement more likely even outside of paid campaigns. ## Interest-Based Targeting vs. Other Targeting Methods Interest-based targeting shares traits with other approaches, but it differs in how audiences are defined and how relevance is determined. - Contextual targeting matches ads to the topic of a page or piece of content, rather than to the user. It delivers strong in-the-moment relevance, but does not capture interest patterns that develop over time. - Behavioral targeting focuses on specific user actions such as clicks, site visits, or purchases. It responds to individual events, while interest-based targeting groups behavior over time to reflect broader patterns of interest. - Demographic targeting relies on static attributes like age, gender, or location. Although useful in some cases, it often depends on assumptions that don’t reflect changing user intent. In practice, interest-based targeting performs best when used alongside contextual and behavioral inputs, creating a more complete and flexible understanding of user intent. ## How Interest-Based Targeting Drives Engagement: The Mechanics Behind the Strategy The effectiveness of interest-based targeting comes from how its components work together to align messaging, timing, and audience relevance. ### Audience Segmentation and Profiling Interest-based strategies start with segmentation. Platforms analyze engagement data and content interactions to group users into interest clusters. These clusters evolve as behaviors change, ensuring that targeting remains current rather than static. Profiling at this level minimizes individual identification. Instead, it focuses on probability modeling, enabling scalable reach without sacrificing relevance. ### Tailored Ad Messaging and Creative Alignment Once segments are defined, your platform of choice then matches your creative assets to the mindset of each audience group. Messaging, visuals, and offers are aligned with what users are already exploring, increasing resonance. In addition to improving engagement, this matching process also reduces creative fatigue, since users are less likely to perceive your ads as unrelated distractions. ### Retargeting and Sequential Engagement No matter how engaging your messaging is, very few users will convert at first glance. Interest-based targeting makes it possible to recognize early interest and continue the conversation with follow-up messaging. That messaging can be timed to match where each user is in the decision-making process. By sequencing ads based on past interactions, brands stay relevant for longer consideration cycles, instead of relying on isolated impressions. ### Efficiency for Advertisers Broad targeting can be inefficient. IBT helps you reduce those inefficiencies by concentrating delivery on interest-aligned users. Campaigns benefit from improved learning signals, helping optimization engines funnel users toward high-performing segments. This improved efficiency is especially useful when you need your targeting to remain accurate as your campaigns scale. ## Challenges, Limitations, and Risks of Interest-Based Targeting Interest-based targeting brings plenty of benefits, but it also introduces some issues for you to navigate. By understanding these challenges, you can set realistic expectations and build more resilient campaigns. ### Over-segmentation and limited scale When interest segments are defined too narrowly, reach can shrink quickly. This can lead to unstable performance, limited deliverability, and slower optimization, due to insufficient volume. ### Shifting or outdated interest signals User interests aren’t static. They can change based on seasonality, life events, or broader trends, making it essential to regularly refresh segments. ### Interest doesn’t always mean intent Showing interest in a topic doesn’t automatically signal a readiness to take action. Regular optimization can help you ensure that your campaigns remain aligned with how audiences actually respond. ### Creative fatigue within narrow interest groups Highly targeted segments may see the same message repeatedly. Without varying creative, you could find engagement declining even with the most accurate targeting. ### Privacy and compliance constraints Platforms have to balance personalization with user consent, data protection standards, and always-evolving regulations. These requirements may limit available interest data, affecting how your platform can execute targeting. ### Signal overlap with other targeting methods Interest-based, behavioral, and contextual strategies often use similar signals. Unless you adjust for it, you may find you’re duplicating your reach or getting confusing performance metrics. These challenges don’t diminish the value of interest-based targeting. Instead, they show the importance of good segmentation, regular refinement, and strong creative strategy to ensure sustained performance and responsible use. ## When and How to Use Interest-Based Targeting: Use Cases and Strategy Fit Interest-based targeting is particularly effective for mid- to upper-funnel campaigns where relevance and engagement matter more than immediate conversion. Common use cases include: - Content promotion. - Product discovery. - App installs. - Brand storytelling. - Re-engagement efforts. Strategically, interest-based targeting works best when paired with clear messaging goals and flexible creative assets that can adapt to different interest profiles. ## Implementation Steps for Interest-Based Targeting Executing an interest-based strategy requires thoughtful setup and ongoing oversight. ### Audience Research Start by identifying which interests naturally align with your product or service. Look beyond surface-level assumptions and analyze engagement data, content performance, and customer interactions to understand what users actively explore and respond to. This research step helps you target based on actual behavior, rather than inferred preferences. ### Segment Definition Once interests are identified, translate them into segments that balance relevance with reach. Segments should be specific enough to reflect meaningful interest patterns, but broad enough to support delivery, learning, and optimization. Defining segments with this balance in mind reduces performance volatility while allowing platforms to optimize efficiently. ### Campaign Setup and Matching Campaign structure should reinforce the logic behind your interest segments. Align each campaign with the most relevant audience group and make sure all creative assets match the mindset, intent, or stage of engagement associated with that interest. Clear alignment between targeting and messaging improves engagement quality and makes performance results easier to interpret. ### Monitoring and Optimization Interest-based targeting works best when it’s flexible. Monitor performance regularly to identify shifts in engagement, efficiency, and audience response. As interests evolve and campaign goals change, adjust segments and creative to maintain relevance and support sustained performance. ## How to Use Interest-Based Targeting with Realize Interest-based targeting only works if you can act on those signals at scale. Realize performance advertising platform, lets you reach audiences by interest across a network of premium publisher sites — and it gives you more than one way to do it. Depending on whether you want to target people by their long-term interests, the content they're consuming right now, or their high-intent signals, you can choose the approach that fits your campaign goal, or combine several. ### Reach Audiences with Taboola First-Party Interest Segments Taboola First-Party (1P) Audiences are interest and intent segments built with Taboola's own machine learning models, drawing on readership signals from more than 9,000 publisher sites (including Yahoo) and transactional data from Connexity, Taboola's retail and e-commerce network. You get more than 500 ready-made segments across more than 200 categories, spanning verticals like automotive, banking and finance, beauty and fashion, consumer packaged goods (CPG), health and wellness, home and garden, media and entertainment, sports and fitness, technology and electronics, and travel and outdoors. Plus, you’ll have access to demographics such as age, gender, language, education, and career. You'll find these segments listed as Taboola 1P Audiences in the Marketplace Audiences section of campaign setup. Through Purchase Intent targeting, Realize provides high-intent targeting bundles like Search Keyword and Mail Domain targeting. For best results, build separate campaigns for each audience profile with creative tailored to that segment. If you narrow your reach by overlaying segments with and targeting, keep it to five segments or fewer — stacking more than that tends to choke campaign scale. ### Match Ads to Content with Topics and Contextual Targeting If you'd rather reach people based on what they're reading in the moment than on their historical profile, Realize offers content-based targeting that doesn't rely on user cookies. Topics Targeting lets you target granular topics closely aligned to your products or services, while Contextual Targeting places your ads in environments that are highly relevant to your vertical. Both are privacy-conscious by design and pair naturally with the user-based segments above for fuller coverage of intent. ### Capture High Intent with Search Keyword and Mail Domain Bundles For audiences closer to the point of action, Realize provides pre-built Search Keyword Targeting Bundles grouped by verticals like automotive, finance, and education, reaching users based on the specific search terms and attributes they engage with across our network. The Mail Domain Targeting Bundles let you target people based on the domains they interact with (such as users receiving specific brand or category newsletters) — offering pre-built vertical bundles like automotive or finance directly in the platform. These signals get you in front of users who are already showing purchase-oriented behavior, not just topical curiosity. ### Combine Your Own Data with Pixel, CRM, and Marketplace Audiences Realize also lets you bring your first-party data into the mix. Using the Taboola Pixel, you can retarget website visitors and build lookalike audiences from people who already engaged with you. You can upload CRM customer files to target known audiences or suppress existing customers, and through Marketplace Audiences you can layer in third-party DMP (data management platform) and CDP (customer data platform) segments — mixing and matching Taboola 1P data with external data to refine who you reach. ## When Interest-Based Targeting Might Not Be the Best Choice As valuable as interest-based targeting can be, it isn’t ideal in every scenario. Highly time-sensitive campaigns or compliance-restricted industries may be better off with contextual or keyword-based approaches. Similarly, products with extremely narrow audiences may benefit more from direct retargeting or first-party data strategies. Understanding your campaign goals and constraints can help you determine whether interest-based targeting is the right approach. In some instances, you may even decide to use it as a secondary tactic, pairing it with another type of targeting. ## Key Takeaways Interest-based targeting connects advertisers with users based on what they care about, not who they are assumed to be. By focusing on relevance, IBT drives stronger engagement, improves efficiency, and supports long-term brand relationships. When handled correctly, it becomes a flexible and privacy-conscious foundation for modern digital advertising. ## Frequently Asked Questions (FAQs) ### How specific should interest segments be when building an IBT strategy? Interest-based segments should strike a balance between clarity and reach. Segments that are too broad may dilute relevance, while overly narrow segments can limit delivery and learning. The goal is to capture actionable patterns without compromising delivery. ### Do I need a lot of first-party data to make interest-based targeting effective? First-party data improves accuracy, but it isn’t mandatory for effective IBT. Platforms can infer interests using aggregated behavioral and contextual signals, enabling advertisers with limited datasets to still perform well. ### How often should I refresh or refine my interest segments? Interest segments should be reviewed regularly, especially as user behavior and market trends evolve. Refreshing segments helps maintain relevance and prevents performance drift. --- ### Campaign Troubleshooting: A 10-Step Framework for Open Web Performance Campaigns URL: https://www.taboola.com/marketing-hub/campaign-troubleshooting/ Last Modified: 2026-06-30 07:56:08 When an open web campaign stalls or underdelivers, determining the root cause can be the biggest challenge. For performance advertisers moving beyond the walled gardens of search and social, the problem is no longer simply resolving errors, it’s also in engineering the right balance between human strategy and machine intelligence. This campaign troubleshooting guide explores the modern diagnostic landscape and leading industry methodologies with a 10-step framework for leveraging AI-driven insights. ## What Is Campaign Troubleshooting? Defining the Diagnostic Process Campaign troubleshooting is the systematic process of identifying, analyzing, and resolving issues that prevent a digital campaign from hitting its KPIs. For open web advertisers specifically, this spans a far wider area than on closed platforms — campaign troubleshooting here touches all aspects of publisher diversity, algorithmic behavior, tracking infrastructure, and creative relevance all at once. A critical distinction separates broken campaigns from underperforming ones. Broken campaigns suffer from technical failures like misfiring pixels, inactive ads, or rejected creatives that result in zero delivery. Underperforming campaigns are running, but failing to optimize. They fall in the gap between potential and reality that often comes from budget constraints, audience misalignment, poor creative, or misconfigured bidding strategies. Both of these categories require a different diagnostic lens, and viewing them as the same is one of the biggest errors that performance teams can make. Effective troubleshooting on the open web also requires an understanding of the open web optimization environment itself. That involves millions of publisher pages, a wide range of audiences, and real-time bidding ecosystems where dozens of variables interact simultaneously. This is why AI has become indispensable within the performance marketing space. ## Troubleshooting Methodologies: The Algorithmic Approach vs. The Auditor There are three dominant performance advertising methods for campaign diagnosis: - AI-first (The Algorithmic Approach): This is the most modern methodology, where machine learning signals can flag anomalies in campaigns. Unusual dropoffs in delivery, conversion rate spikes, or pacing irregularities can all be identified before they become recurring issues. AI-driven troubleshooting enables proactive intervention rather than reactive adjustments. - Top down: Begin with high-level funnel metrics like impressions, CTR, and CVR, and work downward to identify where the dropoff occurs. This method quickly surfaces the stage of failure but can miss technical root causes. - Bottom up (The Auditor): Start with the technical foundations — check tags, pixels, and ad status before moving upward to bidding, audience, and creative. This is the most thorough but time-intensive process. The most effective teams combine all three, using AI to triage first, then applying human judgement via top-down and bottom-up audits to validate and correct. ## Essential Diagnostic Tools for Your Performance Campaign Every performance advertiser needs a curated set of media buying tools organized across four functional categories: - Verification tools: Ad integrity and placement verification platforms like Integral Ad Science (IAS) and DoubleVerify confirm that ads are appearing in brand-safe, viewable environments and are not being served to fraudulent traffic. - Signal checkers: Pixel helpers and API debuggers like Google Tag Manager’s Preview Mode confirm that tracking events are firing correctly and that conversion data is flowing into the algorithm. - Creative analyzers: Heatmaps, attention metrics, and CTR predictors identify which creative assets are resonating and which are creating drag on campaign performance. - AI-powered analytics dashboards: Platforms with native predictive modeling and anomaly detection capabilities allow advertisers to see ahead of performance trends, rather than simply reacting to them. ## The 10-Step Framework for Open Web Troubleshooting When a campaign underdelivers or fails to convert, this campaign diagnostic checklist provides a logical, sequenced path to resolution. ### Step 1: Verify Core Setup, Budgets, and Scheduling Before any sophisticated analysis, confirm the basics are in place. Is the campaign active? Is the budget fully funded? Are campaign flight dates correct? Are daily caps set low enough that the campaign doesn’t end by mid-morning? Human configuration errors are the most common cause of zero delivery and they’re usually easily fixable. ### Step 2: Respect the AI Learning Process Modern programmatic platforms rely on machine learning to optimize delivery. When you launch a campaign or make significant changes, the algorithm enters the algorithmic learning phase, typically a 48- to 72-hour window, during which it collects performance signals to calibrate bidding, audience targeting, and placement decisions. Intervening with manual changes during this window resets the learning cycle, leading to prolonged instability. Monitor this learning status and resist the urge to edit. Budget fluctuations and suboptimal CPAs during this phase are to be expected. ### Step 3: Validate Pixel Tracking and Data Loops Addressing ad tracking issues is foundational to campaign health. If conversion signals aren’t reaching the platform, the algorithm is navigating blind. Use pixel debugging tools to confirm that view and click events are firing, that the correct conversion actions are attributed, and that deduplication logic isn’t suppressing valid signals. Conversion signal optimization requires a clean, high-fidelity data loop to give the algorithm the feedback it needs to find more converting users. ### Step 4: Audit Audience Reach and Scaling Constraints Hyper-segmented targeting is one of the most common causes of delivery issues. When audience parameters are too narrow, the eligible inventory pool shrinks to a point where the algorithm cannot spend efficiently. Audit your targeting layers and, where possible, allow AI-driven audience expansion to surface lookalike inventory across the open web. Scale is often found by loosening, not tightening, your initial audience constraints. ### Step 5: Inspect Inventory and Placement Quality The open web’s strength is in its vast publisher diversity, but it can also be a source of performance inconsistency. Review placement-level reports to identify if certain publishers or placements are consuming budget without contributing conversions. Equally, investigate whether your bids are competitive enough to win impressions on premium, high-quality inventory. A bid that is technically correct, but practically uncompetitive, will result in consistent underdelivery even on the best platforms. ### Step 6: Combat Creative Fatigue via Dynamic Optimization When CTR declines without a corresponding change in audience or bids, creative fatigue is typically the culprit. Users have seen the same ad too many times and engagement has eroded. This is where dynamic creative troubleshooting becomes essential. Dynamic Creative Optimization (DCO) uses AI to automatically test and rotate headlines, images, and calls to action, creating winning combinations without requiring manual intervention. Maintain a rotating library of creative assets and set frequency thresholds that trigger automatic refresh. ### Step 7: Resolve Ad Rejections and Policy Compliance Open web publishers operate under varying editorial and content standards, meaning that one ad that was approved on one platform could be rejected on another. Regularly audit rejection notifications in your dashboard and categorize them by reason, such as image policy, landing page compliance, or prohibited categories. Many automated rejections can be appealed with minor asset revisions. Build a compliance review step into your campaign launch workflow to prevent rejections from silently lowering delivery. ### Step 8: Balance Prospecting and Retargeting Flows A frequently overlooked issue is funnel drying, where retargeting pools shrink because insufficient new users are entering at the top of the funnel. If your retargeting audience is too small, those campaigns will underdeliver regardless of budget. Use AI to continuously identify new top-of-funnel prospects across the open web and feed new, qualified users into the awareness stage so that mid- and low-funnel lists remain scalable. ### Step 9: Optimize Landing Page Performance and UX Click-through rates only tell some of the story. If users are clicking but not converting, the problem may lie in the post-click experience. Use AI-powered analytics to assess landing page load speeds, bounce rates segmented by device and browser, form abandonment patterns, and message-match alignment between ad creative and landing page content. A three-second load delay can reduce mobile conversion by as much as 20%. It’s vital to check these areas to see where your campaigns could be losing vital conversions. ### Step 10: Leverage Predictive Analytics for Proactive Health The most advanced stage of troubleshooting is eliminating the need for it by anticipating potential issues before they occur. By analyzing historical campaign data alongside AI forecasting models, advertisers can identify early warning signs like declining impression share, rising CPCs, or lowering conversion rates, before they turn into missed KPIs. This moves the troubleshooting function from reactive repair to proactive optimization, allowing teams to pre-adjust bids, refresh creative, or expand audiences before performance materially degrades. More troubleshooting recommendations based on Realize performance campaigns could be found here. ## Key Takeaways The evolution of campaign troubleshooting mirrors the broader evolution of digital advertising from manual, reactive processes to intelligent, proactive systems. Effective diagnosis on the open web demands more than fixing what’s visibly broken. Instead, it requires the ability to distinguish between technical failures and optimization gaps, and leaning into where the human-AI partnership can address anomalies and foundational issues. Teams that continually address this type of troubleshooting will outperform those who view it as an afterthought, making this a competitive advantage in open web performance campaigns. ## Frequently Asked Questions (FAQs) ### How does troubleshooting on the open web differ from search and social? The open web involves a greater number of publishers, ad formats, and audience contexts. Search centers on intent signals, whereas social focuses on interest and behavioral targeting within a closed ecosystem. Open web troubleshooting relies heavily on AI-driven behavioral modeling to identify the right users where no single platform is in control. This makes data quality, pixel integrity, and algorithmic trust more critical here than on other channels. ### What are the most common tools used for campaign troubleshooting? Standard tools include pixel and tag debuggers (such as GTM Preview Mode and browser pixel helper extensions), ad verification platforms (like IAS or DoubleVerify for brand safety and viewability), and internal platform analytics that use AI to flag anomalies in pacing, delivery, or conversion rates. For advanced teams, predictive dashboards that model future campaign trajectories based on historical patterns have become an essential part of the troubleshooting process. ### Can AI automatically fix my campaign issues? AI can automate a meaningful range of optimizations, such as reallocating budget toward high-performing creatives, adjusting bids in real time, expanding to lookalike audiences, and rotating ad assets to counter fatigue. However, AI can’t resolve foundational setup errors, such as a broken landing page, an inactive tracking pixel, a misconfigured campaign flight date, or a rejected creative. These require human diagnosis and intervention. The most effective approach treats AI as a powerful partner, where it handles scale and pattern recognition, while human oversight addresses the structural foundations that the algorithm depends upon. --- ### Why Ad Placement Optimization Is the Secret Weapon for Performance Advertisers URL: https://www.taboola.com/marketing-hub/ad-placement-optimization/ Last Modified: 2026-06-24 08:18:16 For performance-driven marketers, the open web is often seen as the last frontier — a massive, sprawling, endless landscape of potential profit that remains notoriously difficult to navigate without the right map. We’ve now officially pivoted into the agentic era, a shift where the old-school grind of manual A/B testing and basic forecasting is being sidelined by autonomous AI agents. These systems don’t just suggest optimizations, they execute them in real-time. It’s an absolute game-changer, so let’s get into how these next-gen tools are revolutionizing ad placement optimization, helping you secure high-viewability spots, respect the user’s journey, and scale your ROAS with precision. ## What Is Ad Placement Optimization? At its core, ad placement optimization is the data-driven process of determining the exact location, format, and context where an ad will get the highest engagement and conversions. It’s not just about being “on the page” anymore. Instead, it’s more about being in the right spot on the right page for the right person. This involves analyzing thousands of variables — from the device the user is holding to the specific paragraph they’re currently reading — to ensure that when your ad appears, it feels like a natural part of the discovery journey, rather than an annoying interruption. ## Why the Open Web Demands a New Approach for Performance Advertisers While it’s true that walled gardens like Meta and Google Search offer a closed, controlled environment, the open web is where the real scale lives. However, that also means lots of unique hurdles and hoops to jump through. With the phase-out of third-party cookies, the old strategy of following the user everywhere is essentially dead. Performance advertisers now have to be smarter, and rely on real-time programmatic efficiency and contextual relevance to generate revenue. On the open web, you aren’t just buying a user — you’re buying a moment of attention. If you can’t optimize your placement for that specific moment, you’re basically just donating money to the internet. ## Advertising in the Agentic Era We’ve moved past simple automation, and are now in a world of autonomous systems that can manage complex workflows without a constant human babysitter. “Agentic” is a word you may have seen being used more frequently these days in the context of digital marketing, so let’s break down what it really means. ### Predictive AI vs. Agentic AI Think of Predictive AI as a GPS: it tells you where the traffic is and suggests a better route, but you’re still in charge of turning the steering wheel and keeping an eye on traffic. Agentic AI would be the self-driving car. While traditional AI analyzes data and waits for a human to adjust the bids, agentic AI autonomously researches, tests, and executes optimal placements in real-time. It doesn’t just suggest a better placement, it goes out and buys it for you while you’re asleep. ### Autonomous Execution and Real-Time Bidding In the agentic era, Real-Time Bidding (RTB) happens at a scale humans can’t even comprehend. AI agents evaluate millions of ad opportunities in seconds. They check out the page, the history of the placement, and the current cost, dynamically shifting your spend toward high-performing, contextually relevant open web placements. The major benefit here is that it eliminates media waste by ensuring you never overpay for a placement that has zero chance of converting. ## Core Strategies for Open Web Ad Placement Securing high-value inventory is only half the battle. You also have to make sure the ad is actually seen and processed by a human brain. ### Prioritizing Viewability and Attention Metrics A “rendered impression” is a vanity metric if that impression happened three scrolls below where the user stopped reading. Performance advertisers are moving toward viewability metrics and attention scores. AI tools now track scroll depth and time on screen to ensure your ad placement optimization strategy favors spots where eyes actually linger. If your ad isn’t at least 50% in view for at least one second, it shouldn’t count toward your budget. ### Strategic Placements: Above the Fold vs. Below the Fold This age-old debate started with physical newspapers and still continues today in digital form. So, what’s better, Above the Fold (ATF) or Below the Fold (BTF)? ATF ads offer immediate, guaranteed visibility, but they can also be the first thing a user ignores. Strategic BTF placements, when inserted at natural reading breaks where a user pauses to digest information, often see much higher engagement. The key is using AI to find those natural breaks so your ad feels like the next logical step in the reader’s curiosity. ### Contextual Relevance in a Cookieless World In a world without cookies, context is king. Modern AI tools use contextual targeting to analyze the text, video, and even audio of a web page. If someone is reading an article about the best hiking boots, your ad for waterproof socks should appear right next to the durability section, or something similar. This hyper-relevancy is what turns a passive reader into a high-intent lead. ## Balancing User Experience (UX) and Ad Revenue More ads doesn’t always mean more money. If you clutter a page so badly that it takes 10 seconds to load, your bounce rate will skyrocket, and your ROAS will tank. ### Lazy Loading and Page Speed Optimization This is where technical ad placement optimization meets UX. By implementing lazy loading — a technique that delays the loading of non-critical resources such as images, videos, JavaScript, or CSS — BTF ads only render when a user scrolls toward them. This keeps the initial page load lightning-fast, ensuring high user retention while still getting your ad in front of the people who are actually engaged enough to keep reading. ### Ad Density and Frequency Capping Nobody likes being followed by the same ad 10 times in an hour. Agentic AI can automatically apply strict frequency capping, ensuring you aren’t annoying your future customers. It strikes the perfect balance between content and monetization, ensuring your ad density remains high enough to be profitable, but low enough to remain brand-safe. ## Selecting the Right AI-Driven Tools for Placement Optimization When choosing your tech stack, look for Demand-Side Platforms (DSPs) that support autonomous workflows. You want tools that offer Dynamic Creative Optimization (DCO) — which adjusts the ad’s look and feel to match the placement — and automated A/B testing platforms that use “multi-armed bandit” type decision making algorithms to favor winning combinations instantly. This is exactly where Realize shines. As a performance engine built for the open web, Realize uses deep learning to handle the heavy lifting of ad placement optimization, placing your creative in front of audiences who are already in a discovery mindset across a direct-to-publisher network of 11,000+ premium sites. By bypassing the usual ad-tech taxes of complex programmatic exchanges, Realize ensures your budget goes directly into high-intent placements that move the needle. ## Measuring Success: KPIs for the Agentic Marketer When AI is at the wheel, you have to look beyond the click-through rate (CTR). You should be obsessing over return on ad spend (ROAS) and cost per acquisition (CPA) instead. Because agentic workflows can optimize for these hard numbers autonomously, your role shifts from the person at the controls to more of a strategist, monitoring the overall health of the funnel while the AI handles the millisecond-by-millisecond bidding wars. ## Key Takeaways The transition from manual guesswork to autonomous, AI-driven strategies is the only way to win on the modern open web. By embracing ad placement optimization, focusing on viewability, and respecting the user’s experience, you can drastically reduce wasted spend and see a massive improvement in ROI. The agentic era isn’t coming. It's here. ## Frequently Asked Questions (FAQs) ### What is the difference between predictive AI and agentic AI in advertising? Predictive AI is like a weather forecast — it analyzes historical data to suggest where an ad might perform best, but it still requires a human to pack the umbrella and execute the change. Agentic AI is truly autonomous. It doesn’t just predict, it acts instantly, analyzing live campaign data and independently executing strategic decisions, such as reallocating your budget to a better-performing placement in real-time without needing an approval button. In the world of ad placement optimization, this means the system is constantly self-correcting your campaigns to ensure your ROAS never dips. ### How does AI improve ad placement optimization? AI processes millions of contextual signals that a human could never track, matching your audience with the most relevant inventory available. It automates the tedious grunt work of A/B testing, adjusts real-time bids based on the likelihood of a conversion, and utilizes dynamic creative optimization to make sure the ad visually fits the environment. On a higher level, ad placement optimization involves using these AI signals to avoid made-for-advertising (MFA) sites and low-quality inventory, ensuring your performance campaigns only appear in high-trust editorial environments where users are actually paying attention. ### What are the best ad placements for generating revenue on the open web? The best placement is the one that balances visibility with user intent. While ATF ads offer immediate eyeballs, they are often skimmed over. BTF ads — especially those placed within the flow of an article using lazy loading — often capture the most engaged users who are in a deep discovery state. High-level ad placement optimization strategy involves a mix: using ATF for reach and frequency, while utilizing BTF native placements to drive the actual high-intent clicks that lead to sales and sign-ups. ### How do I improve ad viewability without hurting UX? Shoot for integration, not interruption. Avoid pop-ups or intrusive formats that hide the content the user actually came to see. Instead, use native ad formats that mirror the layout of the site. Use lazy loading so ads only render when they are about to be seen, preserving page speed. Finally, rely on AI-driven frequency capping. If a user hasn’t clicked after three views, the AI should automatically move your budget to a fresh prospect, preventing both ad fatigue and a cluttered user interface. --- ### Performance Advertising Industry News Updates: June 16, 2026 URL: https://www.taboola.com/marketing-hub/performance-advertising-industry-news-updates/ Last Modified: 2026-06-23 08:58:43 The advertising landscape moves fast — and for performance marketers, staying on top of platform shifts, measurement innovations, and new buying opportunities can make the difference between hitting targets and missing them. Our team of industry experts has curated the week's most impactful updates to keep you informed on what matters for your campaigns. ## 6 of the Most Important Performance Advertising Industry News Updates (And Our Expert's Take on Them) ### Update 1: DSP Competition Intensifies — Here's How Buyers Are Ranking Them Performance advertisers now have more choice than ever, but not all demand-side platforms (DSPs) are created equal. A Digiday Scorecard based on rankings from 13 media buyers shows Yahoo DSP and The Trade Desk leading the pack with aggregate scores of 7.3 and 7.2 respectively, followed by DV360 at 7.0, with Amazon DSP trailing at 6.1. The Trade Desk scored highest on transparency and inventory quality but came up weakest on pricing — a critical tension for performance teams managing tight cost-per-mille (CPM) budgets. Meanwhile, DV360 remains "effectively mandatory for YouTube access," creating a lock-in dynamic that performance buyers have to navigate strategically. Omri Bitan, head of strategy: "This research shows us there are really no clear winners in the DSP landscape — each one has some high points and some glaring disadvantages, without a real ‘go-to.’ — buyers are looking for channels that can offer the advantages they seek, like unique supply and access to data — and that’s especially true in the age of AI." ### Update 2: OpenAI Adds Conversion Tracking for ChatGPT Ads — Closing the Measurement Gap Since ChatGPT launched ads earlier in 2026, performance advertisers have faced a critical question: Do these ads actually drive conversions? To answer that question, OpenAI partnered with LiveRamp to add conversion measurement through LiveRamp's Conversions API (CAPI) Hub, which now allows advertisers to track in-store purchases with no minimum spend required. Online conversion tracking is planned for later this year. This brings ChatGPT toward measurement parity with Google, Meta, TikTok, and Pinterest — and removes a major barrier to budget allocation. OpenAI is anticipating $2.5 billion in ad revenue for 2026, with a target of $100 billion by 2030. Omri Bitan, head of strategy: "Conversion tracking is table stakes for performance platforms, and closed platforms adding third-party measurement APIs shows the market is moving toward more transparent attribution. The real insight here is that advertisers need flexibility in measurement — not dependence on a single platform's opaque conversion model. On the open web, through Realize, you maintain first-party control over your conversion data and audience insights. This means better data portability, cleaner attribution logic, and the ability to leverage your measurement across multiple channels without being locked into proprietary algorithms." ### Update 3: Amazon DSP Adds Pre-Bid Attention Targeting — Filtering Low-Quality Inventory Quality and efficiency go hand-in-hand for performance advertisers. Amazon DSP now integrates Adelaide's AU attention metric for pre-bid targeting, letting buyers filter toward higher-attention placements and away from low-attention inventory. Amazon is also introducing an exclusive AU Quality Floor that excludes the bottom 10% of inventory and made-for-advertising (MFA) sites identified by Jounce Media. The AU metric is already available in Trade Desk, Viant, Yahoo DSP, Adobe Advertising, and Equativ — but Amazon's adoption signals a broader industry shift toward attention as a targeting dimension alongside audience and context. Omri Bitan, head of strategy: "The industry's push to filter out low-quality and made-for-advertising inventory reflects a fundamental truth: supply quality is the foundation of performance. What many advertisers don't realize is that this problem is largely solved by building on the right network from the start. The Realize advantage is that our network is built directly with the world's leading premium publishers — 9,000-plus direct integrations with high-quality editorial environments. This means you're not buying from junk inventory pools and then filtering them out; you're starting with premium supply where brand safety and performance align naturally. That's why many performance advertisers choose Realize — the network design itself prevents the MFA and low-quality inventory problems others are scrambling to solve." ### Update 4: TikTok Repositions as Full-Funnel Performance Platform — Not Just Brand Awareness For years, TikTok was positioned as a brand-awareness channel. That narrative is shifting. TikTok is now actively pitching itself as a full-funnel performance platform, citing a 40% year-over-year increase in daily searches and a collapsed funnel where discovery and conversion happen in a single session. The platform is launching TikTok Funnel HQ and promoting Smart+ and GMV Max automated buying tools. According to TikTok's global head of business marketing, the biggest misconception remains that "TikTok is just a brand platform and not a performance platform." Omri Bitan, head of strategy: "TikTok's repositioning to full-funnel performance is real, but it also highlights a critical blind spot for most performance advertisers: over-concentration on social platforms. While TikTok, Meta, and Google fight for feed dominance, you can diversify your funnel across premium publisher inventory without social algorithm dependency. The collapsed funnel opportunity TikTok is promoting? You can achieve it through contextual targeting and audience strategies that work across a 9,000-plus publisher network." ### Update 5: Agentic Commerce Reshapes Google Ads — New Performance Surface for E-commerce The rise of AI shopping agents is creating a new performance advertising dynamic. Google's "Buy for Me" agentic checkout is now live in AI mode with partners including Wayfair, Chewy, and Quince. Google has also introduced a Universal Commerce Protocol with Shopify, Etsy, Walmart, and Target. The article highlights how product feeds are becoming bidding signals and notes Google's merchant-funded Direct Offers pilot. For e-commerce performance advertisers, this represents a fundamentally different buyer journey — agents compare and transact on behalf of users, changing how conversion signals flow back to campaigns. Omri Bitan, Head of Strategy: "Agentic commerce is a structural shift, and here's the critical insight: Full-on agentic buying is only at its early phases, and the more near-term opportunity is agentic buying of media. Agents value clean, contextual data over algorithmic black boxes. When AI agents evaluate placements, they care about content relevance, audience intent, and advertiser transparency — exactly what the open web delivers through contextual targeting and direct publisher relationships. Performance advertisers who combine first-party data, audience targeting, and contextual placement, will find their inventory more competitive in agentic auctions because it's built on data agents can parse and evaluate. " ### Update 6: TripleLift Launches TL Direct — Unified Self-Serve Buy/Sell Orchestration Fragmented workflows are a drag on performance optimization. TripleLift made its TL Direct self-serve platform generally available, giving buyers and sellers unified control over curated deals across web, mobile, and connected TV (CTV) with user interface (UI), API, and model context protocol (MCP)-based access. A commissioned Forrester study found it can save an average of 40 hours per campaign. Early partners include Raptive and Amerge. For performance teams running campaigns across multiple channels and deal structures, this represents a consolidation opportunity: fewer logins, less manual trafficking, more time optimizing. Omri Bitan, head of strategy: "Platform consolidation is a performance lever people don't talk about enough, but consolidation itself isn't the answer — transparency is. When you're managing campaigns across multiple channels, what matters is clear visibility into inventory quality, audience reach, and conversion efficiency. You need a platform that reduces the need for complex orchestration layers by giving you native multi-channel reach and reporting. The real efficiency gain isn't fewer logins; it's fewer black boxes." ## What This Means for Performance Advertisers The through-line across these updates is clear: measurement is maturing, quality is becoming a competitive advantage, and new channels are forcing performance teams to expand their definition of full-funnel. Whether it's conversion tracking closing gaps at new platforms or AI agents reshaping e-commerce journeys, performance advertisers in 2026 have more tools — and more pressure — to optimize efficiently across a fragmented landscape. The winners will be teams that stay close to these shifts and move quickly to test them. ## FAQs: Questions Performance Advertisers Are Asking ### FAQ 1: Which DSP should I use for performance campaigns in 2026? The short answer: There's no single best DSP anymore — but the real strategic question is whether you're diversifying beyond walled gardens into the open web. For pure-play performance (search, direct response), closed platforms like The Trade Desk offer strong transparency and inventory depth, but they come with vendor lock-in and limited supply diversification and aren’t really geared to performance at their core. For YouTube-heavy campaigns, you're effectively locked into DV360. For CTV campaigns, multiple platforms claim parity, but inventory quality varies significantly. The open web opportunity: Open-web platforms offer a fundamentally different value proposition — direct publisher relationships (9,000-plus integrations), contextual relevance that doesn't rely on cookies, and first-party audience control. This matters especially as privacy regulations tighten and third-party data becomes less reliable. Contextual targeting on premium publisher inventory delivers comparable performance to audience-based DSPs but with better brand safety, transparent inventory, and data portability. Best practice: Build a diversified media mix rather than concentrating on one DSP. Allocate 50-60% to your primary performance channel, 30% to test challengers (including open-web platforms), and 10-20% to experimental channels. Use contextual and audience signals to understand where your highest return-on-investment (ROI) conversions come from — you may find open-web inventory outperforms premium social in unexpected categories. ### FAQ 2: Should I allocate budget to TikTok for performance campaigns, or is it still just for brand awareness? The short answer: TikTok is now a legitimate performance channel, especially for e-commerce, D2C, and mobile app install campaigns. The data speaks: 40% year-over-year increase in search volume and TikTok's own Smart+ and GMV Max automated buying products are built for conversion optimization, not just reach. Early performance advertisers testing TikTok as a conversion channel (not just awareness) are seeing strong results, particularly in categories like fashion, beauty, home goods, and mobile apps where impulse and discovery align. Caveat: TikTok performs differently than Meta or Google. The audience skews younger, the creative expectations are different (authenticity > polish), and the conversion window is often same-session or next-session. If your audience is Gen Z or millennial, and your product has impulse-buy appeal, test it. If your audience is 45-plus, or your product requires long consideration, hold off for now. Best practice: Run a 30-day test with a dedicated budget allocation (start with 5-10% of your ad spend). Use TikTok's conversion tracking and Smart+ optimization to gather performance data. If return on ad spend (ROAS) meets your threshold, expand. If not, you've learned it's not the channel for you without over-committing. ### FAQ 3: How should I think about conversion tracking and attribution when AI agents are doing the buying? The short answer: Traditional last-click attribution breaks down when AI agents are in the loop, but open-web performance platforms have a structural advantage: first-party data control and transparent audience signals. Here's the opportunity: In agentic commerce, the agent's decision-making is based on data it can evaluate transparently — product data, pricing, audience context, and conversion signals. Closed platforms with proprietary algorithms are actually at a disadvantage in agentic systems because agents can't reliably parse them. Open-web platforms and contextual signals are naturally more agent-friendly because they're built on transparent content analysis and first-party audience data. What to do: - Shift your focus from campaign metrics to feed quality and audience data transparency: Are your product descriptions contextually aligned? Is your audience data clean and first-party sourced? - Invest in contextual relevance: Agents evaluate content-to-product fit. Using contextual targeting strategies ensures your ads appear in high-relevance contexts where agent recommendations will favor your product. - Use first-party audience data as your anchor: Build and own your audience segments (lookalikes, website visitors, CRM data). Agents can understand and value transparent first-party signals better than opaque third-party audiences. - Track feed-level and context-level performance, not just final conversions: Measure which content contexts drive the highest-value conversions, then optimize your contextual targeting and product positioning to match those contexts. Best practice: Start auditing your first-party audience data and product feed alignment now. Advertisers can use contextual targeting to understand which content contexts convert best for your products, then build predictive audiences based on those high-performing contexts. This gives you a native advantage in agentic auctions because your inventory and audience signals are transparent, contextually relevant, and agent-parseable. --- ### Premium Inventory: Understanding Highly Desirable Ad Locations and How to Use Them URL: https://www.taboola.com/marketing-hub/premium-inventory/ Last Modified: 2026-06-22 10:45:05 Premium inventory has long had a place in a marketer’s toolkit, though the locations and technology have changed in the digital era. Conceptually, though, premium inventory refers to high-quality, top-tier ad space on reputable platforms. Each industry and platform might be different, but generally, premium inventory is available space that an advertiser considers the highest value, guaranteed to get attention and engagement from the right viewers. It also comes at a high cost, so premium inventory has to be used wisely, and measured carefully, as part of an ad campaign. ## What Is Premium Inventory? Premium inventory ad placements vary widely depending on the industry, time of year or season, and website. While these may have been a full-page or above-the-fold ad in the newspaper era, now, premium inventory might refer to a coveted newsletter position, an exclusive website banner, a sole sponsorship of digital content, or other prime website real estate. Examples of premium inventory in the modern digital era include: - Exclusive sponsorships, such as high-visibility ads on top 100 sites (think New York Times or Forbes), or branded content. - TV and video, like sound-on ads that can’t be skipped during streaming content or as commercials in live sports. - High-impact spots, like above-the-fold banners, newsletter positioning, or homepage takeovers. - Private marketplaces, where advertisers bid on invite-only, high-quality inventory. - Targeted placements, where ads are guaranteed (due to first-party data use) to appear in high-quality, relevant environments. ## Core Characteristics of Premium Inventory ### Visibility Premium inventory ads are premium because they’re highly visible, meant to be seen by as many qualified users as possible. They might be above the fold on a site, or in another high-traffic area, or even out of home (OOH), like a billboard in Times Square. ### Safety Premium inventory space is also safe for brand reputation. Ads will show up on trusted, reputable sites, without the risk of appearing near harmful content. ### High-Performance Premium inventory should perform well in terms of conversions, because the ads are designed to generate high user interaction, such as in premium video placements or full-screen mobile spots. Or, they’re targeted to niche or high-value demographics to drive engagement. ## How Premium Inventory Is Valued ### Session Depth and Placement Priority Within digital advertising, premium inventory valuation takes into account a combination of high placement priority and low session depth. That means that ads are served early in a user’s visit, or toward the top of the page, to the most engaged users. So, these ads put a premium on the first impressions of a user’s visit, and apply settings to make sure the most valuable ads come up first. ### Market Supply and Demand Dynamics The factors of market supply and demand dynamics within premium advertising take into account the economic concepts of supply and demand. Premium inventory for advertising, which could include high-demand ad space, luxury goods, or infrequently available spots, can command higher prices because of its high demand, such as its uniqueness or the value it offers. Premium inventory pricing is often scarcity pricing. It can also be related to timing, such as during a peak season, when prices go up as demand goes up for a certain amount of time. Time-sensitive inventory, like hotel rooms during the Super Bowl, depend on speed to sell before the inventory becomes worthless (or at least, worth much less) the day after the game. Finally, brand-safe, targeted inventory can command higher prices because advertisers know they’ll see better return on investment (ROI) and engagement. ### Pricing and Guaranteed Deals Within the premium inventory market, programmatic guaranteed deals are a way for advertisers to avoid auction volatility uncertainty. They can fix the audience and placement ahead of time and are able to plan the cost as well, which can offer peace of mind and predictable performance for a campaign. Publishers can also benefit, since they don’t have to rely on unpredictable markets themselves. Premium inventory pre-set pricing can be useful for predictability and planning, too. Pre-negotiated pricing can often be more expensive for premium ad inventory, but the guarantee of a prime ad spot might be worth it to an advertiser, depending on their goals and the situation. ### Programmatic vs. Direct Sales Channels There are a few options for premium inventory sales, depending on what an advertiser wants to achieve. Programmatic sales use real-time bidding and can bring lower costs, plus add scale and efficiency, if an advertiser can bid or use an advertising platform to take care of bidding. When efficiency, reach, and performance are the top priorities, programmatic sales can help a brand get more for their money through automation and careful audience targeting. Plus, this option can help to get the most performance out of leftover inventory. Direct sales require higher, guaranteed prices (CPM) that ensure exclusive and brand-safe ad inventory. This is a high-value, low-volume strategy, but it gives an advertiser full control over where the ad goes and what content is around it. For brand-building, takeovers, or other high-impact ads, this can be impactful. It’s also possible to use a hybrid approach, choosing direct sales for higher-tier inventory and programmatic sales to get more for your money. ## Premium Inventory Channels and Formats With many creative options for premium inventory advertising, these are the channels and formats to know: Format Description Performance Benefit Motion Ads Short, dynamic looping clips, or the “Ken Burns” documentary effect with static images and no sound. Subtly captures attention without being as intrusive as a full video. Vertical Video Ads 9:16 aspect ratio assets, originally designed for social, such as Reels and TikTok. Ideal for mobile-first premium environments and app integrations; high visual engagement. Carousel Ads Interactive multi-card units that allow users to swipe through products or steps. Excellent for storytelling or showcasing a product line; drives higher intent through interaction. Rich Media Interactive elements like polls, quizzes, or gamified CTAs within the ad unit. Drives deeper engagement and qualifies the user before they even reach the landing page. Story Ads Full-screen, mobile-optimized Instagram “Story” style formats appearing in premium feeds. Leverages the familiar vertical-swipe behavior of social users in a premium editorial context. ## Benefits of Premium Inventory for Advertisers ### Targeting Precision Premium inventory is premium because it’s in a prime position, so lots of users will view it. Along with that benefit, premium inventory includes precise targeting of specific audience segments based on high-quality data. You’re able to better control who sees your ads, where, and when, bringing guarantees that aren’t possible with other ad inventory selections. Targeting these days is often AI-driven, bringing better ROI and increased reach. ### Brand Protection Depending on your brand reputation and goals, you may choose premium inventory to make sure your content won’t be associated with any inappropriate ads, and cut down on the risks of exposure to bots or fraudulent traffic. Premium inventory ads only appear on reputable, trusted sites. ### Better Performance Using premium inventory as part of an ad campaign will likely show better performance, whether those metrics are click-through rates (CTR), engagement, conversions, or others. One study found that ads in premium, high-quality environments increased purchase intent by 40%. ## Challenges and Limitations of Premium Inventory ### Cost It’s not always possible to use premium inventory in an ad campaign, since these spots are expensive and often out of reach, especially for super-premium, more rare placements. At some point, the high cost might be impossible to justify against return on investment measurements. ### Availability Premium inventory spots are premium for several reasons, but availability is often one of them due to scarcity, demand surges, timing, and more. You may need to plan far in advance or forgo the premium spots when availability dwindles quickly. ### Suitability A B2B company may never require a Super Bowl advertising spot, because the audience might not overlap at all with their target audience. Creative formats can also play into premium inventory decisions, since the format, audience, and placement all have to line up to make it worthwhile. ## Alternatives to Premium Inventory Premium inventory can be an important tool for advertisers in particular industries or with a robust budget. There are other ways to think about targeting and engaging audiences, though. Consider these alternatives to premium inventory. ### Direct Deals Programmatic direct deals can be useful for niche or contextual ad networks that are well-matched to the brand. These deals offer a first look at ad inventory at set prices, and cost less than guaranteed premium inventory. It’s a way to target audiences without going over budget, or to use occasionally or when guaranteed spots aren’t as crucial. ### Remnant Inventory So-called remnant inventory is unsold, lower-cost ad inventory that’s long tail or especially targeted. Consider this when you’re working with a smaller budget and can use real-time bidding to find just the right spot for your creative and audience. ## How to Measure Performance on Premium Inventory Ad Placement When you’re paying premium inventory prices, measuring and evaluating performance will be essential. With continuous performance monitoring, advertising teams can adjust bids, targeting, and creative for each premium placement. Keep these factors in mind to gauge whether and how well premium inventory ads are working for your business. ### Key Metrics for Performance Evaluation Some metrics will be more useful than others when evaluating the performance of premium inventory ads: - Gross margin ROI to evaluate the profit generated from investing in inventory, especially for high-value items. - Stock-to-sales ratio compares stock levels to sales to make sure that premium items are in stock, but not overstocked. - Inventory turnover ratio to measure how often premium inventory turns over, or is sold and replaced, with a high ratio indicating desirable or fast-moving spots. - Sell-through rate measures how much inventory was sold versus the amount received; higher rates mean there’s a strong demand for premium goods. - Inventory carrying cost is the total cost of holding inventory, which should be low to protect margins when it comes to premium inventory. Other premium inventory performance metrics might include perfect order rate, backorder rate, and deadstock rate. ### Viewability, Engagement, and Conversion Outcomes When measuring premium inventory performance, keep in mind there are generally much higher benchmarks than open exchange inventory. So, premium inventory should have higher viewability metrics — meeting or exceeding the standard of 50% pixels in view for one to two seconds. Premium placements also lead to higher viewability. One study found these placements can lead to 73% viewability compared to 45% for below-the-fold ads. Viewability can even get to 100% for full-screen, unskippable ads. Engagement measures the interaction and focus of users with premium ads. Research found that premium inventory outperformed open exchange inventory by 143% in terms of attention and engagement. Video performance is also high with premium inventory, with 90% or more completion rates with higher quality content. The essential measurement of conversion matters greatly with premium inventory. It’s possible to use premium inventory to help move users from attention to conversion. The nature of premium inventory can improve conversion rates, too: brand-safe impressions drive a 233% lift in conversion rates, while combined viewable and brand-safe impressions drive a 57% conversion rate. ### Attribution Challenges Premium ad inventory attribution is part of the bigger picture of how premium placements are performing and whether they’re worth the cost. Challenges to understanding attribution include data loss due to privacy regulations; cross-device, nonlinear journeys; and last-click models that don’t or can’t value brand awareness efforts. In the case of some premium spots, like streaming services, the positive signals remain within that silo, making it difficult to incorporate them into broader metrics tracking. Premium inventory placements often drive long-term brand lift or offline-to-online effects versus easily trackable clicks in the moment. ### Aligning Inventory With Campaign Objectives Using premium inventory as part of an ad campaign has to take into account the attribution challenges as well as the goals or desired outcomes. Especially because of the costs likely involved, advertisers should have strategic goals that align with using premium inventory placements. These high-quality ad placements promise brand safety, engagement, and viewability, which often line up well with brand awareness, conversions, or guaranteed access to new or desirable audiences. Premium ads can also be a way to augment audience targeting data when trying to increase user engagement and attention. ## Key Takeaways Premium inventory for advertisers can bring great visibility, engagement, and performance, but it also comes at a cost. Premium inventory might be the top of the page on a hugely popular website, a site takeover on a launch day, or ads running during a most-watched streaming show. Advertisers can take advantage of premium inventory to get guaranteed, brand-safe placements, but at that cost, they also have to consider the best ways to use premium inventory under budget. Consider how, when, and how often to incorporate premium inventory into ad campaigns to get the most out of this option. ## Frequently Asked Questions (FAQs) ### Why should advertisers prioritize premium inventory in their campaigns? Premium inventory typically offers advertisers some enticing factors: high-quality audiences, brand-safe environments, and reliable visibility. These all help to increase engagement, improve campaign performance, and reduce risks associated with low-quality or non-brand-safe placements. Depending on budget and campaign goals, premium inventory can offer a lot of return for advertising teams. ### How can advertisers optimize campaigns across multiple premium inventory sources? There are lots of premium inventory sources available today, whether high-visibility, popular websites or streaming services. Advertisers can optimize campaigns across multiple premium inventory sources by continuously monitoring performance metrics for each source and examining them as a whole. They can also adjust bids, targeting, and creative for each placement continuously to make sure they’re spending wisely. Using programmatic tools and cross-channel analytics helps maintain consistent reach and engagement across different publishers. ### How does creative format influence performance on premium inventory? There are lots of creative ad formats to consider, and the choice of creative format — such as video, native, or interactive media — can have a big impact on engagement and conversion rates when it comes to premium inventory. Advertisers should match the format to the audience and placement context to maximize visibility and user interaction. --- ### The AI ROI Crisis: Why Zero-Click Search Is Pushing Advertisers to the Open Web URL: https://www.taboola.com/marketing-hub/zero-click-search-impact-on-advertisers/ Last Modified: 2026-06-16 11:18:05 Paid advertising was previously a clean exchange, where you paid for intent and gained a visitor. But, these days, that’s not the case: AI-generated answers are now satisfying user intent directly on the search results page, before a single click ever happens. With studies suggesting that nearly 60% of searches now end on the SERP itself, the engine you’ve been relying on to drive traffic is increasingly keeping users for itself. If your ad budget is funding a platform that’s turned into a walled garden, it’s worth asking what you’re really getting in return. ## The Evolution of Search: From “Blue Links” to “Answer Boxes” Not long ago, a Google search was a portal that pointed you somewhere. Type in “best running shoes for flat feet” and you’d get 10 blue links to articles, product pages, and reviews. You clicked on one, you visited a website, you maybe returned to the SERPs and clicked on another link to another website. Brands captured your attention and drove you to their sites. That model is fast eroding. The Search Generative Experience (SGE) impact has rewritten what happens between search engines and users. What began as Google’s AI Overviews (AOI) feature — a box of synthesized answers above the organic results — has now become the default interface for hundreds of millions of queries. This is no longer in beta, but the main first-result across informational, navigational, and commercial search queries. The intent of the user is still there, but the click is not. Users get what they need in terms of answers, comparisons, and recommendations without ever landing on your product page. Without clicking, they’re likely not converting. This is a deliberate architectural shift, but still ultimately falls in line with what Google has always tried to do — provide answers to user questions. AI has simply made it better at doing so without sending users somewhere else. For businesses large and small that built their entire acquisition funnel on search traffic, this is a significant change. ## Is Traditional PPC ROI Declining? The 50% CTR Drop Case Study Initial thoughts around AI Overviews hurting click-through rates is now backed by hard data. The picture that’s emerging is one of a structural decline in PPC efficiency in AI-era conditions. This is not a blip on the radar; it’s a new baseline. Across a range of industries, marketers are reporting that campaigns which previously had healthy ROAS results are now requiring significantly more budget to generate the same revenue. CAC is increasing, impression share is shrinking, and even when ads do appear, few people are scrolling beyond the AI answers to find them. The numbers don’t lie. Research from Seer Interactive reveals that when an AI overview appears on a results page, paid click-through rate drops from around 21.27% to only 9.87%, a decline of more than 50%. That means for every 100 users who would have clicked on your ad before AI Overviews, fewer than 50 are now doing so. Seer Interactive’s 2025 analysis also found similar results, documenting that high-funnel paid search efficiency is cratering, specifically for queries where AI answers appear. These aren’t isolated cases, but increasingly common problems impacting consumer product categories. The concept of zero-click search in 2025 has firmly brought these problems to the forefront of marketers’ minds. When users search and get a three-paragraph AI answer that makes them stop scrolling, ad budgets absorb impressions with zero returns. This performance reality isn’t going anywhere any time soon. ## The “Double Hit” to Search ROI: Rising Costs and Falling Visibility Declining CTR is only part of the picture. What happens to the economics of ad slots when AI Overviews are taking up valuable top-of-SERP real estate? AI Overviews now dominate this above-the-fold space. On mobile, where the majority of e-commerce searches now happen, a fullscreen AI overview can push every paid ad below the fold quite significantly. The total inventory of high-visibility ad positions has greatly contracted over the last year, but the number of advertisers competing for those positions has not. Supply is down, but demand is steady, so what does this mean for marketers? Cost per click goes up. Advertisers fighting for the last visible placements are paying premium prices for diminished returns. What’s happening is that you’re bidding in an auction where the prize has shrunk, but the competition has become fiercer. Meanwhile, Google Ads versus AI Overviews is becoming a tension point inside Google itself. Google has a financial incentive to keep ads valuable, but it also has a product incentive to make AI overview as useful as possible for users. For now, those two objectives are in conflict and advertisers are the ones paying for it. ## The Strategic Pivot: Why Advertisers Are Migrating to the Open Web For performance marketers, the core strategic question becomes, “If Google’s AI is answering questions instead of sending users to our site, where do we go next?” The answer for a growing number of growth-focused brands is the open web. The walled garden versus open web distinction has never been more commercially important than now. Inside walled gardens like Google Search, Meta’s feed, or Amazon’s placements, the platform has full control over the user experience, the algorithm, and now, the answer. Your brand is simply a guest and rules can change overnight. Outside of those walls on the open web — home to news sites, lifestyle publishers, niche media, and vertical content — the dynamic is different. There’s no AI overview intercepting user attention before it reaches your ad. Users arrive at a publisher site because they chose to. They’re reading and in a discovery-focused mindset. When your ad natively appears in the content, it’s part of their total experience, rather than a distraction from it. This is what makes the migration to an open web approach strategically compelling: It’s a channel where you can own the interaction, from impression to landing page, without an AI layer filtering your audience before they ever get to see your offer. ## Beyond the SERP: Diversifying Into Publisher Ecosystems An open web advertising strategy typically involves three channels working together — programmatic display, native advertising, and direct-to-publisher partnerships. For a small-to-medium business that’s product focused, native advertising is often the best entry point. Native ads are designed to match the form and function of the editorial environment in which they appear. A sponsored article on a health publication, a recommended product story on a lifestyle site, or a content recommendation at the bottom of a relevant post are all contextually relevant rather than interruptions, and that’s exactly why they convert. The publisher ecosystem marketing model also gives you something search advertising increasingly cannot: transparency and control. You know where your ad is running and you can align your message with specific content verticals. You can test ad creative without entering into a bidding war. Because users are going to a landing page, rather than competing with an AI answer block, you can shape the post-click experience in a meaningful way. The native advertising ROI case is strengthening as search ROI weakens. Brands that diversified their acquisition mix early are now seeing their open web channels absorb volume that previously came from paid search and, typically, at lower CPAs. For merchants with around 10-20 unique products, that’s a significant advantage. ## Measuring Success in the Age of AI Search One reason brands have been slow to adapt is their focus on the wrong metrics. Search volume or paid search impressions now measure a world that isn’t the same, if it even exists at all. Gartner predicts that traditional search volume will decline by 25% in 2026 as users shift to conversational AI and direct answer interfaces. First-party data growth, direct attribution, customer lifetime value, and blended ROAS across channels are all the new metrics that need greater attention. The emergence of agentic commerce trends adds another layer to the urgency of this recalibration. When an AI agent is doing the buying, traditional keyword targeting becomes less relevant. AI Overviews ROI is directly competing in the age of AI search, and the brands winning aren’t clinging onto search dominance. Instead, they’re building different acquisition funnels across a full digital landscape. ## Key Takeaways AI Overviews have fundamentally altered the search experience by answering user questions directly on the results page, reducing the need for users to click through to a website. For advertisers, this translates into a measurable compression of click volume, spikes in CPC for remaining inventory, and a widening gap between spend and revenue. The strategic response here is not to abandon search entirely, but to diversify. The open web, accessed through native advertising and programmatic placements, offers a compelling alternative with engaged audiences, brand-controlled experiences, and first-party data that search increasingly cannot provide. ## Frequently Asked Questions (FAQs) ### Are AI Overviews reducing PPC clicks? Yes, evidence from multiple studies has found that when an AI overview appears on a results page, paid click-through rates can fall by more than 50%. ### Will traditional search volume continue to drop? The signs are there — Gartner predicts a 25% decrease in traditional search engine volume throughout 2026 as users move to answer-first interfaces like AI chat tools. ### What is the “open web” in advertising? The open web is a universe of independent publishers and digital media sites that exist outside of closed platforms like Google or Meta. Advertisers use native and programmatic ads here to reach users in environments where the brand controls the landing experience. --- ### How D2C Brands Can Run CTV as a Successful Performance Channel URL: https://www.taboola.com/marketing-hub/ctv-for-d2c-brands/ Last Modified: 2026-06-11 06:53:53 The era of cheap Facebook audience acquisition is officially over. CPMs are up, audiences are saturated, and the incremental return from Meta campaigns is shrinking every quarter. Connected TV (CTV) has been on the radar of many brands as a natural progression channel from search and social, but performance marketers are often wary of this leap when clicks and last-touch conversion have been the priority. What was once an exclusively brand-awareness channel, though, has evolved into a performance TV advertising option that can generate measurable, lower funnel results when approached with a strategic mindset and the right infrastructure. ## 4 Strategies for D2C Brands to Run CTV as a Performance Channel ### 1. Solve the Cookie-Less Attribution Problem For most performance marketers, the biggest concern around CTV is attribution. It’s a legitimate concern: CTV ads run on shared household televisions, carry no cookies, and can’t be clicked. When a conversion happens, there’s no pixel firing, no UTM parameter, no direct signal to connect that purchase back to the TV impression. This is the central CTV measurement challenge — closing the loop between living room ad exposure and a purchase on a personal device. The primary way to overcome this is to use IP-based identity matching, where a connected household IP that’s used to receive the CTV impression is later used to make a website visit or a purchase. It’s not click-level precision, but it’s a credible and widely used attribution option. Beyond IP matching, cross-device attribution can be strengthened using third-party identity graphs that put together household device clusters using hashed email addresses, login data, and other deterministic signals. These allow you to follow the user’s journey from TV exposure to mobile browsing to desktop checkout with a higher level of confidence. One platform built to close this gap is Realize. Using advanced audience matching to retarget users across personal devices after a CTV exposure, Realize leverages a network of over 9,000 premium open web publishers to follow up a TV impression with relevant, action-oriented advertising. Partnerships with major streaming platforms like Paramount mean that CTV exposure can now be directly linked to downstream sign-ups or purchases, within a single attribution view. Complementing identity-based attribution, incrementality testing offers another layer of CTV proof. By splitting your audience into exposed and holdout groups and comparing their conversion rates, you can measure the true lift of your CTV campaign, independent of other channels. Branded search lift — tracking whether CTV-exposed households show increased branded searches in the days post exposure — is a lighter version of the same principle, which can be set up without a major measurement infrastructure investment. ### 2. Overcome Audience Fragmentation With Programmatic Buying Unlike walled garden systems like Meta, CTV inventory isn’t a single place you go to buy ad space. Roku, FireTV, Apple TV, Peacock, and more all have their own ad stack, audience identifiers, and frequency logic. For brands used to the unified simplicity of Meta, the fragmentation here can feel daunting. The solution is programmatic TV buying through a Demand Side Platform (DSP). A DSP aggregates inventory across the CTV ecosystem, letting you buy audiences rather than just placements. Instead of negotiating separate deals with individual streaming apps, you define your target audience (demographics, purchase intent, content preferences, etc.) and let the DSP find those people wherever they’re watching. This is the most efficient and effective method, especially for brands without dedicated media buying teams. Programmatic TV also gives you first-party data activation. Your existing customer list can be uploaded into a DSP and used for suppression, so you’re not wasting CTV budget on retargeting those who have already purchased, and for lookalike modeling, where you can build audiences that mirror your best customers’ profiles. This is the CTV equivalent of Meta’s custom audience feature and lookalike audience functionality, and it’s one of the features that makes OTT advertising for D2C brands a more viable performance channel. Incremental reach is the key here — you’re not just re-reaching the same people you’re already targeting on search and social. CTV lets you extend your reachable audience to cord-cutters and streaming-first households that are largely invisible on digital channels. ### 3. Produce Performance Grade Creative on a Budget One of the most persistent myths about CTV is that it requires TV-level production budgets to create a decent ad. It doesn’t, but it does require a different creative approach than what you’re used to on social. For most D2C brands, your existing social video assets are a great starting point. Short-form social content can be adapted for CTV by extending the runtime to around 15-30 seconds, adding a clear verbal and visual call-to-action. Including a branded URL or QR code that gives users a direct path to action can also work well. The storytelling rhythm that works on social needs to be slowed a little for this format, but the core message and visual identity can carry over seamlessly. The creative framework that typically works best on CTV is problem > solution > proof. Open with a relatable pain point your customer has, then introduce your product as the solution, closing with social proof like a customer result, a before and after, or some kind of credibility signal. This structure works in 15 seconds and scales to 30, so you have options. It’s direct response logic applied in a television format, and what separates CTV ads that drive D2C CTV campaigns with measurable outcomes from ones that simply look good but do nothing. While it’s tempting, resist the urge to run a pure brand spot to “be on the big screen.” The point of performance CTV is that every impression should be working for you. Build creative that tells viewers exactly what to do next and you’ll end the campaign with a measurable signal, rather than only vague brand recall. ### 4. Scale Spend Without Destroying CPA (the Frequency Trap) CTV’s fragmented inventory landscape can quickly create a budget-burning environment due to uncontrolled frequency. Without a centralized mechanism to track how many times a given household has seen your ad across different apps and publishers, it’s entirely possible to serve the same viewer your 30-second spot a dozen times in a single week. Each of those redundant impressions costs and none of them are moving the viewer closer to a purchase. This is where frequency capping becomes not just a best practice, but a financial necessity. The goal is to set a universal household-level cap, usually around three to five exposures a week, that applies across your entire CTV buy, regardless of app or publisher. Enforcing a cross-platform frequency cap requires that you either run all your CTV buys through a single DSP with universal frequency logic, or use an identity solution that can recognize the same household across different inventory sources. Neither is a perfect solution, but even partial control is better than none. Beyond protecting against waste, frequency management directly impacts your CTV ROAS. Campaigns that run without caps tend to see CPA inflate over time as the same households get over-served, while untouched audiences go unreached. Keeping your frequency capped preserves budget for genuinely new impressions, expands your effective reach, and works on decreasing CPA as your campaign matures. ## Key Takeaways CTV is no longer a channel you add to a media plan for brand awareness and hope for something more. Instead, with the right identity infrastructure, it can function as a conversion source that reaches audiences your existing channels can’t. The problems of attribution can at least be partially solved with IP-based matching, identity graphs, and incrementality testing, while inventory fragmentation can be avoided through DSP-based programmatic buying. Instead of thinking of CTV as a standalone channel, it should be used as a top of funnel acquisition tool that works alongside your search and social retargeting. As the costs on these channels continue to rise, brands building CTV campaigns (and knowing how to measure and test on this platform) will be the ones who succeed long term and have an acquisition advantage in the next wave of social inflation. ## Frequently Asked Questions (FAQs) ### How can DTC brands measure CTV performance effectively? The most practical approach is IP-based cross-device attribution with incrementality testing. This lets you quantify the actual life your CTV campaigns are driving, while connecting households to visits or purchases. Combining these approaches gives a credible performance picture without requiring click-level data. ### Is CTV advertising too expensive for small D2C brands? Not with programmatic buying. Unlike traditional television, which requires a large upfront commitment, programmatic CTV lets you set your own budget, define your audience precisely, and pay only for impressions that reach households of your target profile. This dramatically reduces wasted spend, so it’s ideal for brands with smaller budgets. ### What is the ideal frequency for a CTV campaign? Most CTV performance marketers find that three to five exposures a week per household is the sweet spot between building recall and triggering fatigue. Below that, viewers won’t retain enough information to act. Above, you’re spending money on impressions that annoy people who have already made a decision about your brand and product. --- ### What is Supply Path Optimization? Understanding SPO URL: https://www.taboola.com/marketing-hub/supply-path-optimization/ Last Modified: 2026-06-10 10:44:27 Ever feel like your ad budget is a traveler trying to get from Point A to Point B, but every time it takes a turn, a mysterious toll booth appears? In the programmatic advertising world, intermediaries, hidden fees, and confusing detours clutter that journey. If you’ve ever looked at your reporting and wondered why your working media isn’t working as hard as it should, you’re ready to talk about supply path optimization (SPO). Think of SPO as the Marie Kondo of digital advertising. It helps you declutter your supply chain, keeping only the paths that “spark joy” — or, in this case, deliver actual value. No more paying fees to intermediaries who aren’t pulling their weight. No more duplicate auctions consuming your budget. Only clean, efficient routes to quality inventory so that your dollars actually reach the destination: your audience. ## What Is Supply Path Optimization? Digital advertisers use SPO to find the most direct, cost-effective, and transparent route to buy digital ad space. This strategy helps advertisers find supply-side platforms (SSPs), ad exchanges, and publishers that deliver the best bang for the buck. In the early days of programmatic, we thought more paths were better. If you could buy an ad on a premium site through five different exchanges, why not? Header bidding changed the game, though. Suddenly, publishers could offer the same impression through multiple SSPs simultaneously. Advertisers found themselves bidding against themselves, for the same impression, multiple times. SPO emerged as the solution. In lieu of participating in every auction, savvy advertisers began mapping out the paths that actually delivered unique value. They began consolidating partners, eliminating redundancies, and focusing spend where it mattered most. Boiled down, SPO helps you make choices: Which auctions should you participate in and which should you skip? This supply-chain strategy removes unnecessary layers while preserving and improving access to quality inventory. ## Why SPO Matters: Benefits for Advertisers and Publishers The modern programmatic supply chain has a lot of bloat. Studies suggest up to 15% of every programmatic dollar vanishes into hidden fees, redundant tech layers, and inefficient routes. ### Cost Efficiency and Spend Optimization Do you know what happens if you bid through five different SSPs for the same impression? You’re potentially paying five different fees because everyone gets a cut. It makes sense, then, to eliminate redundant paths and reroute your budget toward actual impressions. ### Transparency and Control SPO shows where your ads appear, who’s handling them, and where your money goes. It also tells you: - Which SSP delivers the best cost per mille (CPMs). - Where the hidden fees hide. - Whether you’re paying for bot traffic. ### Better Inventory Quality and Fraud/Risk Mitigation Your supply chain is only as strong as your weakest SSP. One sketchy partner can expose your entire campaign to fraud, brand safety issues, or low-quality placements that tank your performance. SPO acts as your quality filter, scrutinizing supply partners and eliminating those prone to fraudulent activity (think bot traffic, domain spoofing, or click farms), dramatically reducing your risk. ### Improved Data Fidelity and Performance Here’s something most marketers don’t often think about: every hop an impression takes through the supply chain degrades data quality. Critical metadata, like page context, user signals, and viewability information, gets lost or distorted with each additional intermediary. Poor data means poor targeting. Poor targeting leads to wasted spend and missed opportunities. SPO preserves data fidelity by minimizing hops between advertisers and publishers. The fewer the intermediaries that touch your impression, the cleaner your signals. Better data means better optimization, more accurate bidding, and stronger campaign performance. ## How SPO Works: Mechanics and Key Considerations SPO isn’t magic. It’s methodology. The process starts with data collection and analysis, usually involving an algorithm or set of manual rules that evaluate SSPs and look at: - Auction mechanics: Is this exchange running a fair second-price auction? Is it slower than molasses, causing you to miss an auction entirely? - Fees: How much are they taking off the top? - Unique access: Does this path give me something I can’t get elsewhere? Once you gather this intel, you rank your supply partners. Prioritize the best performers — those offering direct publisher relationships, transparent pricing, quality inventory, and strong fraud protection. Renegotiate or cut the underperformers. Now, some advertisers take a conservative approach. They consolidate to a handful of proven SSPs and call it a day. Others embrace artificial-intelligence (AI)-driven optimization, letting algorithms continuously evaluate paths and shift spend dynamically, based on performance. The most sophisticated advertisers set guardrails through preferred supply lists while allowing machine learning (ML) to find efficiencies within those parameters. ### Key Considerations Include: - Auction mechanics: Know whether the SSP is running a second-price vs. a first-price auction. You’ll pay slightly more if you’re bidding in a second-price auction. - Data leakage: Some platforms leak information that can increase costs, especially if your SSP shares bid data with competitors. SPO helps you identify and avoid these partners. - Resellers vs. direct relationships: An SSP with direct publisher integrations typically offers better value than one reselling another exchange’s inventory. - Signal quality: Platforms with code-on-page integrations capture richer signals than those that rely on header bidding wrappers or other indirect connections. This capability matters a lot for AI optimization. ## Challenges and Limitations of SPO SPO’s biggest risk is over-optimization. Cut your supply paths too aggressively, and you limit scale because you’ve got fewer options to explore and might miss something. Finding the sweet spot between efficiency and reach requires constant calibration. You need enough paths to maintain scale and discovery, but not so many that redundancy and waste creep back in. Another challenge is incomplete transparency. Identifying legit direct relationships isn’t always straightforward, especially when exchanges access publishers through header bidding wrappers or other intermediary tech. Communication gaps also complicate matters. Buyers doing SPO in isolation with no input from publishers or SSPs may fall short of goals. Publishers have preferences, too (that’s demand path optimization, or DPO). Misalignment can create inefficiencies neither party intended. A caveat: SPO isn’t a set-it-and-forget-it solution. The programmatic landscape shifts constantly. Publishers change partners. SSPs adjust fee structures. New fraud schemes emerge. The best supply path today might fall short tomorrow. ## When (and How) to Use SPO Prioritize SPO whenever you’re running complex programmatic campaigns (especially if you plan to scale). It’s particularly valuable if you: - Notice unexplained cost increases. - See inconsistent performance across similar campaigns. - Have limited visibility into where your ads actually run. - Suspect you’re paying for duplicate impressions. These are all red flags insisting you need a supply path analysis. If you’re in a regulated industry or care about adjacency (and you should), SPO helps ensure your ads only appear through vetted, trustworthy partners. Use SPO proactively, not reactively. Don’t wait until you’ve blown through budget on underperforming inventory. Make it part of your campaign planning process from the start. ## How to Implement a Supply Path Optimization Strategy ### 1. Inventory and Supply Audit Start by auditing your current programmatic setup. Identify which exchanges are delivering your top-performing impressions. Document the relationships: is this SSP directly integrated with publishers, or is it reselling through another exchange? Look for duplicate access — are you reaching the same publishers through multiple SSPs? That’s your first optimization opportunity. ### 2. Performance and Risk Assessment Data time! Analyze historical performance across all supply paths. Calculate key metrics for each SSP: - Win rate. - Average CPM. - CTR and conversion rates. - Viewability scores. - Fraud/invalid traffic rates. Assess qualitative factors, too: - How transparent are each partner’s fees? - How quickly does each partner close auctions? - Do they provide adequate reporting? - Have they had any recent brand safety incidents? ### 3. Build a Preferred Supply List/Blue List Create a whitelist of trusted partners. Focus on direct relationships and transparent exchanges. - Tier 1 (blue list): Your best performers — direct publisher partnerships, transparent pricing, quality inventory, and strong fraud protection. Prioritize these partners in your bidding strategy. - Tier 2: Solid performers with a little room for improvement or specific use cases where they excel. You’ll keep using them, but may negotiate better terms or apply stricter targeting. - Tier 3: Underperformers or redundant paths. You might reduce or cut these candidates. ### 4. Leverage Tools or Platforms with SPO Support Don’t go it alone. Modern DSPs and advertising platforms offer SPO features, including: - Automated path optimization based on performance. - Transparency reports showing supply chain details. - Built-in fraud detection and brand safety tools. - Direct publisher integrations. If you’re managing multiple DSPs, consider consolidating wherever possible. Fewer platforms mean better data, clearer insights, and more leverage in negotiations. ### 5. Monitor, Measure, and Iterate As I mentioned previously, SPO isn’t once-and-done. Set up regular reviews and track how your optimizations impact key metrics: - Are CPMs decreasing without hurting reach? - Is inventory quality improving (higher viewability, lower fraud)? - Are conversion rates stable or improving? - Is brand safety incident rate declining? A caveat: a supply partner performing well today might falter tomorrow. New partners with better access might appear. Market conditions shift. Stay agile. ## Key Takeaways Supply path optimization (SPO) is the shortest distance between budgets and customers, helping to eliminate redundancies, reduce intermediaries, prioritize transparency, focus on quality over quantity, and continuously optimize based on performance data. SPO also helps you reclaim control over advertising spend. Whether you manage campaigns across multiple DSPs or use an integrated platform like Realize, you’ll cut costs, improve performance, reduce fraud risk, and gain the transparency necessary for making informed decisions. ## Frequently asked questions (FAQs) ### How do I know if SPO is necessary for my campaigns? SPO is most valuable when you want to improve transparency, reduce intermediary fees, ensure high-quality inventory, and optimize spend efficiency. It’s particularly relevant for programmatic campaigns with multiple SSPs or complex supply chains — basically any campaign beyond very basic, small-scale efforts. Clear signs you should prioritize SPO: unexplained cost increases, inconsistent performance, brand safety issues, or an inability to tell where your ads are running. ### What are the risks of not implementing SPO in programmatic buying? Without SPO, campaigns often stress their budgets by paying multiple resellers for the same inventory. The result is wasted spend, higher exposure to low-quality/fraudulent inventory, and opacity in reporting and performance metrics. Skip SPO considerations, and you risk: - Less than optimal content distribution paths. - Inefficient budget allocation. - A reduced ability to monitor content delivery across preferred supply sources. ### How often should I review or adjust my supply paths? Don’t set-and-forget. Schedule formal audits and adjustments based on performance data, inventory changes, and shifts in market dynamics. Conduct monthly deep-dive reviews and consider weekly check-ins for large campaigns or during critical periods (like the holiday shopping rush). The goal is catching expensive leaks before they drain your budget and jumping on new, high-performing paths when they appear. --- ### 8 Performance Advertising Platforms for Financial Services URL: https://www.taboola.com/marketing-hub/performance-advertising-platforms-financial-services/ Last Modified: 2026-06-10 10:33:14 Financial services marketers face tough competition online. It’s expensive to acquire new customers, the industry is highly regulated, and conversion cycles are longer than in most other industries. In other words, people take more time to make financial decisions. At the same time, traditional growth channels like search and social media are becoming crowded and harder to stand out in. As a result, lenders, insurers, fintech companies, and wealth managers are exploring new performance advertising opportunities across the open web. Choosing the right performance advertising platform can be challenging, though. It’s no longer just about reach: The platform needs to be precise and compliant, while delivering measurable outcomes. Below, you’ll find a list of the most widely used platforms for financial services advertisers who are looking to generate qualified leads, scale acquisition efficiently, and maintain control over brand safety and performance optimization. ## 8 Performance Advertising Platforms for Financial Service Advertisers Compared Platform Why It’s essential Core Use Cases and Supporting Features Best for (Financial Niches) Pricing Model 1. Realize Open-web performance platform with AI optimization and publisher partnerships for conversion outcomes. Performance targeting beyond search/social; native, display, vertical video; conversion-focused AI bidding; advanced audience targeting. Lead gen for lenders, wealth management, and banking conversion campaigns. Performance-based model; campaigns billed on cost-per-click (CPC) basis, or cost per mille (CPM) for programmatic. 2. Google Ads A leading ad platform that offers advanced targeting tools and powerful automation features. Search, display, and YouTube advertising with intent signals, audience segmentation, automated bidding strategies, and robust conversion measurement tools. Personal loans, credit card signups, and insurance quote generation. CPC, cost per acquisition (CPA), and automated Smart Bidding models. 3. Microsoft Advertising Search platform with strong professional audience reach and lower auction competition compared to Google. Search and audience ads are enhanced by LinkedIn profile targeting, demographic segmentation, and business-focused intent signals. Investment firms, B2B financial services, advisory, and retirement planning campaigns. CPC or CPA bidding structures. 4. PropellerAds Global traffic network that supports multiple ad formats and helps businesses scale customer acquisition campaigns. Push notifications, popunder ads, native placements, and device-level targeting are supported by automated optimization tools and reporting dashboards. Forex offers, trading platforms, and international credit campaigns. Self-serve CPC and CPM pricing. 5. ROIads Focused on driving conversions, with detailed bid controls and AI tools that help improve campaign performance. Push and pop traffic formats combined with detailed targeting by geography, browser, and device, supported by automated optimization rules. Loan lead generation and fintech mobile acquisition campaigns. CPC or CPM with minimum funding requirements. 6. Adsterra Multi-format ad network offering broad global reach with built-in fraud protection and targeting flexibility. Native, banner, popunder, video, and social-style ad units supported by anti-fraud systems and customizable targeting layers. Financial publishers and global lead generation campaigns. CPC, CPM, or CPA, depending on campaign structure. 7. RichAds Performance-focused ad network that offers detailed optimization controls and precise campaign filtering. Push, native, and pop formats with automated optimization rules, micro-bidding, and detailed analytics dashboards. Credit offers, lending funnels, and investment sign-up campaigns. CPC or CPM with deposit requirements. 8. HilltopAds Built for audience targeting and global reach, with a strong focus on user privacy. Multiple ad formats supported by anti-ad-block technology and international traffic monetization capabilities. Small-to-mid-sized financial campaigns targeting emerging markets. CPC and CPM pricing models. ### 1. Realize Why it’s essential: Realize is an AI-powered performance platform that helps financial services brands reach high-intent customers across the open web by leveraging predictive matchmaking and deep contextual targeting. It serves as a comprehensive hub for scaling campaigns beyond social media, using real-time signals to connect advertisers with users who are actively researching financial products on premium news and business sites. In the context of financial services, you would use Realize to drive lower-funnel actions such as credit card applications, insurance quotes, or mortgage inquiries, while maintaining strict brand safety and compliance. The platform is used to automate complex bidding strategies and transform existing marketing assets into high-performing native or motion ad formats that blend seamlessly into trustworthy editorial environments. Showcased features: - Automated Maximize Conversions bidding strategy that uses real-time signals to prioritize spend on opportunities most likely to convert. - High-intent targeting that allows brands to reach users who have recently searched for specific financial terms across the open web. - Select, a brand suitability solution that ensures campaigns run exclusively on a curated, Made for Advertising (MFA)-free ecosystem of premium, high-tier publishers. - Gen-AI landing page tool (currently in Beta) that builds campaign-ready, localized advertorial landing pages designed to capture attention and drive conversions in specific markets. - Automated SpendGuard algorithm that improves efficiency by identifying and blocking under-performing sites in real time to minimize wasted budget. - Contextual targeting that places ads alongside relevant editorial content, such as personal finance or banking articles, without relying on cookies or personal identifiers. Best for: Realize is ideal for high-consideration financial services brands (such as insurance, banking, and mortgage lenders) that require a balance of scale and precision. It’s particularly helpful for organizations with limited internal AI support that need automated tools to manage the complexities of open-web prospecting, while ensuring their ads only appear in trusted, brand-safe environments. Pricing model: Performance-based model; campaigns billed on a CPC basis, or CPM for programmatic. Pros: - Leverages predictive algorithms to achieve, on average, 15% lower CPA and 50% higher conversion rates through automated bidding. - Provides robust brand safety through certifications and curated premium publisher lists, ensuring ads appear next to reputable financial news. - Audience targeting capabilities include retargeting segments, predictive segments, wide scale of Taboola 1P and 3P audiences, search keyword, and mail domain targeting to reach relevant finance-specific audiences. Cons: - Privacy regulations and compliance for finance advertisers can put constraints on pixel implementation, which is required across key steps in the advertiser's funnel to optimize campaigns. - High-performance features like Maximize Value or the Performance Simulator often require a learning phase and a baseline of conversion data to be effective. - Certain high-intent targeting tools, such as Mail Domain and Search Keyword targeting, are currently restricted to specific global markets. ### 2. Google Ads Why it’s essential: Google Ads remains one of the most important acquisition channels for financial services marketers because it captures users at the exact moment of intent. When someone searches for terms related to loans, insurance, investing, or credit products, they are often already deep in the decision-making process. This makes Google Ads particularly effective for bottom-funnel performance advertising where conversion likelihood is high. For financial brands, the platform serves as both a demand-capture engine and a testing ground for messaging. Advertisers can quickly validate offers, landing pages, and value propositions before expanding into broader channels, such as the open web. Showcased features: - Combines search, display, and video advertising within a single platform. - Uses machine learning optimization to improve campaign performance. - Adjust bids dynamically based on signals such as device type, time of day, user behavior, historical campaign performance, etc. - Offers advanced conversion tracking to measure results more accurately. - Enables detailed audience segmentation for better targeting. - Includes remarketing tools to re-engage potential customers over time, which is ideal for financial products with longer decision cycles. Best for: Google Ads is particularly effective for high-intent acquisition campaigns, such as personal loan applications, credit card sign-ups, insurance quotes, and refinancing inquiries. Financial companies with strong landing page experiences and compliance-ready messaging typically see the strongest results. Pricing model: Campaigns typically operate on CPC or CPA bidding, with Smart Bidding automating optimization toward defined conversion goals. Pros: - Extremely strong intent targeting driven by user search behavior, often producing high conversion rates for financial products. - Advanced machine learning bidding strategies continuously optimize toward conversion goals using large-scale data signals. - Deep integration with analytics, CRM platforms, and attribution tools enables sophisticated performance measurement. Cons: - Finance-related keywords are among the most expensive in digital advertising, increasing customer acquisition costs. - Heavy competition can make scaling difficult without strong landing page optimization and conversion infrastructure. - Increasing automation reduces manual transparency into bid decisions and optimization logic. ### 3. Microsoft Advertising Why it’s essential: Microsoft Advertising reaches a distinct audience segment that often differs from Google’s user base, including professionals and older demographics with higher average household income. For financial services advertisers, this can translate into higher-quality leads, particularly for investment and retirement-related products. Because competition tends to be lower than on Google, advertisers frequently find opportunities to achieve comparable results at reduced CPC levels. Showcased features: - Integrates with LinkedIn profile data for enhanced audience targeting. - Allows targeting based on company, industry, job function, etc. - Creates strong opportunities for promoting financial advisory services, institutional financial products, and B2B financial solutions. - Easy to adapt to the platform as campaign management tools closely resemble Google Ads. - Supports campaign imports from Google Ads for faster setup and scaling. - Adds extra demographic targeting layers to refine audience reach. Best for: Investment advisors, wealth management firms, retirement planners, and B2B financial services providers seeking professional audiences. Pricing model: Primarily CPC-based bidding with automated optimization options. Pros: - Typically has a lower CPC than Google Ads due to lower competition. - LinkedIn profile targeting enables precise segmentation by profession, industry, and company attributes. - Strong desktop usage and professional audience demographics often improve lead quality for financial services. Cons: - Smaller overall search volume limits large-scale campaign growth. Fewer third-party tools and ecosystem integrations compared to Google’s advertising stack. - Performance can vary significantly depending on geography and audience segment. ### 4. PropellerAds Why it’s essential: PropellerAds provides access to a large global inventory base, enabling financial advertisers to expand their reach beyond traditional search-driven environments. This is especially useful for campaigns targeting international markets or emerging financial audiences where search demand may be limited. The platform enables advertisers to test acquisition strategies quickly across multiple traffic sources and formats. Showcased features: - Supports multiple ad formats, including push notifications, popunder ads, native ads, and in-page ad formats. - Includes automated optimization tools. - Real-time reporting for ongoing performance tracking. - Advanced targeting options based on device type, geographic region, operating system, etc. - Automation tools support bid optimization and traffic filtering. - Campaigns automatically improve over time as performance data builds. Best for: Forex platforms, trading services, and global fintech campaigns focused on high-volume user acquisition. Pricing model: Self-serve campaigns operating on CPC or CPM pricing. Pros: - Access to large global traffic volumes allows for rapid campaign scaling and international expansion. - Multiple ad formats allow advertisers to test different acquisition strategies quickly. - Self-serve platform enables fast campaign launches and flexible optimization. Cons: - Traffic quality may vary widely depending on the targeting setup and optimization experience. - Some ad formats may require careful brand-safety considerations for regulated financial advertisers. - Requires active monitoring and optimization to maintain consistent lead quality. ### 5. ROIads Why it’s essential: ROIads focuses heavily on performance efficiency, positioning itself as a platform for direct-response marketers seeking granular control over campaign outcomes. Financial advertisers looking to optimize toward measurable conversions often use it to refine acquisition funnels. Its structure emphasizes optimization over reach alone, appealing to advertisers who prioritize return on investment (ROI) over brand exposure. Showcased features: - Supports push notification and pop traffic ad formats. - Uses AI-assisted bidding to help optimize campaign performance. - Offers micro-targeting capabilities for more precise audience reach. - Segments audiences based on browser, device, geography, and user environment. - Automated rules adjust bids using real-time performance signals. - Dedicated account support for campaign guidance. - Optimization recommendations to help improve results as performance data grows. Best for: Loan lead generation, fintech app installs, and conversion-driven acquisition campaigns. Pricing model: CPC or CPM campaigns with minimum deposit requirements. Pros: - AI-assisted bidding and micro-targeting provide strong control over conversion-focused campaigns. - Detailed targeting options allow advertisers to refine audiences at a granular level. - Dedicated account support can help optimize campaigns more efficiently. Cons: - Limited ad format variety compared to larger performance advertising platforms. - Minimum deposit requirements may create barriers for smaller advertisers. - Scaling potential may depend heavily on the availability of niche traffic. ### 6. Adsterra Why it’s essential: Adsterra provides a versatile advertising environment that combines global reach with multiple ad formats, allowing financial advertisers to diversify acquisition strategies. Its flexibility makes it useful for campaigns operating across multiple regions or audience segments. The platform is often used to complement primary acquisition channels by introducing incremental traffic sources. Showcased features: - Supports native, banner, video, and social-style ad formats. - Includes anti-fraud technology to help protect campaign quality. - Customizable targeting controls for audience refinement. - Reporting dashboards with performance insights. Best for: Financial content publishers, affiliate-driven finance campaigns, and international lead generation initiatives. Pricing model: Supports CPC, CPM, and CPA structures depending on campaign setup. Pros: - A wide variety of ad formats supports a diverse range of acquisition strategies. - Global reach enables advertisers to quickly expand into new markets. - Built-in anti-fraud technology helps improve traffic reliability. Cons: - Performance consistency may require ongoing manual optimization. - Some traffic sources may produce lower-quality leads without careful filtering. - Interface and reporting tools may feel less advanced compared to enterprise platforms. ### 7. RichAds Why it’s essential: RichAds positions itself as a performance-oriented advertising network designed for advertisers who want strong optimization controls and detailed campaign visibility. Financial marketers often use it for scalable testing environments focused on measurable outcomes. Its emphasis on filtering and performance analytics helps advertisers refine traffic sources over time. Showcased features: - Supports push, native, and pop ad formats. - Automated rules to help manage and optimize campaigns. - Micro-bidding capabilities for more precise bid control. - Offers a performance analytics dashboard for tracking results. - Adjust campaigns automatically based on conversion trends and traffic behavior. - Optimization filters help remove underperforming audience segments. - Prioritizes higher-value audiences to improve overall campaign performance. Best for: Credit offers, lending funnels, and investment signup campaigns focused on direct response results. Pricing model: CPC or CPM models with minimum funding thresholds. Pros: - Advanced optimization tools (such as automated rules and micro-bidding) enhance campaign control. - Detailed performance analytics help advertisers identify profitable traffic segments. - Designed specifically for performance marketers focused on measurable outcomes. Cons: - A steeper learning curve for advertisers unfamiliar with performance traffic networks. - Limited brand-building formats compared to native or premium publisher environments. - Requires sufficient testing budget to identify high-performing segments. ### 8. HilltopAds Why it’s essential: HilltopAds provides global advertising reach with a focus on audience segmentation and privacy-conscious distribution. Financial advertisers expanding into new markets often use it to access traffic beyond traditional Western advertising ecosystems. Its infrastructure supports campaigns targeting diverse geographic regions. Showcased features: - Supports multiple ad formats. - Includes anti-ad-block technology to maintain ad delivery in restricted environments. - Offers targeting based on geographic location and device type. - Supports international campaign expansion with global targeting. - Offers regular payouts for predictable payment cycles. Best for: Small to mid-sized financial campaigns targeting global audiences or emerging markets. Pricing model: CPC and CPM. Pros: - Global inventory allows advertisers to reach audiences in emerging and underserved markets. - Anti-ad-block technology helps maintain delivery rates across challenging environments. - Flexible pricing models make it accessible for smaller campaigns. Cons: - Fewer advanced automation and AI optimization features compared to larger platforms. - Limited premium publisher inventory may affect brand perception for financial advertisers. - Campaign performance may require more manual oversight and testing. ## More About Performance Advertising for Financial Services ### Top Ad Networks for Financial Services Lead Generation To effectively generate financial services leads, you’ll need to work with advertising platforms that can balance reach with accurate targeting. Unlike e-commerce ads, financial decisions usually involve more research and longer decision timelines. Platforms that combine contextual targeting, intent signals, and predictive optimization tend to perform best because they reach users while they are actively considering options, not just casually browsing. Open web advertising platforms are increasingly used alongside search campaigns, helping advertisers reach users earlier in their research journey. This expands customer acquisition opportunities beyond traditional keyword-based advertising. ### Which Advertising Channels Have the Best ROI for Finance Companies? Search advertising has traditionally delivered strong ROI because users show clear intent when searching for financial products. However, growing competition has driven up acquisition costs. Native advertising and open web placements often offer more efficient scaling because they create new demand, instead of competing only for existing search traffic. The most successful financial advertisers use a mix of channels, combining search campaigns to capture intent with open web campaigns focused on discovery and ongoing engagement. ### Compliance Rules for Advertising Financial Products Financial advertising must follow strict regulations that vary by region. Advertisers need to ensure disclosures are clear, claims are properly supported, and targeting methods follow privacy laws. Platforms with contextual targeting and brand safety controls help reduce compliance risks by placing ads in trusted environments. Clear messaging and transparent landing pages are important not only for compliance, but for building and maintaining user trust. ### Key Performance Indicators for Financial Services Advertising Campaigns Unlike retail campaigns that focus on immediate purchases, financial marketers often track results further down the customer journey. Metrics such as cost per qualified lead, approval rate, customer lifetime value (CLV), and funded account conversions are usually more important than total lead volume. Accurate conversion tracking, often supported by CRM integrations, is essential for optimizing return on ad spend. ### How to Target High-Net-Worth Individuals With Digital Ads Reaching affluent audiences requires a mix of contextual placement and behavioral signals. Advertising within premium publisher environments, investment-focused content, and professional audience segments often performs better than relying only on demographic assumptions. Placing ads alongside financial news or investment analysis content can signal strong intent without using personal identifiers. ### Cost of Advertising on Google Ads for Finance Keywords Finance keywords are consistently among the most expensive in paid search due to intense competition and high CLV. Keywords related to mortgages, insurance, and credit products often have very high CPC rates. Because of this, strong conversion rate optimization and high-quality landing pages are essential for maintaining profitability. ## Key Takeaways Performance advertising in financial services requires both precision and flexibility. No single platform solves every customer acquisition challenge. Search platforms capture intent, open web solutions expand reach, and performance networks provide scalable environments for testing campaigns. The most effective strategy combines multiple platforms aligned with specific campaign goals. Advertisers that diversify beyond traditional channels while maintaining strong performance measurement systems are better positioned to control acquisition costs and scale sustainably. ## Frequently Asked Questions (FAQs) ### What are the best channels to reach high-intent finance prospects beyond search and social? Open web advertising platforms, contextual native placements, and premium publisher networks are increasingly effective for reaching users actively researching financial decisions outside traditional search environments. ### How can AI and first-party data improve conversion for financial services campaigns? AI systems analyze behavioral patterns and optimize bidding in real time, while first-party data improves targeting accuracy and measurement reliability. Together, they enable more efficient budget allocation and stronger conversion rates than traditional audience-targeting methods. ### What are the best advertising platforms for a new mortgage broker? New mortgage brokers often benefit from starting with intent-driven platforms, such as search advertising, while gradually expanding into open web performance platforms that can scale lead generation once conversion tracking and messaging are optimized. --- ### 3 Ways to Master Creative Variation at Scale Using Realize URL: https://www.taboola.com/marketing-hub/creative-variation-at-scale/ Last Modified: 2026-06-10 10:21:15 Your competitors aren’t uploading ad creatives one by one. They’re deploying hundreds of variations across campaigns in minutes, testing new angles before lunchtime, and optimizing by the end of the day. If your team is still building ads manually, you’re not just wasting time, you’re falling behind. The Realize platform gives performance advertisers tools designed to close that gap. Built for mass ad customization and efficient user interface (UI) campaign management, it lets you launch, modify, and refresh massive volumes of creative without the busywork that slows everything down. Whether you’re an agency managing multiple accounts or a small- or medium-sized business (SMB) ready to scale creative variations, these three workflows will transform the way you do business. ## Creative Variations At Scale Using Realize: 3 Successful Strategies ### 1. Launching Mass Variations via Creative Bulk Upload The more campaigns you run, the more creative variations you need, and the less sense it makes to build each one by hand. That’s the logic behind the creative bulk upload CSV feature in Realize. Instead of entering ads individually, you prepare everything in one structured file and push the entire batch live at once. Here’s how it works. On the Create Ads page in your Realize dashboard, you’ll select the “Bulk Upload” option. From there, you choose which campaigns should receive the new creatives — and yes, you can select multiple campaigns at once. Then you upload the CSV file, which maps out the essential components of each ad: - Landing page URL: The destination page where each ad will drive traffic. - Title: The headline that appears alongside your creative. - Ad image: The thumbnail or visual asset tied to each variation. Once you upload the file, Realize automatically populates your campaign with the ads, URLs, and titles from your files. You review the results and hit submit. That’s it. Dozens or even thousands of variations are distributed across your campaigns in a single action. What makes this especially powerful is the flexibility of the template. Descriptions and call-to-action (CTA) buttons can be included in the creative bulk upload CSV for stronger engagement, but they aren’t required fields. This means you can get creatives live quickly, then circle back to optimize with additional copy elements later. For advertisers who need to test different headline angles, swap out landing pages for regional campaigns, or simply get a high volume of creatives into rotation, this feature turns what used to be hours of manual entry into a five-minute workflow. And, because you’re working from a structured template, the risk of human error — mismatched URLs, duplicated titles, or missing images — drops significantly. ### 2. Updating Live Ads with Excel Bulk Operations Launching creatives at scale is one thing. Keeping them updated is another challenge entirely. Campaign priorities shift, landing pages get refreshed, and seasonal messaging needs to rotate in and out. Without a system for making Ad ID bulk edits across your account, even small changes can eat up an entire afternoon. That’s where Realize bulk operations come in. This feature uses an Excel template that lets you modify existing campaigns and ads without rebuilding anything from scratch. The process starts from the same Bulk Upload section of Realize, where you can either download a blank template or export your current campaign data into a pre-filled spreadsheet. The real efficiency gain here is how updates work. When you’re modifying existing creatives, you don’t need to fill in every field. The template is designed so that: - For campaign updates, only the Campaign ID is required, alongside whatever field you want to change. - For ad-level changes, only the Ad ID is needed, plus the specific variable you’re updating. - For all other fields, no input is needed, since Realize will keep the existing values intact. This means if you need to swap out a landing page URL across 300 ads, you simply populate the Ad ID column and the new URL column. The same logic applies if you’re pausing a bunch of underperforming creatives, updating tracking codes, or changing ad titles for a new promotion. You can update multiple campaigns in Taboola’s Realize with a single file upload, rather than clicking through each one individually. Once you upload the completed Excel template, Realize processes the changes and generates a processing report. This step is critical and shouldn’t be skipped. The report flags any discrepancies — mismatched IDs, formatting errors, unsupported values — so you can catch and fix issues before they affect live campaigns. For advertisers who also need to update media assets in bulk, Realize supports uploading a ZIP file alongside your Excel sheet. The file names in the ZIP must match the references in your spreadsheet exactly, and the platform will link them automatically during processing. The Excel template ad updates workflow is especially valuable for agencies and larger advertisers who manage heavy creative rotations across campaigns, regions, or product lines within a single account. Instead of training team members on complex processes, you can standardize your updates in a single spreadsheet format that anyone can use. ### 3. Automating Continuous Variation with RSS Feeds CSV uploads and Excel templates are powerful for batch operations, but what about campaigns that need a constant stream of fresh content? If you’re an e-commerce brand adding new products daily, or a publisher promoting new articles every few hours, manually uploading each new item isn’t realistic. This is where RSS feed integration becomes your biggest time-saver. When you automate campaign items with RSS, every new page or article added to your feed is automatically converted into an active campaign item inside Realize. No manual uploads, no CSV formatting, no spreadsheet wrangling — just a continuous flow of fresh creatives pulled directly from your content. Here’s what makes that approach so effective: - New content goes live automatically: As soon as a new item appears in your RSS feed, it shows up in your Realize campaign without any action on your part. - Outdated content removes itself: If an item drops out of your feed — say, a product goes out of stock, or an article is archived — it’s removed from your campaign automatically. - Items stay active until the feed changes: Creatives synced via RSS continue running until you manually pause them, or they fall out of the feed. For best results, Realize recommends setting your RSS lookback period as far back as you can. Realize’s recommended best practice is to keep as many active items as possible running in your campaign, which gives the algorithm more creative variations to optimize against. More variations mean more data points, and more data points mean better performance over time. This hands-off approach is especially well-suited for content-driven advertisers — think blogs, news sites, affiliate publishers, or any brand with a frequently updated product catalog. Rather than treating creative uploads as a recurring task on your to-do list, RSS automation lets you set up the system once so that your content pipeline can do the rest. It’s worth noting that RSS feeds work best when your content is structured consistently. Each feed item needs a clear title, URL, and image for Realize to build an effective ad unit. If your feed is well-formatted, the transition from content publication to live ad is seamless. ## Key Takeaways Scaling creative variations doesn’t have to mean scaling your workload. By using the right combination of tools within the Realize platform, advertisers can move faster, test more, and optimize without burning hours on manual processes. Creative bulk upload CSV lets you launch thousands of ad variations across multiple campaigns in a single upload, eliminating one-by-one data entry. Realize bulk operations via Excel templates allow you to make targeted updates to live ads using only an Ad ID or Campaign ID, with no need to re-enter unchanged fields. RSS feed automation keeps campaigns stocked with fresh content automatically, syncing new pages and products directly into your campaigns in real time. Together, these three workflows give performance advertisers the speed and agility they need to stay competitive. Instead of spending time on logistics, you can focus on what actually moves the needle: strategy, creative testing, and results. ## Frequently Asked Questions (FAQs) ### Can I update existing campaigns using the Realize bulk operations feature? Yes. When you download the bulk operations Excel template, you can modify both campaigns and individual ads by entering the relevant Campaign ID or Ad ID alongside the specific field you want to change. Any fields that don’t need updating can simply be left blank, and Realize will preserve the existing values. After uploading your changes, always review the processing report to confirm everything was applied correctly. ### Do I have to include descriptions and CTAs in my bulk upload template? No. The creative bulk upload CSV template requires landing page URLs, titles, and images to process your ads, but descriptions and CTA buttons are optional fields. Including them can improve engagement and click-through rates, but they aren’t mandatory to get your creatives live. You can always add them later through individual edits or a follow-up bulk update. ### What is the fastest way to add new creatives automatically in Realize? Linking an RSS feed to your campaign is the most hands-off method available. Once the feed is connected, any new pages or articles added to it are automatically converted into active campaign items — no manual uploads required. The campaign dynamically reflects whatever is currently in your feed, so content that’s removed from the feed is also removed from the campaign. This makes RSS integration ideal for advertisers who publish new content frequently and want to keep their campaigns fresh without ongoing manual effort. --- ### Escaping the Black Box: How to Regain Control of Your AI Campaigns URL: https://www.taboola.com/marketing-hub/agentic-ai-campaigns-control/ Last Modified: 2026-07-06 08:43:46 AI has transformed campaign management, but speed and scale come with a catch. Hand the keys to a black-box AI solution, and you risk off-brand messaging, wasted spend, and audiences optimized for entirely the wrong signals. Currently, 60% of organizations have no way to terminate a misbehaving AI agent, and 63% can’t enforce limits on what those agents are authorized to do. For performance marketers, it's a cost per action (CPA) and return on ad spend (ROAS) problem.  Scaling fast only amplifies the damage. This guide shows where control breaks down, how to regain control of your AI campaigns, how to build the right guardrails, and what transparent AI campaign management actually looks like in practice. ## The Hidden Dangers of a Black-Box AI Solution AI-powered campaign management delivers real efficiency gains, but when the system making decisions is completely opaque, efficiency can quickly become a liability. A black-box AI solution is one where inputs go in, outputs come out, and the logic in between is invisible to you. For marketers, that means campaigns running on assumptions you can’t audit, creative optimizing toward signals you didn’t approve, and budget allocating itself based on criteria you’ll never see. The algorithm is working, just not necessarily for you. Three failure patterns show up consistently: ### Generic, off-brand output Without visibility into how content decisions are made, messaging drifts toward whatever the system has learned performs broadly. ### Disconnected customer journeys Black-box systems optimize individual touchpoints in isolation. The result is a funnel that converts on paper, but fragments the actual customer experience. ### Misaligned optimization targets Closed algorithms are built to maximize platform-level performance metrics. That’s not always the same as your CPA, your ROAS, or your long-term customer value. The deeper risk is structural. When you can’t see how decisions are made, you can’t course-correct before damage is done, and by the time poor performance shows up in your reporting, the budget is already spent. ## Reclaiming Your Data and Strategic Narrative Regaining control starts before the algorithm ever runs. It starts with who owns the inputs. Most black-box systems are built around your data, but processed in ways you can’t access or interrogate. Feed first-party data into a walled-garden platform without a clear strategic framework, and you’ve handed over more than targeting parameters — you’ve handed over your brand positioning. The fix isn’t to use less AI, but rather, to walk in with a stronger brief: ### Own your audience architecture Define segments from your own customer data, not platform-inferred proxies. ### Set messaging boundaries upfront Brand voice, claim hierarchy, and call-to-action (CTA) structure shouldn’t be left to algorithmic inference. ### Control the context Funnel stage, intent signals, and seasonal relevance all shape how the system optimizes. Define them, or the platform will. When the inputs are yours, the outputs become accountable. AI becomes a tool that executes your strategy, instead of replacing it. ## Establishing Crucial Guardrails for Campaign Safety Giving AI more autonomy without clear boundaries isn’t scaling, it’s gambling. AI marketing guardrails are what separate a system that works for your brand from one that works against it. Every AI campaign needs three categories of guardrail in place: ### Brand Compliance Rules Define what the AI can and cannot say. Documented tone guidelines, approved claims, restricted messaging territories, and creative boundaries it cannot cross regardless of predicted performance. AI brand compliance isn’t a creative preference, it’s a risk management function. ### Human-In-The-Loop AI Workflows and Checkpoints Not every decision needs human approval, but the high-stakes ones do. Budget reallocation above a set threshold, new audience segments, landing page variations: build mandatory review points into these moments before the system acts. That friction is the point. ### Hard Action Limits Spending caps, bid ceilings, audience exclusion lists, frequency limits. These are the non-negotiables that prevent runaway optimization from burning budget or damaging audience relationships. Set them before launch, not after something goes wrong. Remember, guardrails don’t limit what AI can do: they define the conditions under which it’s trusted to operate. ## Enter the Transparent Agent: Steering With Realize+ The alternative to a black-box AI solution is better visibility into how that automation makes decisions. Realize+ (in BETA) is built around the concept of algorithmic transparency, something that’s increasingly called a “glass-box” approach to AI campaign management. Every action the system takes is auditable, every optimization decision is explainable, and every output is traceable back to the rules you set. https://www.youtube.com/watch?v=QUYxkx8KFfg In practice, that means: ### Full Decision Visibility Realize+ surfaces why budget shifted, why a creative was prioritized, and why an audience segment was expanded, in plain language, not platform black-box logic. ### Guardrail-Native Architecture The guardrails you establish aren’t a layer on top of the system, they’re built into how it operates. Brand compliance rules, spending limits, and audience parameters are enforced at the decision level, not reviewed after the fact. ### Transparent Quality assurance built in Every campaign output is checked against your brand and compliance rules before it goes live, no manual review layer required. ### Explainable AI marketing in action Every campaign action generates a clear rationale. Teams can audit decisions, identify drift early, and course-correct without having to reverse-engineer what the algorithm did. With Realize+, AI doesn’t replace your judgment, it executes within it, with full transparency into every step it takes to get there. ## Key Takeaways AI campaign management is only as effective as the control structure around it. A black-box AI solution optimizes for outputs you can see but decisions you can’t, and that gap is where brand integrity and budget efficiency get lost. Regaining control starts with owning your inputs: first-party data, audience architecture, and messaging boundaries defined before the algorithm runs. Always remember, AI marketing guardrails aren’t optional safeguards, and spending caps, brand compliance rules, and human-in-the-loop checkpoints are structural requirements for running AI campaigns safely at scale. Transparent agent AI changes the equation. When every decision is auditable and every action is explainable, marketers stop reacting to what the algorithm did and start directing what it does next. Realize+ is built for this. It gives performance marketers the visibility, guardrails, and explainable AI marketing framework needed to scale campaigns without ceding control. The goal isn’t to unplug the AI, but to make sure you’re always the one holding the steering wheel. ## Frequently Asked Questions (FAQs) ### Why is a black-box AI solution dangerous for marketing? When you can’t see why the algorithm made a decision, you can’t catch off-brand messaging before it runs, identify why spend is being misallocated, or course-correct before the damage shows up in reporting. Closed systems also tend to optimize for platform-level metrics that don’t align with your CPA targets or brand objectives. Lack of visibility isn’t just a transparency issue, it’s a budget and compliance risk. ### What are AI guardrails in marketing campaigns? Guardrails are predefined rules that keep your AI system aligned with your business goals. In practice, they cover three areas: brand compliance rules that define what the AI can and cannot say; human-in-the-loop checkpoints that require approval before high-stakes actions; and hard limits on spending, bidding, and audience targeting. Together, they prevent the system from optimizing in directions you never approved. ### What is a transparent AI agent? A transparent agent is an AI system built on explainable AI principles. Unlike opaque models, it shows its reasoning, so marketers can trace every optimization decision back to the data and rules that drove it. Instead of asking, “Why did the algorithm do that?” you already know. That visibility is what makes AI campaign management auditable, adjustable, and safe to scale. ### How does Realize+ help regain control of campaigns? Realize+ replaces black-box logic with full decision visibility. Every optimization action is explainable, every campaign decision is traceable, and the guardrails you set are enforced at the system level, not reviewed after the fact. Marketers stay in control of strategic direction while Realize+ handles execution within the boundaries they define. It’s the difference between handing your campaigns to an autopilot and having a system that shows its work. --- ### 7 Campaign Scheduling Best Practices for Performance Advertisers URL: https://www.taboola.com/marketing-hub/campaign-scheduling-best-practices/ Last Modified: 2026-06-10 08:36:55 When it comes to digital advertising, timing is crucial to your results. Showing ads to the right people at the right moment is what separates campaigns that convert from ones that waste budgets. That’s where campaign scheduling best practices come in. By taking control of when your ads run, you can reach your audience during their highest-intent moments, pull back spend during low-performing windows, and get more from every dollar you invest. ## What Is Campaign Scheduling? Campaign scheduling, also known as dayparting, is a feature that gives advertisers greater control over their ad delivery schedule. Rather than running ads continuously, you define specific days and hours during which your ads are eligible to appear. This allows you to focus your budget and bids on the moments that matter most to your audience. ## Best Practices for Performance Campaign Scheduling ### 1. Start Broad to Gather Initial Data Before making any scheduling decisions, you need data. The best way to get it is to run your new campaigns 24 hours a day, seven days a week, for the first seven to 10 days after launch. This initial phase gives you a baseline across all hours and days of the week, so you can see when users naturally engage with your ads before you start restricting delivery. Cutting hours too soon means you could be removing windows that actually perform well; they simply don't have enough data to prove it yet. Patience during this phase pays off significantly when it comes time to optimize your campaigns in subsequent weeks. ### 2. Pinpoint Peak Performance Times Once your campaign has run for at least a week, look at your performance reports to identify peak performance windows. You’re looking for specific hours of the day that consistently drive your lowest cost per action (CPA) and highest conversion rate (CVR). These peak performance times are the base for an effective and efficient ad schedule. Look for patterns: maybe Tuesday mornings and Thursday evenings convert at twice the rate of other times. Those are the windows to prioritize. Most advertising platforms provide hourly and daily breakdowns that make this analysis easier. It’s also worth tracking these windows over time, since peak hours can shift with seasons, promotions, or changes in audience behavior. Building a habit of regular reporting reviews keeps your schedule aligned with current performance, rather than assumptions based on older data. ### 3. Align Schedules with Customer Habits Data only tells part of the story, and understanding your audience's daily routines tells the rest. The goal is to synchronize your campaign schedule with target audience behavior so your ads appear when intent is highest. For instance, if you’re running a subscription service and you notice that sign-ups spike on weekends, that is a clear signal to concentrate your budget on Saturdays and Sundays. If you work in B2B, your audience is likely most engaged during business hours on weekdays. Mapping your schedule to real user behavior is one of the best ways to optimize ad campaign performance. ### 4. Duplicate Campaigns to Test Schedules One of the best ways to test whether a scheduling change actually improves results is to duplicate your campaign and run both versions simultaneously with different schedules. This A/B testing approach keeps your data clean because you aren’t changing variables mid-flight on a single campaign. For example, run Campaign A on a broad schedule and Campaign B restricted to your identified peak hours. After a statistically significant period, compare the cost per action, conversion rate, and return on ad spend (ROAS) for each. This process is central to performance campaign optimization and helps you make scheduling decisions based on evidence, rather than guessing. Make sure the two campaigns don’t target overlapping audiences in a way that causes them to compete against each other, as this will skew your results. ### 5. Adjust Budgets and CPCs for Specific Timeframes Not all hours of the day deserve equal investment. Once you know which windows drive the best results, use bid and budget adjustments to optimize performance in your highest possible conversion window. Set a higher cost per click (CPC) during your prime converting hours to stay competitive and protect your placement when it counts. During lower-performing hours when you still want to maintain some presence, lower your CPC to keep spend minimal and efficient, without going dark. This tiered bidding approach is a core part of any solid dayparting strategy, and it ensures your budget is weighted toward the moments with the highest return potential, rather than spread evenly across hours that may not convert as well. ### 6. Optimize Time Zones for Your Audience When you set a campaign to run from 9 a.m. to 5 p.m., that schedule is only effective if “9 a.m.” reflects your audience’s local time zone, not the default time zone in your ad account. If your campaign is targeting users in multiple regions, misaligned time zone settings can mean your ads run at 3 a.m. local time when you thought they were running during the workday. Before launching any scheduled campaign, check your account’s default time zone setting and confirm it matches your target market. For native advertising campaigns running across multiple geographies, you may need separate campaigns per region to handle this accurately. ### 7. Use Platform Shortcuts for Efficiency Setting up schedules manually hour by hour can take some time, especially when you’re managing multiple campaigns. Realize campaign setup tools include several shortcuts designed to save time without sacrificing precision. Look for preset options like “Every Day,” which applies the same hourly schedule across all seven days at once, and “Weekdays,” which limits delivery to Monday through Friday with a single click. When you have a specific time block you want to replicate, the “Copy timeframe to all” feature applies it across each day instantly, rather than requiring you to set each day individually. Taking advantage of these shortcuts keeps your setup efficient and reduces the chance of manual errors when building complex schedules. ## Key Takeaways Strategic campaign scheduling is not just a tactical detail, but one of the most direct options you have for improving ad performance. By starting with broad data collection, identifying your highest-converting windows, aligning your schedule with how your audience actually behaves, testing through duplication, adjusting bids to match value, verifying time zone settings, and using platform tools efficiently, you turn your ad delivery schedule into a successful performance engine. The result is a campaign that works harder during the moments that matter and conserves budget when it doesn’t. Start with one or two of these practices, measure the impact, and build from there. ## Frequently Asked Questions (FAQs) ### What is dayparting in advertising? Dayparting is the practice of scheduling your ads to run only during specific hours or days of the week, so your budget is focused on the times your audience is most likely to engage or convert. ### How long should I run a campaign before adjusting the schedule? Run your campaign on a 24/7 schedule for at least seven to 10 days first, so you have enough data across all times of day to make informed decisions about which hours to prioritize. ### Does campaign scheduling work for all campaign types? Yes, scheduling can benefit most campaign types, though the optimal windows will vary depending on your industry, audience, and campaign goal. ### Can I run different schedules for different audiences? Duplicating campaigns and assigning different schedules is a recommended approach for testing and for targeting audiences in different time zones, or with different behavioral patterns. ### How often should I revisit my campaign schedule? Review your schedule at least once a month, or whenever you see a significant shift in performance metrics, since audience habits and competitive dynamics can change over time. --- ### AI Guardrails in Agentic Advertising: How Agencies Retain Human Control and Mitigate Risk URL: https://www.taboola.com/marketing-hub/agentic-advertising-ai-guardrails/ Last Modified: 2026-06-10 08:09:18 The world of digital ads has been turned upside down by agentic AI — a technology that doesn’t just suggest a strategy, but plans and executes your entire media buy while you focus on something else. But, for agency buyers, it’s more complex than that. After watching unchecked bots set fire to budgets on low-quality clicks, or turn a sophisticated brand into “AI slop” — a catch-all term for generic, soulless content that feels like a robot wrote it (because it did) — some marketers are concerned that this level of autonomy can be a liability. To keep things from going off the rails, the smartest players are setting up AI guardrails. By disabling tools like auto-generated creative, they’re making sure humans stay in the driver’s seat for the stuff that actually matters: strategy, voice, and landing page quality. ## What Are AI Guardrails in Agentic Advertising? In advertising, there needs to be oversight when moving from AI that helps to AI that acts. We’ve quickly shifted from AI that can write catchy headlines to agentic AI that decides which ads to run, adjusts your bids at 3 a.m., and moves your budget from one platform to another without asking for permission. While that speed is incredible, on less transparent platforms, it creates a black box where no one knows why, e.g., the bot just moved $20,000 to a niche gaming page. This is where AI guardrails in agentic advertising come in. They’re the technical stay-in-your-lane rules and permissions that ensure the bot operates within your brand’s safety zone. ## The Agency Dilemma: Skepticism Over the Black Box Advertising agency AI skepticism around these black box solutions isn’t just about being against new technology — it’s about recognizing opaque decision-making processes and not wanting to get fired for a bot’s bad decision. For an agency, a campaign that looks busy but fails to drive revenue is a disaster. Without human oversight, you could be looking at a serious AI budget waste problem, which is why we’re seeing the move toward human-in-the-loop ad optimization. Let the machines crunch the data, but let the humans set the rules. As incredible as AI is, it still benefits from human-set limits. ## The Risk of AI Slop: Why Brands Are Losing Trust One of the biggest autonomous media buying risks is the rise of AI slop. It’s bland, mass-produced content people can now spot a mile away, and it’s an instant turn-off. Audiences are getting really good at identifying these ads, which is a problem, because fully AI-generated ads often fail to stick in someone’s memory, actually requiring higher quality standards to even get noticed. ## Applying Guardrails: Retaining Human Control Over Ad Copy and Landing Pages Some veteran ad tech advocates argue that to unlock the true power of automation, you have to hand over the keys completely and let the machine write the copy, build the images, and dynamically change the landing pages. In reality, that can be a recipe for a brand-safety disaster. The winning move for modern agencies is highly specific: disabling AI creative generation entirely, while letting the analytical agent handle the heavy lifting of real-time bidding, budget pacing, and platform optimization. When configuring an enterprise agentic setup, agencies should disable options like: - Automated Copy Expansion: Features that allow the machine to rewrite human headlines into “optimized” robot variations. - Dynamic Asset Bundling: Systems that independently slice and dice images or overlay generic graphic design elements without design review. - Generative Landing Page Variations: Automation modules that alter layout copy on the fly to match a user’s perceived profile. By keeping these generative tools switched off, you firmly focus on retaining human control in AI marketing. This approach ensures that your brand voice stays perfectly intact, your landing pages remain polished, and your messaging retains that genuine, empathetic human spark that LLMs can struggle to simulate. ## Designing a Decision Architecture for Media Buying Safely handing the steering wheel over to an AI agent means moving past a lazy on/off mentality and actually mapping out who does what. You can’t just flip a switch and hope for the best: instead, marketing teams need to clearly define which tasks are fully automated, which ones require a co-pilot setup, and which ones absolutely demand a human signing off in ink. Think of it as a five-tier spectrum of machine control: you've got Level 1 (old-school manual grinding), Level 2 (human-led insights), Level 3 (human-in-the-loop co-piloting), Level 4 (human-in-the-loop active monitoring), and Level 5 (total machine autonomy). If you’re a performance agency looking to scale aggressively without accidentally bankrupting your clients, the absolute sweet spot is a mix of augmented intelligence and strict veto power — specifically, Levels 3 and 4. ## Essential AI Guardrails Every Advertising Agency Needs ### 1. Data and Input Guardrails Your AI agent is only as sharp as the data you feed it. If you want your optimization paths to actually make sense, you have to plug the system directly into clean, first-party CRM and server-to-server conversion data. Feed it trash, and you’ll get trash in return. If your bot starts making bidding decisions based on loose, guessed pixels or surface-level vanity metrics, it will slide right into algorithmic drift, meaning it will start hunting down cheap, useless audience segments that look fantastic on a slide deck, but do absolutely nothing for your actual revenue. ### 2. Budget and Bidding Caps Never let an autonomous bot out of its cage without an absolute financial ceiling. You need to hardcode unbreachable programmatic limits directly into your platform, to cap exactly how much an agent can spike a bid or shift an overall budget during a single campaign flight. Think of these caps as a massive circuit breaker: they prevent a sudden data glitch or tracking error from sending the algorithm into a runaway doom-loop that burns through a client’s entire monthly budget while your team is offline for the weekend. ### 3. Human-in-the-Loop (HITL) Approvals HITL architecture is your ultimate insurance policy. This rule states that even if the AI can crunch a billion data points in milliseconds and queue up the perfect campaign adjustment, it’s still structurally blocked from pushing that change live until a real, breathing human reviews and clicks “approve.” Gating high-stakes inflection points — like launching brand-new audience definitions, entering unverified markets, or scaling up massive ad lines — is how you keep automated mistakes from tanking your brand’s reputation. ## The Future of AI in Advertising: Balancing Automation and Trust As the advertising space evolves, transparent AI governance is fast becoming an agency’s single biggest competitive advantage. At the same time, the initial market rush, where tech platforms sold buyers on the magic of an unmonitored black-box AI solution, is rapidly fading away. Smart brand leaders are demanding to see exactly where their ad dollars land, and precisely how their agencies manage automated risks. Agencies that step up and show off their custom guardrail architectures during pitches are winning the highest-value accounts. Instead of pitching a black-box engine and asking clients for blind trust, modern agencies should pitch cutting-edge tools paired with rigorous human oversight. This communicates to clients that you’re leveraging the hyper-efficiency and immense scale of the open web to truly maximize return on investment, without ever exposing their brand to the unpredictable pitfalls of unchecked automation. ## Key Takeaways Maximizing your performance on the modern open web doesn’t mean you need to hide from automation, it just means you have to stop letting the machine drive without a license. The winning play is all about guiding that AI horsepower through strict, real-time boundaries. By locking down tight AI guardrails — like stripping the bot of its creative privileges, setting unbreachable budget caps, and keeping your thumb firmly on the veto button — agencies can completely wipe out manual grunt work. You’ll get all the fast scaling speed of a machine, while ensuring your client campaigns stay insulated from algorithmic drift, middle-of-the-night budget fires, and brand-killing slop. ## Frequently Asked Questions (FAQs) ### What is agentic AI in advertising? Agentic AI refers to highly autonomous software systems that go beyond just basic data analysis or text generation to actively plan and execute tactical media buying maneuvers. Working within predefined rules, you can use API connections to independently manage bids, adjust budgets, and reallocate campaign resources across networks in real-time. ### Why are agency buyers skeptical of unchecked AI? Media buyers are wary of fully autonomous systems because they often operate as a non-transparent black box, chasing surface-level vanity metrics or cheap, low-intent clicks to hit mathematical optimization goals. Without the proper oversight, unchecked automation can result in significant ad budget waste and produce generic, uninspired “AI slop” that actively harms client brand trust. ### Why do agencies disable AI-driven creative generation? When you turn on fully autonomous generative ad features, you’re essentially letting a robot write your brand story. The result is often a wave of generic, soulless copy and imagery that audiences can spot from a mile away and instantly tune out. Worse, if an algorithm decides to slightly alter a headline to chase a cheap conversion goal, it might accidentally violate a strict legal compliance rule, use an unapproved phrase, or ruin a brand’s hard-earned voice. Agencies switch this feature off to retain strict human control over the ad copy and landing pages. ### What is a Human-in-the-Loop (HITL) guardrail? Think of a Human-in-the-Loop guardrail as your ultimate campaign insurance policy and the big red “don't break things” button. It’s an operational framework where the machine does 99% of the exhausting grunt work — processing data, tracking down hidden trends, and queuing up optimizations — but is structurally blocked from pushing those changes live until an actual human media buyer looks at it and clicks approve. --- ### The 7 Audience Targeting Platforms Every Performance Advertiser Should Know Of URL: https://www.taboola.com/marketing-hub/audience-targeting-performance-platforms/ Last Modified: 2026-06-18 06:48:43 Reaching the right person matters more than reaching more people. That’s always been true in advertising, but it’s become the defining challenge of 2026. Privacy regulations have tightened. Third-party cookies are effectively gone. iOS attribution changes have punched holes in what used to be reliable data. Audiences that were easy to find on social platforms two years ago are increasingly fragmented across the open web, streaming platforms, and niche publisher environments. The good news is that the platforms in this space have responded. AI-driven audience modeling, contextual targeting, and first-party data activation have all matured significantly. Still, “audience targeting platform” now covers a lot of ground, from social walled gardens to open-web demand-side platforms (DSPs) to programmatic specialists, and picking the wrong one for your use case means wasted budget, not just suboptimal targeting. This guide covers seven of the best audience targeting platforms. You’ll see what makes each one worth using, where each one falls short, and which use cases each one is best for. ## 7 Best Audience Targeting Platforms for Performance Campaigns at a Glance Platform Why It’s Essential Core Use Cases and Features Best for Pricing Model 1. Realize One of the most powerful open-web performance targeting platforms, using AI and predictive audiences across thousands of publisher sites. Predictive audience modeling, native ad targeting, contextual signals, first-party data integration, large publisher network distribution. Performance marketers, ecommerce brands, and agencies wanting open-web user acquisition beyond walled gardens. Performance-based model; campaigns billed on CPC basis, or CPM for programmatic. 2. Google Ads Dominates intent-based targeting with search data and massive reach across search, YouTube, and the display network. Keyword targeting, in-market audiences, remarketing lists, YouTube ads, smart bidding, conversion optimization. Businesses targeting users actively searching for products/services. CPC/CPA/ CPM auction bidding. 3. Meta Ads Manager One of the most advanced social audience targeting systems, with billions of users and deep behavioral signals. Interest targeting, lookalike audiences, Advantage+ AI optimization, retargeting, cross-platform ads (Facebook and Instagram). Consumer brands, e-commerce, mobile apps. CPC/CPM auction model. 4. LinkedIn Ads The strongest B2B audience targeting platform because of professional identity data. Targeting by job title, company size, industry, skills, seniority, account-based marketing, lead gen forms. B2B companies, SaaS, recruiting campaigns. CPC/CPM (premium pricing). 5. The Trade Desk A leading programmatic DSP enabling data-driven targeting across web, CTV, audio, and mobile. AI bidding (Koa), cross-device identity graph, Connected TV (CTV) targeting, data marketplace integrations. Enterprise advertisers and large agencies. CPM programmatic bidding. 6. Amazon DSP Leverages Amazon shopping behavior data to target consumers close to purchase. Retail intent targeting, audience retargeting, off-Amazon display ads, streaming TV placements. E-commerce brands and retail advertisers. CPM programmatic model. 7. StackAdapt Known for strong intent-based and contextual targeting using AI for programmatic advertising. Native ads, contextual targeting, audience intent signals, omnichannel campaigns (display, video, CTV). B2B marketers, mid-market brands, agencies. CPM/CPC depending on format. ### 1. Realize Why it’s essential: Realize is an AI-powered performance platform designed to help advertisers find and engage their ideal customers across the open web, using proprietary data signals. It moves beyond basic demographic targeting by leveraging deep learning to analyze user behavior, content consumption, and conversion patterns. In practice, Realize is used to deploy a privacy-first targeting strategy by combining contextual relevance with predictive modeling. Advertisers can choose to target specific interests or leverage the platform’s AI to find users who act like buyers, regardless of their historical profile. This approach ensures that your message reaches the right person at the moment they’re most receptive to your offer. Showcased features: - Predictive Targeting: Uses machine learning to identify and reach users most likely to convert, based on real-time behavioral signals rather than static profiles. - Contextual Segments: Matches your ads to specific article topics and editorial environments, ensuring your audience is reached while they are in a relevant mindset. - Lookalike Modeling: Analyzes your existing customer data to find “seed-based” audiences on the open web that share the same characteristics as your best-performing users. - Interest Targeting: Accesses a massive library of pre-built audience segments based on verified consumption habits across premium news, lifestyle, and tech publishers. - Search Keyword Targeting: Allows you to reach users who have recently searched for specific terms on the open web, capturing intent similar to traditional search engines. - First-Party Data Onboarding: Enables the secure upload of customer relationship management system (CRM) lists to re-engage existing leads or exclude current customers from prospecting campaigns. Best for: Realize is highly effective for performance marketers in high-consideration verticals — such as insurance, real estate, and education — who need to reach specific personas without relying on social media data. It’s particularly helpful for brands that need to scale their mid-funnel engagement by finding new pockets of relevant users that their current search and social campaigns are missing. Pricing model: Performance-based model; campaigns billed on a CPC basis or on a CPM basis for programmatic. Pros: - Utilizes AI and contextual signals that remain effective in a cookieless and privacy-regulated landscape. - Provides access to specific professional and interest-based segments found on premium publisher sites that social platforms often overlook. - The platform’s AI automatically shifts budget toward the audience segments with the lowest actual CPA. Cons: - Predictive and lookalike targeting require a baseline of conversion data to accurately identify and scale new audiences. - Certain high-intent targeting tools, like Mail Domain or Search Keyword Targeting, have limited availability in specific global regions. - Onboarding first-party CRM data for advanced targeting requires a more rigorous technical configuration than standard interest-based ads. ### 2. Google Ads Why it’s essential: Google Ads is built on something no other platform can replicate at scale: active search intent. When someone types a query, they’re announcing exactly what they’re thinking about. That signal is the most direct audience targeting available in advertising. Beyond search, targeting extends across Display, YouTube, and Performance Max. Custom intent audiences let you reach users who have been searching for specific keywords across Google properties. Showcased features: - Keyword-Based Search Targeting: Reach users at the exact moment they’re searching for your product or category. - Customer Match: Upload CRM data to re-engage prospects across Search, YouTube, Gmail, and Display. - In-Market Audiences: Target users actively researching or comparing products in specific categories. - Performance Max: Automated campaign type that serves across all Google channels, optimizing audience and creative combinations simultaneously. Best for: Advertisers whose customers actively search for their product or category. Strong for direct-response campaigns capturing existing demand. Pair with an open-web platform for awareness and discovery beyond active search. Pricing model: Primarily CPC for search. CPM and target CPA/return on ad spend (ROAS) bidding across display, video, and Performance Max. Pros: - Unmatched intent data from search behavior at scale. - No minimum spend — accessible at any budget level. - Customer Match integrates first-party data reliably. Cons: - Performance Max offers limited visibility into what the AI is actually doing. - Search CPCs have risen sharply in competitive categories. - Attribution accuracy has declined post-iOS 14, particularly for longer conversion windows. ### 3. Meta Ads Manager Why it’s essential: Meta Ads Manager covers Facebook, Instagram, WhatsApp, and the Meta Audience Network. The targeting strength comes from behavioral data Meta collects across its platforms, and the scale of that data is genuinely hard to match for consumer audiences. Meta Advantage+ has become the dominant campaign type for performance advertisers, replacing manual audience selection with Meta’s algorithm. It works well for e-commerce brands with sufficient conversion history. The tradeoffs are considerable, though, including iOS 14+ attribution losses, rising CPMs, and shrinking advertiser control over targeting decisions as Meta moves more of that logic behind its AI. Showcased features: - Custom Audiences: Upload customer lists or website visitors for precise re-engagement across Facebook and Instagram. - Lookalike Audiences: Meta’s lookalike modeling is among the strongest available for consumer brands. It builds statistically similar audiences from your best customers. - Advantage+ Shopping Campaigns: Fully automated campaign type that handles audience, creative, and budget for e-commerce advertisers with sufficient conversion data. - Cross-Platform Reach: A single campaign setup covers Facebook, Instagram, and the Audience Network. Best for: Consumer brands and e-commerce advertisers with good creative assets and enough conversion data for Meta’s algorithm to learn from. Less suited for B2B or professional persona-based targeting. Pricing model: CPC and CPM. Auction-based pricing varies significantly by industry and audience competition. Pros: - Scale and consumer reach are unmatched. - Lookalike modeling is best-in-class for direct-to-consumer (DTC) brands with clean first-party data. - Advantage+ reduces manual campaign management overhead. Cons: - Attribution accuracy has declined significantly post-iOS 14. - CPMs have risen sharply in competitive verticals. - Less advertiser control as Meta shifts targeting decisions to its AI systems. ### 4. LinkedIn Ads Why it’s essential: LinkedIn is the only major ad platform built on verified professional identity. Job title, company size, industry, seniority, and skills are all targetable. Unlike interest-based targeting elsewhere, these are self-reported and regularly updated by users with a professional incentive to ensure their accuracy. All of this makes LinkedIn the strongest platform for B2B targeting by a clear margin. However, LinkedIn ads can get expensive. CPCs frequently run $8–$15+ in competitive B2B categories, which demands campaigns with a clear return on investment (ROI) path to justify the spend. Showcased features: - Professional Attribute Targeting: Target by job title, function, seniority, company name, company size, industry, and skills, all from verified profile data. - Matched Audiences: Upload CRM contact or account lists to reach specific people or companies directly. Core to any account-based marketing (ABM) strategy. - Website Retargeting: Re-engage LinkedIn users who visited specific pages on your site. - Thought Leader Ads: Sponsor posts from individual employees, letting messaging carry personal credibility rather than a brand logo. Best for: B2B advertisers targeting specific professional personas, particularly at mid-market and enterprise accounts. Strong for ABM, pipeline generation for high annual contract value (ACV) products, and decision-maker brand building. Pricing model: CPC, cost per mille (CPM), and cost per lead (CPL). Minimum $10/day per campaign. Pros: - Only platform with reliable professional targeting at scale. - ABM by company, title, and seniority simultaneously — uniquely available here. - Revenue attribution reporting connects campaigns to pipeline. Cons: - CPMs are significantly higher than those of other platforms. - Audience sizes are smaller, limiting reach for broad awareness. - Ad fatigue happens quickly in niche professional segments. ### 5. The Trade Desk Why it’s essential: The Trade Desk is one of the leading independent DSPs for programmatic advertising across the open internet. Unlike walled gardens, it gives advertisers access to display, video, CTV, audio, native, and digital out-of-home (DOOH) inventory with a level of transparency and control that platform giants don’t offer. The Kokai AI system reached widespread adoption in 2025 and delivered measurable performance improvements for most advertisers who migrated to it. Audience Unlimited, launched early 2026, bundles third-party data segments at simplified pricing, directly addressing the long-standing issue of data costs consuming 20%+ of media budgets in programmatic buying. Showcased features: - Kokai AI Platform: Handles audience modeling, bidding, and budget allocation. Advertisers choose between Performance Mode (full AI control) or Control Mode (manual oversight with AI recommendations). - Audience Unlimited: Bundled access to curated third-party data segments at simplified pricing, significantly reducing data costs. - CTV and Streaming Targeting: Extensive connected TV inventory access with audience data layered on top of content and device signals. - First-Party Data Activation: Integrates with LiveRamp and other identity infrastructure for CRM activation across programmatic. Best for: Enterprise brands and large agencies running multi-channel programmatic campaigns, particularly those investing heavily in CTV. Pricing model: Percentage of ad spend. Pricing not disclosed publicly; requires sales engagement. High minimum spend applies. Pros: - Best-in-class transparency for programmatic buying. - CTV and streaming inventory access is among the strongest available outside walled gardens. - Over 95% client retention rate for 11 consecutive years. Cons: - High cost of entry — not accessible below a significant spend threshold. - The contracting process has documented flexibility issues, per user reviews. ### 6. Amazon DSP Why it’s essential: Amazon DSP gives advertisers access to Amazon’s purchase intent data, one of the most commercially valuable targeting signals in advertising. When someone searches for a product, adds it to a cart, or buys in a specific category, that behavior feeds audience segments that can be activated both on Amazon properties and across the broader programmatic ecosystem. For brands in retail and e-commerce categories where Amazon shopping behavior is a reliable proxy for buyer intent, the targeting depth is genuinely unique. Competitors’ transaction-based audiences at this resolution don’t exist anywhere else. Showcased features: - In-Market Audience Targeting: Reach users actively browsing and purchasing in specific product categories based on real transaction and browsing data. - Lifestyle Audiences: Segments based on recurring purchase behavior — actual spending patterns, not self-reported interests. - Customer Remarketing: Re-engages users who viewed your product pages, added to cart, or purchased, across both Amazon and off-Amazon inventory. - Advertiser Audience: Activates your own CRM data within Amazon’s ecosystem for re-engagement and lookalike expansion. Best for: Brands actively selling on Amazon and consumer packaged goods (CPG) or retail advertisers, where Amazon shopping behavior reliably signals category intent, regardless of where the final purchase happens. Pricing model: CPM-based. Managed service requires $50k+ monthly minimum. Self-service available for brands within the Amazon ecosystem at lower minimums. Pros: - Purchase-based audience data is the most commercially valuable targeting signal available. - On-Amazon and off-Amazon campaign coverage from one platform. Cons: - $50k+ managed service minimum is prohibitive for most mid-market advertisers. - Less relevant for B2B or categories where Amazon isn’t a meaningful shopping destination. - Walled garden structure limits transparency compared to open-web DSPs. ### 7. StackAdapt Why it’s essential: StackAdapt has grown from a native advertising specialist into a full-channel programmatic DSP covering display, video, CTV, audio, DOOH, in-game, and email. It’s consistently rated among the highest for customer satisfaction in the DSP category. Responsive account managers and strong support are recurring themes in reviews, which stand out in a space where enterprise platforms often deprioritize both. For B2B advertisers, StackAdapt’s ABM capabilities and access to professional third-party audience segments make it a credible complement to LinkedIn Ads at meaningfully lower CPMs. The Ivy AI assistant reduces the expertise barrier for teams without dedicated programmatic traders. Showcased features:  - Ivy AI Assistant: Natural language interface for campaign planning, audience recommendations, and performance interpretation. - ABM Targeting: Target by company, industry, job function, and seniority across the open web — similar reach to LinkedIn without the walled garden CPMs. - StackAdapt Data Hub: Centralizes first-party data for privacy-first audience activation across channels. - Multi-Channel Programmatic: Unified buying across native, display, video, CTV, audio, DOOH, and in-game from a single campaign setup. Best for: Mid-market brands and agencies needing full-funnel programmatic capabilities without the complexity or minimum spend of enterprise DSPs. Especially effective for B2B advertisers extending reach beyond LinkedIn. Pricing model: Usage-based: CPM, CPC, and cost per engagement (CPE). Starting price around $5,000 a month; varies by channel and spend level. Pros: - Customer support quality consistently rated excellent across software review platforms G2 and Capterra. - Broad channel coverage without enterprise-level complexity or minimums. - Strong B2B targeting outside LinkedIn, at lower CPMs. Cons: - About $5,000/month starting point for best results excludes smaller advertisers. - Reporting user interface (UI) has a meaningful learning curve. - Performance can vary, depending heavily on campaign setup quality. ## More About Audience Targeting for Performance Advertising ### What Are Audience Targeting Performance Platforms? Audience targeting platforms are advertising tools that help you reach specific groups of people based on behavioral, demographic, professional, or contextual signals. Instead of buying broad ad placements and hoping the right person sees them, these platforms let you define who you want to reach and serve ads only to people who match that profile. Good audience targeting platforms often win on data quality. Without quality data, you cannot predict how accurately the platform will identify the right person and how reliably it will reach them at the right time. ### How to Measure the Performance of Audience Targeting Campaigns Start by separating vanity metrics from business metrics. Impressions and reach tell you about exposure, but rarely tell you how well your campaign is working. Here are some metrics to keep your eyes on: Cost per lead (CPL): This metric measures what you’re paying for each qualified inquiry, form fill, or demo request. Cost per acquisition (CPA): With this stat, you get a measure of what you’re paying per conversion, whether that’s a purchase, sign-up, or other defined action. Return on ad spend (ROAS): Revenue generated per dollar spent. This metric is most relevant for e-commerce where purchase value is trackable. Audience quality indicators: Time on site, pages per session, and conversion rate from ad traffic tell you whether the audience you’re reaching is actually relevant, or just cheap to reach. While it’s crucial to track all these metrics, keep in mind that attribution is the hard part. Most buyers interact with multiple channels before converting, and each platform might claim credit for the same conversion. A multi-touch attribution model, or at minimum a consistent last-click methodology applied across all platforms, is essential for making accurate budget decisions. ### Key Features of Audience Targeting Software Before committing to any platform, here’s what separates strong targeting tools from ones that just look good in a demo. #### Data Quality and Sourcing Audience segments are only as useful as the data behind them. Some platforms use verified first-party behavioral data; others aggregate third-party signals of varying reliability. Ask where audience data comes from, how recently it was collected, and how it’s maintained. Outdated or loosely defined segments lead to broad, low-converting reach, which is worse than no targeting at all, because it appears to be working. #### Privacy-readiness The platforms still relying heavily on third-party cookies or cross-site tracking are operating on borrowed time in most markets. Look for platforms that have invested in contextual targeting, first-party data activation, and identity solutions that hold up under current privacy regulations and won’t collapse when the next update hits. #### Channel and Format Coverage Some platforms are single-channel specialists. Others run across display, video, CTV, audio, native, and DOOH through a single interface. If you’re managing campaigns across multiple environments, a platform with unified targeting logic across channels is significantly more efficient than stitching together separate tools. #### Transparency and Control Walled gardens (Meta, Google, Amazon) offer powerful targeting but limited visibility into exactly where your ads run and why they’re reaching the audiences they do. Open-web DSPs typically offer greater transparency, more control over placements, and greater flexibility in defining and refining audiences. #### Minimum Spend and Accessibility Enterprise DSPs like The Trade Desk and Amazon DSP have meaningful minimum spend requirements that price out smaller advertisers. Self-serve platforms like Google Ads and Meta Ads Manager are accessible at any spend level. Knowing where a platform sits on this spectrum matters before you get to a sales call. #### AI and Optimization Capabilities Most platforms now claim AI-powered optimization. The meaningful distinction is whether the AI improves audience discovery, creative optimization, or both. Platforms that combine predictive audience modeling with automated budget reallocation toward top-performing segments save the most operational time. ### Audience Targeting Platform Pricing Models Here are the common pricing models for audience targeting platforms: Cost per click (CPC): You pay each time someone clicks your ad. Platforms like Realize, Google Ads, Meta, and LinkedIn use this model. It works well for direct-response campaigns where clicks correlate with intent. Cost per thousand impressions (CPM): For this model, you pay for the exposure regardless of clicks. It’s standard for programmatic DSPs like The Trade Desk and Amazon DSP. You’d find this model useful if you’re running awareness campaigns. Cost per acquisition/action (CPA): You pay when a defined conversion happens. This model is available through smart bidding on platforms like Realize, Google, and Meta. Cost per view (CPV): This model is common on video platforms like YouTube. You pay when a viewer watches past a defined threshold. ### How to Set Up Audience Retargeting for Better Performance Retargeting reaches people who already know you, which makes it one of the highest-converting tactics available, but one of the most wasted when set up poorly. Do the following to get the best out of your retargeting campaigns: Segment your retargeting audiences: Not everyone who visits your sites should see the same ad. Someone who read a blog post is at a different stage than someone who visited your pricing page multiple times. Set frequency caps: Showing the same person the same ad multiple times reaches a point of diminishing returns, where your ad repels them rather than attracting them. Most platforms let you cap the number of times a user can see a given ad within a given time window. Use the feature! Define your retargeting window carefully: A 30-day window makes sense for a high-consideration purchase, but you might need a shorter window for products with a short decision cycle. Match the window to your actual sales cycle. Refresh creative regularly: Retargeting audiences are small and see your ads repeatedly. Creative fatigue sets in faster than in prospecting campaigns, so make sure to rotate your creatives frequently. ## Audience Targeting Best Practices ### Match the platform to the signal, not the budget The cheapest platform is rarely the right one. Use platforms where their data advantage is right for your specific audience, for instance, LinkedIn for professional personas and Realize for open-web discovery. ### Give AI campaigns enough budget to learn Underfunded AI campaigns never exit the learning phase. If a platform’s algorithm needs 50 conversions per week to optimize, a budget that generates 10 conversions per week will never perform well. ### Build audiences from first-party data first Your CRM, your website visitors, and your existing customers are your most valuable targeting seeds. Use them to build lookalike audiences and to exclude people who are already past the stage you’re targeting. ### Don’t run every channel simultaneously at launch Start with one or two platforms, establish baseline performance, then expand. Running six platforms at once with thin budgets means none of them learn fast enough to be useful. ### Audit audience overlap regularly If you’re running Google, Meta, and a DSP simultaneously, you’re almost certainly serving ads to the same people across all three and crediting multiple platforms for the same conversions. Use platform-level overlap reports and a consistent attribution model to understand what’s actually driving results. ### Test audiences as rigorously as the creative Most advertisers A/B test ad copy and images constantly, but rarely test audience definitions with the same discipline. Small changes in audience segmentation, such as narrowing an age range, adding a behavioral signal, or excluding a job function, can meaningfully affect performance. ## Key Takeaways The best audience targeting platform is going to be the one most suited to your specific needs, whether that’s B2B, e-commerce, or brand awareness. Whichever platform you choose, though, you should ensure that it provides transparent data, accounts for constantly-updating privacy laws, and covers the full breadth of your campaign where possible, to avoid having to track data across multiple platforms. ## Frequently Asked Questions (FAQs) ### What are some audience targeting best practices for e-commerce performance? Start with your customer purchase data as a seed audience, then build lookalike segments from it. Segment retargeting by intent level — you cannot expect cart abandoners and product page visitors to be the same. Set frequency caps on small retargeting pools to prevent creative fatigue, and for prospecting, pair a social platform with an open-web DSP to cover discovery and capture demand across the full purchase path. ### Which are the best audience targeting platforms for small businesses? Google Ads and Meta Ads Manager are the strongest starting points. They have no minimums, have self-serve access, and enough targeting depth to drive real results. Google works best when customers are actively searching for what you sell, while Meta works better for consumer brands building demand. LinkedIn Ads is the right call if you’re targeting a B2B audience, but bear in mind that it has higher CPCs. If you have a budget starting from $5,000/month, you can consider StackAdapt. ### How do you start using AI for e-commerce advertising? Get conversion tracking right first. AI campaign types such as Meta Advantage+, Google Performance Max, and Realize’s predictive targeting need conversion data to learn from. Once you have the conversion data, give whichever platform you choose enough budget to generate 30-50 conversions per week. Feed it strong inputs like your CRM list, product catalog, and a wide creative set. ### The Trade Desk vs. Realize audience targeting capabilities: What are the differences? The Trade Desk is built for enterprise advertisers running multi-channel programmatic at significant scale, with high spend minimums and a sales-led process. Realize is built for performance marketers who want AI-driven open-web targeting at more accessible spend levels, with predictive audience modeling and a CPC-based model. --- ### 4 Essential Steps to Ensuring Ad Compliance and Reducing Creative Rejections URL: https://www.taboola.com/marketing-hub/ensure-ad-compliance-and-reduce-creative-rejections/ Last Modified: 2026-06-04 09:34:19 Ad rejection is a direct drag on ROI, scale, and momentum. Every time a platform kicks back your creative, you lose more than a couple of hours of production time: you also lose the first days of a learning phase. You also lose windows for time-sensitive promotions, and you risk account flags that silently suppress delivery just when you need volume the most. Most marketers respond to this by tweaking bids, adjusting audiences, and reshuffling budgets, but if your creative is constantly getting rejected, you’re basically pouring media dollars into a leaky bucket. The only way to fix this at scale is to move away from reactive “appeal and resubmit” habits and build a proactive, automated compliance system that keeps your ads running and your spend flowing toward approved, working media. ## Step 1: Pre-validate your creative against platform policies When you strip away platform-specific nuances, most rejections boil down to a few fundamental issues: unsubstantiated claims, prohibited content, spammy formatting, privacy violations, and weak landing page alignment. If you can systematically prevent these problems inside your creative workflow, you eliminate the majority of avoidable rejections before they ever reach a review queue. That starts with a substance-first mindset. ### Substantiate All Claims A Substance-First Checklist means you don’t let clever copy or flashy visuals leave the draft stage until they’ve passed a basic truth and compliance test. Every performance claim should trigger a simple question: can we prove this, right now, with a concrete and current piece of evidence? Claims like “99% effective,” “double your results,” “lowest price anywhere” are immediate red flags, both for AI screeners and human readers. If you can’t easily prove a claim, either adjust the claim (“clinically tested,” “proven in customer trials,” “competitive pricing”) or drop it altogether. ### Avoid Prohibited Content Platforms are increasingly aggressive about anything that looks like misrepresentation, and absolute language and guarantees are the easiest signals to pick up in automated review. At the same time, prohibited content should never be a difficult judgment call for advertisers. Advertising platforms put a hard wall around illegal products, discriminatory or hateful language, and anything designed to shock or disturb. If a piece of creative needs a debate about whether it crosses a line, that alone is a signal to rework it. ### Maintain Professional Tone and Formatting Most ad systems default on the side of caution, and so should you. Tone and formatting are key signals to screeners and readers that the copy is making promises it can’t keep. ALL CAPS HEADLINES, strings of punctuation, and gimmicky symbol use may feel like “thumb-stopping” tactics, but from an AI’s perspective they look indistinguishable from spam. That means a higher probability of rejection and lower quality scores, even when the creative technically passes. Professional, readable copy doesn’t just help you clear review, it also converts. ### Respect Data Privacy  Privacy is the next non-negotiable. Under frameworks like GDPR and CCPA, you can’t imply that you have specific information about a user’s health, finances, or identity. That means no “We saw your credit score” or “We know you’re struggling with debt” phrasing, and no ad text that suggests you’re targeting people based on sensitive attributes. Even if your targeting is compliant, the copy itself can make it look like a violation, which is enough to get you rejected or escalated. ### Observe Text-to-Visual Ratio Finally, visuals aren’t just eye-grabbing, scroll-stopping mind candy. They need to be treated as part of the compliance surface. Overloaded images, heavy text overlays and confusing layouts are more likely to be flagged by platform quality filters. A clean visual hierarchy, minimal text on images, and honest representation of the product or service help you avoid quality downgrades and keep your creatives on the right side of automated checks. ## Step 2: Optimize Your Landing Page Once the asset itself is compliant, the next layer is its destination: the landing page. ### Match Message to Destination Relevancy is a major trust signal for platforms, and a misalignment between promises and experience is one of the fastest ways to get an otherwise clean ad rejected. The rule of thumb is simple: the story your ad tells must match the story your landing page continues. These are three rules of thumb to follow: - If you promote a 20% discount, that discount should be clearly visible above the fold when a user clicks. - If you show a specific product, it should be present and purchasable on the destination, or one click away via obvious navigation. - If the ad suggests one core benefit and the page pushes something totally different, you’re inviting trouble. ### The Three-Click Rule A practical heuristic here is the three-click rule. Someone who lands on your page from an ad should be able to find the advertised product, service, or core action within three clicks, maximum. If they can’t, that’s a signal not just of bad UX but of potential misrepresentation. Fixing this often means simplifying navigation, linking directly to product or offer pages, and stripping away distractions that bury the core experience. ### Targeting Alignment Targeting alignment is another subtle but critical piece. Platform rules around discrimination are strict, especially for sensitive verticals like housing, employment, and credit. Your targeting controls need to comply, and your copy cannot imply that certain groups are being excluded or singled out based on protected characteristics. Even unintentional wording can create the appearance of bias. ### Technical Compliance Broken links, pages that take too long to load, forced downloads, excessive pop-ups, or autoplay audio can all trigger quality issues or hard rejection. You can have flawless creatives and still fail review because the landing page is slow or unstable. A basic QA pass should be part of your standard pre-launch process. Check load speed and mobile responsiveness before you upload the ad, and eliminate as far as possible the presence of disruptive elements. When you combine substance-first creative, privacy-safe messaging, clean formatting, and tightly aligned landing experiences, you create a baseline environment where policy violations are the exception rather than the rule. That’s the foundation you need before you even consider automation. ## Step 3: Vet and Test Your Creative Prior to Submission Even with a strong policy foundation, manual oversight will eventually fail under the pressure of scale. Once you’re running dozens or hundreds of creatives across multiple platforms, markets, and formats, a purely human review process becomes slow, inconsistent, and expensive. This is where automation stops being a nice-to-have and becomes essential infrastructure for performance marketing. ### AI for Future-Proofing The first layer of automation is pre-submission creative vetting and testing. Instead of relying on individual marketers to remember every nuance of every policy update, you introduce AI-powered tools that sit inside your workflow and act as an intelligent safety net. ### Pre-Submission Vetting These systems can scan ad copy, images, and landing page URLs against a live map of platform restrictions: banned phrases, restricted verticals, claim language that requires substantiation, privacy-sensitive wording, or risky imagery. The goal is to catch violations before your ads ever hit the review queue. That alone can save hours or days of lost time, especially when you’re working under launch deadlines and promotion windows. ### A/B Testing Compliance This same layer of automation should also reshape how you think about A/B testing. Historically, marketers might throw several edgy or aggressive variants into the mix to see what sticks. In a stricter policy environment, that approach is dangerous. Every non-compliant variation you submit doesn’t just get quietly rejected: it contributes to an account-level risk profile. A disciplined setup means running your variants through automated checks first and only allowing fully compliant creative into tests. ### Real-Time Ad Compliance Real-time ad compliance extends this concept beyond submission. Policies change, promotions expire, landing pages get updated, and what was compliant yesterday can become problematic tomorrow. If you’re not monitoring live campaigns, you may only discover issues after a wave of rejections or a sudden performance drop. Automated monitoring tools can regularly ping your active ads, re-check the landing pages and alert you if broken links, a changed offer, or a newly restricted phrase shifts your ads out of bounds. Instead of waiting for the platform to penalize you, you get a chance to fix or replace the asset proactively. ## Step 4: Automate Stop-Loss and Pausing The second major automation layer is what you can think of as “stop-loss” logic for creative. Just as traders use stop-loss orders to cap downside risk in markets, performance marketers can use automated rules to detect and neutralize threats from rejected or problematic ads as soon as they appear. ### Custom Rules for Compliance Custom rules can be configured to listen for specific signals: an ad’s status changing to “rejected” or “limited,” a spike in policy-related disapprovals, a sudden drop in delivery tied to review feedback. When those conditions are met, the rule automatically pauses the offending ads, excludes the creative from ad sets or campaigns, or routes it into a remediation queue. There are two upsides to creating custom rules. First, you avoid wasting budget on campaigns where a rejected creative is effectively blocking spend or throttling delivery. Second, you limit the number of rejections accruing to your account, which reduces the risk of deeper reviews or suspensions. ### Proactive Budget Protection Proactive budget protection is the natural extension of this. With automated guardrails, your media spend stays concentrated on approved, active creatives that are actually capable of driving results. Instead of manually scanning dashboards for red warning icons or “rejected” tags, your system does the watching, and your team focuses on optimizing messaging, audience strategy, and funnel performance. Over time, this combination of pre-submission vetting, real-time monitoring, and automated stop-loss rules becomes a sort of compliance nervous system for your marketing operation. It continuously senses, interprets, and reacts to risk signals so you don’t have to catch every issue with human eyes. ## Key Takeaways Creative compliance is a core performance lever that decides whether your campaigns get a fair shot at scale, or die in the review process. When you build a substantiation-first creative workflow rooted in factual claims, clean formatting, privacy-safe messaging, and tightly aligned landing pages, you drastically reduce the likelihood of rejection before automation even enters the picture. Layering in platform intelligence and AI-powered tools like Realize to scan, monitor, and protect your campaigns in real time turns that strong foundation into a resilient, scalable system. The net effect is fewer interruptions, fewer last-minute fire drills, stronger brand safety, and a higher share of your media spend flowing to approved, working ads. In an environment where competition is fierce and review standards are tightening, treating compliance as a performance driver is what separates fragile campaigns from truly scalable performance marketing. ## Frequently Asked Questions (FAQs) ### How can I speed up creative approvals and reduce rejections by catching policy issues before I submit? To speed up creative approvals and minimize rejections, you can make use of integrated AI-powered creative generation tools trained on platform policies. Realize, for example, can scan your copy, imagery, and landing page alignment in real time, flagging prohibited content and suggesting immediate fixes before you ever hit submit. Realize is built on a proprietary framework trained with best practices to ensure high-quality, compliant output. Additionally, Realize’s ABBY AI assistant provides real-time fixes for creative rejections, helping you catch policy issues early and significantly reduce approval delays. ### If an ad gets rejected, how can I stop budget wastage and prevent the issue from impacting my account health? To prevent budget wastage and protect your account health when ads are rejected, you must use automated custom rules configured for compliance. These rules automatically check ad status and instantly pause any rejected ads or ad sets, ensuring you don’t spend money on disabled creatives or accumulate negative policy flags. Realize enhances this protection through its Custom Rules for Campaign Optimization. It features specific stop-loss functionality, allowing you to automatically pause underperforming or rejected ads based on real-time campaign results. This ensures your budget is strictly protected and channeled only toward scaling the creatives that work. ### How do I ensure my landing page is compliant and matches my ad, especially regarding relevance and disclosure? Ensuring landing page compliance and relevance relies on prioritizing the user experience. Realize supports this goal with Codeless Conversions, which enables easy, no-code setup of event-based tracking (such as button clicks). You can use Realize to implement strict relevance checks like the “three-click rule” to ensure the advertised product is immediately accessible. Realize’s optimization capabilities also ensure the entire ad experience, from the creative to the landing page, is fully compliant and seamlessly aligned to drive user conversions. --- ### Meta Advantage+: The Complete Strategy Guide for Performance Advertisers URL: https://www.taboola.com/marketing-hub/meta-advantage-plus/ Last Modified: 2026-06-01 11:02:08 The manual Facebook ads playbook — building custom audiences, layering interests, splitting budgets across ad sets — was built for a signal-rich environment that Apple’s privacy changes and browser restrictions have largely dismantled. Meta’s answer to this new reality is Meta Advantage+, a suite of AI-powered tools that can automate audience targeting, placements, budget optimization, and creative testing. For advertisers used to managing campaigns through detailed settings and segmented ad sets, relinquishing control to AI can feel like taking your hands off the wheel before you fully trust the autopilot. This guide covers what’s in the suite, how it differs from standard Advantage features, and how to structure campaigns to get the most out of it. ## What is Meta Advantage+? Meta describes Advantage+ as a suite of products that help advertisers use AI to optimize campaigns in real time and match ads to the people most likely to take action. The suite includes both end-to-end campaign solutions and single-step automations that allow you to apply AI to only specific parts of the campaign, such as audience, placement, creative, destination, and budget. In practice, Meta Advantage+ is a form of Meta Ads automation that helps determine who sees your ads, where they appear, how spend is distributed, and which creative variations are most likely to drive results. That makes it part of the broader move toward agentic AI ad campaigns, where the platform doesn’t just execute fixed settings, but uses campaign goals, available signals, modeled behavior, conversion data, and machine learning to continuously optimize toward a defined outcome. As an advertiser, you still set the objective, budget, creative inputs, conversion event, and guardrails, but instead of manually defining every path to performance, you give Meta’s system more flexibility to find the combinations most likely to convert. ## Meta Advantage vs. Meta Advantage+: What’s the Difference? The naming can be confusing, but understanding Meta Advantage vs. Advantage+ is important. Meta Advantage refers to individual automated features within a manually structured campaign, such as Advantage Detailed Targeting (which slightly expands your defined audience) or Advantage Lookalike (which broadens the lookalike range). You stay in control of the overall setup, but can use these as optional assists. Meta Advantage+ is a fully automated campaign type. While you set the objective, creative inputs, and total budget, the audience, placements, budget pacing, and creative testing are all handled automatically by the system. The simplest way to think about it is this: Advantage enhances a manual setup. Advantage+ replaces it. ## Core Tools in the Meta Advantage+ Suite Each Advantage+ tool automates a campaign decision that advertisers used to manage manually. ### Advantage+ Shopping Campaigns (ASC) ASC, also known as Advantage+ sales campaigns, is built for e-commerce brands. It consolidates prospecting, retargeting, and broad reach into a single budget pool, dynamically allocating spend based on where conversions are most likely to occur. You connect your catalog and conversion signals, upload creative assets, and choose the objective, and Meta’s system optimizes delivery toward the combinations most likely to drive purchases. For brands with enough conversion data, ASC can reduce account complexity and improve efficiency. Meta reports 20% improved cost per action or acquisition (CPA) when using Advantage+ sales campaigns. ### Advantage+ Audience Advantage+ Audience uses AI-driven audience discovery and machine learning ad targeting to move beyond rigid, manual segments. It uses inputs such as customer lists, website traffic, engagement data, lookalikes, interests, and demographics as signals, rather than strict targeting limits. Meta prioritizes people who match those suggestions, then expands beyond them when it predicts better performance. Advertisers can still set firmer controls for location, minimum age, language, and custom audience exclusions. ### Advantage+ Placements Advantage+ Placements distributes spend across Meta’s eligible inventory, including Facebook, Instagram, Stories, Reels, Messenger, and the Audience Network, shifting budget in real time toward the best-performing placement for each creative-audience combination. That flexibility can improve efficiency, but advertisers with strict placement or brand safety requirements should review where ads can appear before relying on full automation. ### Advantage+ Creative Advantage+ Creative applies automated enhancements and dynamic creative optimization (DCO) to uploaded assets, including visual touch-ups, AI-generated text variations, music, translations, subtitles, overlays, product tags, and background generation. It helps adapt creative across placements without requiring manual versioning, but the output still depends on the strength of the source material. As targeting broadens, creative quality becomes a stronger performance signal because the system learns which messages, formats, and offers attract different users. ## Advantage+ vs. Manual Campaigns: The Trade-Off The efficiency gains can be real, but Advantage+ also changes how advertisers manage performance. Advantage+ is strongest for efficiency, speed, and return on ad spend (ROAS) optimization, especially for e-commerce brands with sufficient conversion volume. Manual campaigns are stronger for granular control, data transparency, precise targeting requirements, and cleaner testing conditions. For many performance advertisers running direct-response campaigns at scale, Advantage+ is a strong default to test. Manual setups still have a role when campaigns require narrow audiences, strict geographies, regulated targeting, brand safety controls, or a clearer view into why something worked. ## The Meta Advantage+ Strategy for Performance Advertisers Switching to Advantage+ isn’t enough on its own. The system performs best when you give it the right conditions: - Go broad. Avoid over-layering interests or narrowing the audience too much. Broad targeting gives the algorithm more room to identify patterns. - Budget realistically. There’s no universal minimum, but many advertisers use around $100 per day as a practical testing benchmark. The right number depends on your CPA, conversion rate, and learning needs. - Avoid constant interruptions. Frequent budget changes, audience edits, and creative swaps disrupt learning and make performance harder to evaluate. - Use customer lists as signals. Customer relationship management system (CRM) data, website visitors, and past purchasers can help guide the system, even when Meta expands beyond those inputs. - Measure against the right baseline. Compare Advantage+ to your broader manual setup, not just to a single ad set. ## Why Creative is the New Targeting When Meta controls more of the “who,” advertisers need to focus more on the “what.” Each asset gives the system clues about who may care, what message resonates, and which offer is most likely to convert. Feed Advantage+ campaigns three to five meaningfully different creative variations. Don’t rely on minor copy swaps. Test across three areas: - Creative type: static, short-form video, carousel, Reels-native, Stories-native. - Message: product benefit, social proof, problem/solution, offer-led. - Intent stage: cold prospecting, warm consideration, purchase-ready. The algorithm can optimize delivery, but it can’t fix weak positioning or invent a stronger offer. Creative strategy remains one of the advertiser’s most important levers. ## When Not to Use Meta Advantage+ Meta Advantage+ is useful, but it's not the right fit for every campaign. Manual campaigns may be better when you’re targeting narrow B2B segments, specific job titles, limited geographies, or niche verticals where broad modeling could dilute the audience. They may also make more sense for regulated categories or brands with strict placement requirements. Manual setups are also useful for controlled testing. If you need to isolate a specific audience, message, placement, or offer, a fully automated campaign can make results harder to interpret. Low-signal accounts should also be cautious. If you don’t have enough conversion volume, Advantage+ may need more time and testing before it becomes reliable. ## Key Takeaways Meta Advantage+ is Meta’s clearest signal yet that AI-led advertising is the default, not the exception. For performance advertisers, the adjustment is to stop treating automation as a loss of control and start managing the inputs that still shape results: creative quality, budget, conversion data, audience signals, and campaign stability. A hybrid approach is often strongest. Use Advantage+ as a primary performance engine when you want scale and efficiency. Keep manual campaigns for precise segments, specific tests, or situations where visibility and control matter more than automation. ## Frequently Asked Questions (FAQs) ### What is the difference between Advantage and Advantage+? Advantage+ is Meta’s more automated campaign experience, where AI manages audience, placement, budget, and creative together. Standard Advantage features apply automation to individual settings within a campaign that you otherwise control manually. ### Do I need a large budget for Meta Advantage+ campaigns? You don’t always need a large budget, but you need enough spend to generate useful learning data. Many advertisers use around $100 per day as a practical benchmark, though the right budget depends on your CPA, conversion volume, and objective. ### Does Meta Advantage+ completely replace manual ad targeting? For many direct-response and e-commerce campaigns, it can reduce the need for detailed manual targeting. Advantage+ Audience uses broad signals and machine learning to find likely converters. Manual targeting still matters when you need strict control, niche segmentation, or clean tests. ### Are Advantage+ Shopping Campaigns (ASC) better than manual setups? Often, yes, for e-commerce advertisers with enough conversion data and strong creative inputs. Advantage+ Shopping Campaigns (ASC) can optimize across audiences, placements, creative, and budget faster than a manual structure. Manual campaigns may still be better when control, transparency, or precise testing matter more than scale. --- ### How Adaptive Learning Algorithms Improves Ad Performance URL: https://www.taboola.com/marketing-hub/adaptive-learning-algorithms-performance-advertising/ Last Modified: 2026-06-01 08:56:49 If you’re still reviewing your ad campaigns once a week, or even at the end of each day, you’re already behind. Consumer intent shifts with the news cycle, competitor bids spike without warning, and auction prices can swing dramatically between breakfast and lunch. Your ad strategy needs to move as fast as the market does. That’s the promise of adaptive learning in advertising: machine learning systems that don’t just respond to change, they anticipate it, continuously recalibrating bids, targeting, and creatives so your campaigns are always competing at their best. ## What Is Adaptive Learning in Digital Advertising? The term “adaptive learning” has its roots in education technology, where it described systems that adjusted lesson content based on how individual students were progressing, personalizing the experience in real time based on what’s actually working for each learner. In AdTech, adaptive learning refers to machine learning engines that continuously update their own predictive models based on incoming performance data. Rather than operating from a fixed rulebook, these systems review live signals like user behavior, conversion patterns, and engagement rates, and use them to refine targeting pathways and bidding logic on an ongoing basis. This type of system is less like a campaign manager who reviews performance weekly and more like a trading algorithm that’s always processing new information, making micro-corrections, and optimizing toward a defined outcome. ## The Shift from Static Campaigns to Continuous Learning Models Traditional media buying has always leaned heavily on historical data. You look at what performed well last quarter, build a campaign around those insights, launch it, and optimize. This model works, but it’s fundamentally reactive. By the time you’ve gathered enough data to make an adjustment, the window of opportunity has often already closed. Continuous learning models are the opposite of this. Instead of periodic check-ins, they operate on live feedback loops. Every impression, click, scroll, and purchase feeds directly back into the system, which updates its models without a delay. This doesn’t make human expertise obsolete, but redirects it toward higher-leverage decisions like setting goals, shaping creative strategy, and building the data infrastructure that powers the algorithm. ## Why Your Ad Strategy Must Change Every Hour When advertisers first see their data, they’re often surprised that the same ad placement can cost dramatically different amounts depending on the hour of day, and perform at dramatically different conversion rates depending on who’s seeing it and when. User intent is volatile. Someone browsing at 7 a.m. on a commute is in a very different mindset than that same person researching a purchase at 9 p.m. on the couch. Auction pressure fluctuates as competitor budgets deplete mid-day or surge around key events. Creative fatigue sets in faster than most teams realize. An hourly ad strategy built around adaptive algorithms can catch these micro-windows, capitalizing on a dip in competitor spend, shifting budget toward a high-intent audience segment that’s suddenly more active, or rotating a creative before engagement drops. A human buyer checking in every 24 hours simply cannot act on those signals fast enough. The algorithm can, and does, continuously. ## How Adaptive Learning Algorithms Work for Ads At its core, an adaptive ad system is an engine built around a continuous input-output loop. On the input side, it ingests data points that most traditional campaigns either ignore or examine too infrequently, like clicks, scroll depth, hover time, micro-conversions like add-to-cart events, and video completion rates. Real-time bidding algorithms use this data to make immediate decisions, not just whether to bid, but how much to bid, for which user, on which device, at what time of day. The system is constantly running experiments, evaluating outcomes, and updating its model weights to reflect what’s actually working right now. This is what separates machine learning ad performance from rule-based optimization. A rule-based system does what you tell it to, like “increase bids when CTR exceeds 2%.” A machine learning system figures out why click-through rate (CTR) is exceeding 2% in certain contexts and proactively finds more of those contexts before you have to ask. ## Dynamic Creative Optimization (DCO): Adapting the Message Adaptive learning doesn’t just adjust bids, it changes what users actually see. Dynamic creative optimization (DCO) is the mechanism by which algorithms personalize ad creatives at the individual level, testing combinations of headlines, images, copy lengths, and calls to action (CTA) to find the variation most likely to resonate with a specific user at that moment. Rather than choosing a single top creative and running it to the whole audience, DCO treats every impression uniquely. A user who recently browsed running shoes might see a product-focused headline and a performance image. Someone earlier in the funnel might get a brand story and a softer CTA. The algorithm learns which combinations drive conversions across thousands of user profiles simultaneously, at a scale no manual creative testing process could match. ## Combating Ad Fatigue with Algorithmic Rotation Even the best-performing creative eventually wears out. Users who have seen the same ad five or 10 times stop registering it or start to tune out the brand entirely. Ad fatigue prevention is one of the clearest benefits for adaptive systems, and it’s something manual campaign management consistently struggles with. Adaptive algorithms detect the earliest signals of fatigue, like a gradual drop in CTR, a rising cost per acquisition (CPA), or decreasing engagement depth, before they become obvious problems. Once those signals cross a threshold, the system automatically rotates in fresh creative assets without any human intervention required. The result is that campaigns maintain consistent performance over longer periods. Rather than pushing a creative to the point of diminishing returns and then rushing to refresh, adaptive systems keep the rotation dynamic from day one. ## Real-World Examples: Meta Advantage+, Google PMax, and Realize+ Meta’s Advantage+ campaigns and Google’s Performance Max (PMax) both rely on programmatic advertising AI to make real-time decisions about placement, audience, bidding, and creative delivery. Both systems look at signals from across their respective ecosystems, like search history, on-platform behavior, and conversion data from the advertiser’s pixel, and use them to continuously improve campaign performance without manual bid adjustments. These platforms aren’t without criticism, of course, the biggest being that they operate as black boxes. Advertisers can see outcomes but have limited visibility into why the algorithm made specific decisions. Realize+ takes a different approach, delivering the same adaptive learning capabilities with real-time bidding, dynamic creative testing, and audience refinement, while providing advertisers with direct, first-party code to publishers. That transparency matters for brands that need to understand and justify their ad spend, not just review the results. ## The Data Foundation: Feeding Your Adaptive Engine A common mistake advertisers make when adopting adaptive systems is assuming the technology does the heavy lifting from day one, which isn’t the case. Predictive ad targeting is only as accurate as the data feeding it, and poor data infrastructure is the single most common reason adaptive campaigns underperform. First-party data collected with user consent, server-side tracking through a conversions API, and clean customer relationship management system (CRM) integration are essential for the algorithm to connect ad exposure to actual business outcomes, such as return on ad spend (ROAS). ROAS optimization depends on the algorithm having an accurate signal of what a conversion is actually worth, not just whether a click happened. Without that signal quality, the system optimizes toward the wrong objective and produces results that look good on vanity metrics, but miss on revenue. ## Overcoming the Learning Phase in Adaptive Advertising Every adaptive system needs time to calibrate. During the learning phase, the algorithm is running experiments, building user profiles, and accumulating enough conversion data to make reliable predictions. Performance during this period is often volatile as CPAs change rapidly, spend paces unevenly, and results can look troubling if you’re not expecting any of this. Algorithmic ad buying depends on structural conditions to exit this phase quickly. Audience pools need to be large enough to generate meaningful data without over-segmentation, and daily budgets need to be sufficient to accumulate conversions at pace. Typically, a minimum of 50 conversions per week per ad set is the rough industry benchmark, and creative assets need to be in place before launch, rather than added in stages after. The biggest mistake advertisers make during the learning phase is intervening too early. Changing budgets, audiences, or creatives resets the clock. Set structural parameters upfront and let the system build its model on stable inputs. ## How to Transition Your Team to an Adaptive Ad Strategy The shift to adaptive systems isn’t just a technology change but a workflow change. Media buyers who previously spent their days adjusting bids, toggling audience segments, and swapping creatives manually, now need to redirect that work. Instead, set campaign constraints and business objectives the algorithm optimizes toward. Build and maintain the data pipeline, ensuring first-party data is flowing cleanly, conversion events are firing correctly, and attribution is accurate. Develop creative assets at volume, because an adaptive system’s personalization is only as good as the creative library you give it. Monitor for structural issues rather than tactical ones at first, so that you can leave the algorithm to handle the decisions, while you take care of the big-picture strategy. Teams that make this transition well will find that results improve significantly, not because they’re working harder, but because human effort is aligned with problems that actually require human judgment. ## Key Takeaways Adaptive learning represents a shift in how advertising campaigns are managed; not an incremental improvement on existing approaches, but a fundamentally different model. The manual, check-in-based optimization cycle that defined digital advertising for its first two decades is being replaced by systems that recalibrate continuously and improve their own decision-making as they go. For marketing teams, this means greater efficiency, less wasted spend, and the ability to compete in auction environments that move faster than any human process can keep up with. The brands that move ahead will be the ones that treat their ad platforms as learning systems, feeding them quality data, giving them clear objectives, and trusting the algorithm to find the path. ## Frequently Asked Questions (FAQs) ### What is adaptive learning in advertising? Adaptive learning in advertising is a machine learning system that never stops optimizing. It continuously reads performance signals, like clicks, conversions, and engagement patterns, and uses them to automatically refine bids, audiences, and creatives in real time. The goal is to remove the lag between data and action that manual campaign management has always suffered from. ### Why should an ad strategy change every hour? Because the conditions shaping your campaign’s performance, auction prices, competitor activity, and user intent, are shifting constantly. An ad strategy using yesterday’s data is already out of date. Hourly recalibration lets algorithms catch and act on those shifts before they cost you conversions. ### How does adaptive learning affect ad creatives? Rather than running one version of an ad to your entire audience, adaptive learning uses DCO to build and test thousands of creative combinations simultaneously, matching the right message to the right person based on their behavior and context. ### Do adaptive algorithms replace human media buyers? Not at all, though they do fundamentally change what the job looks like. The algorithm handles the high-volume, real-time decisions that no person could reasonably make at scale. Media buyers shift their focus to the things machines can’t do, like interpreting business context, developing creative strategy, and making sure the data pipeline feeding the system is clean and complete. --- ### Broad Targeting: Scaling on the Open Web With AI URL: https://www.taboola.com/marketing-hub/broad-targeting/ Last Modified: 2026-06-04 09:36:39 Performance advertisers spent years refining hyper-granular audience segments by layering interest categories, demographic filters, and behavioral signals in pursuit of the “perfect” user. That approach worked when data was more accessible, tracking was more dependable, and inventory was relatively cheap, but those conditions no longer hold in the same way. Privacy regulations are tightening, third-party signals are weakening, and the walled gardens of search and social are increasingly saturated and expensive. As a result, advertisers are shifting toward AI-powered broad targeting on the open web, using real-time contextual and performance signals to find high-intent users at scale, without the tracking infrastructure marketers once depended on. ## What Is Broad Targeting? Broad targeting is an advertising strategy built on minimal constraints. Instead of relying on narrowly defined audience segments, you set a few basic parameters, such as geography, device, or age range, and give the platform room to determine who’s most likely to convert. With a broad targeting strategy, machine learning takes on more of the audience discovery work, finding performance patterns across a wide range of signals and scaling in ways manual segmentation can’t. ## The Evolution of Audience Reach: Why Manual Targeting Is Fading Granular interest targeting was the gold standard when data was abundant and customer journeys were more linear. Conversion paths now stretch across devices, channels, and moments, while the signals that powered behavioral targeting continue to erode. Simultaneously, algorithms have become better at identifying likely converters in real time, making hyper-specific audience builds more expensive, less scalable, and less effective than they used to be. ## How AI Is Revolutionizing Broad Targeting Modern AI doesn’t need interest labels to understand intent. It analyzes content consumption, contextual relevance signals, device signals, and engagement patterns across big datasets to identify users most likely to act at that instant, then fine-tunes delivery based on what’s actually driving performance. AI-powered ad optimization replaces the static audience segment with a self-learning, self-correcting system. ### Processing Real-Time Contextual Signals When an ad opportunity becomes available, AI processes page content, device, time of day, and placement, while also weighing historical engagement patterns to predict which users are most likely to respond. On platforms like Realize, this means reaching users in relevant moments without tracking them across the web. ### Moving From Demographic to Predictive AI Models Static demographics answer who someone is. Predictive audience modeling answers whether they’re likely to convert right now. A user who fits a demographic profile may still be far from a purchase, but someone engaging with relevant content in the right moment may be ready to act. Predictive models learn from real-time engagement, producing more accurate conversion forecasts than fixed audience profiles alone. ## Navigating Signal Loss and a Privacy-First Future Broad targeting is well-suited to a cookieless world. It relies less on cross-site tracking and static user profiles, and more on contextual signals, first-party data targeting, and machine learning to guide delivery. Those inputs are more durable as privacy-first advertising becomes the norm. As you plan for the next few years, cookieless targeting solutions powered by machine learning ad delivery should be the foundation, not the fallback. ## Broad Targeting vs. Lookalike and Retargeting Audiences Lookalike and retargeting strategies capitalize on existing intent, but both are limited by the size and quality of the audience data you already have. Broad targeting works earlier in the process, reaching potential buyers before intent is explicitly signaled. Rather than relying on a pixel pool or customer list, it gives AI room to discover net-new audiences at scale, often with lower CPMs. Retargeting helps convert known interest, lookalike targeting extends it, and broad targeting uncovers new demand that feeds the funnel. ## Why Your Ad Creative Is the New Targeting Tool As audience targeting becomes less manual, the ad itself plays a bigger role in who responds. Creative diversification — building varied messages and formats for different buyer personas, pain points, and value propositions — gives the algorithm more signals to work with, helping it match the right creative to the right user segments. As performance data comes in, delivery naturally shifts toward what converts, with the creative doing more of the targeting work. ## The Strategic Benefits for Performance Advertisers Broad targeting on the open web delivers several compounding advantages for performance advertising scale: - Lower CPMs from wider access to underused premium publisher supply. - Faster scalability without the ceiling imposed by narrow audience definitions. - Organic audience discovery of high-performing segments across geographies, content categories, and device patterns. - Reduced overhead with fewer audience toggles to manage, freeing teams to focus on creative and strategy. ## When Should Advertisers Choose Broad Targeting? Broad targeting is a strong fit for new product launches, where there’s no seed audience; geographic expansion, where historical data is limited; and conversion campaigns that have plateaued, where narrow targeting is often the bottleneck. In each case, give the campaign enough budget to move through its learning and optimization phase, as that early ramp-up is part of the process, not a short-term test. ## Best Practices for Scaling Broad Campaigns on the Open Web Broad targeting works best when you give the algorithm room to learn while supplying strong signals. Setup tips: - Strip unnecessary constraints: Start with geography and device only; add filters back only when business rules require them. - Feed strong conversion signals: A well-implemented conversion pixel or server-side API, like the Realize Pixel, gives the system the data it needs to optimize. Thin or delayed signals slow learning. - Give it time: Don’t judge broad campaigns in the first few days. Let the learning phase play out before drawing conclusions. - Run a hybrid strategy: Pair broad targeting for discovery with custom audiences to capture known intent, and use each to inform the other. ## Measuring Success and Analyzing Performance Data Evaluate broad campaigns at the campaign level, not the segment level. Focus on blended CPA and overall ROAS rather than per-audience breakdowns. Post-purchase analytics can also reveal geographic, demographic, or contextual patterns the AI discovered organically — signals worth using in future campaigns. ## Key Takeaways Privacy regulations and cookie deprecation have made traditional targeting harder to sustain, while machine learning has made it easier to scale without relying on narrow audience definitions. Broad targeting on the open web, powered by AI and strengthened by creative diversification, is better aligned with where open web advertising is headed. Platforms like Realize reflect that shift, combining open web scale with AI-driven optimization to help advertisers grow beyond the limits of traditional targeting. ## Frequently Asked Questions (FAQs) ### What is the difference between broad targeting and interest targeting? The core difference is where audience-building happens — in your campaign settings, or inside the algorithm. With interest targeting, it happens before the campaign runs; you select specific behaviors, demographics, or hobbies upfront. Broad targeting moves that process into the platform itself, using only a few basic parameters like location or device as a starting point, while AI uses real-time signals to find the users most likely to convert. ### Why is broad targeting effective on the open web? The open web gives advertisers access to a much broader range of publisher environments than search and social alone. On the open web, users show intent in different ways depending on what they’re watching, reading, or researching. Broad targeting lets AI use those contextual cues to spot likely buyers as they engage with content, often before they’ve shown more obvious intent signals elsewhere. ### How does AI improve broad targeting campaigns? AI improves broad targeting by making faster, more dynamic decisions than manual targeting can. For every impression opportunity, it weighs a mix of live and historical signals to predict the likelihood of a response, then adjusts delivery as new performance data comes in. Over time, that feedback loop helps the system find more efficient opportunities without depending on rigid audience rules. ### Does broad targeting mean I lose control over who sees my ads? Not exactly, but control does move. Instead of specifying your audience upfront, you shape it through your creative. When you build distinct ads for different buyer personas and pain points, AI routes each one toward the users most likely to respond to that specific message. The creative becomes the targeting mechanism, which is often more precise than a manually defined segment. --- ### Modern Performance Marketing: How to Scale Smarter Beyond Walled Gardens URL: https://www.taboola.com/marketing-hub/scale-smarter-open-web-modern-advertising/ Last Modified: 2026-05-31 06:57:04 Not long ago, performance marketing was simply launching ads, tracking clicks, and manually optimizing bids to scale. In 2026, that’s no longer the case. Today, advertisers can find themselves stuck inside the closed ecosystems of the “Big 2”: Google and Meta. A May 2026 survey of 200 senior performance marketers makes this dependency concrete: 74% allocate more than a quarter of their entire budget to Paid Search, and 67% do the same for Paid Social — a level of concentration that makes diversification feel structurally impossible, not merely strategically inconvenient. It’s a situation that can come with fragmented data, burned-out ad managers, and the relentless daily updates that define manual campaign management. The brands moving out of this grind aren’t working harder, but smarter. They’re handing off the heavy lifting to agentic AI marketing platforms that operate as autonomous copilots, leaving teams free to focus on strategy that actually moves the needle. ## Performance Marketing vs. Brand Marketing in the AI Era Before we explore where modern performance marketing is going, it’s important to distinguish what makes this different from traditional brand marketing, and how the line between them is becoming increasingly blurred. Brand marketing plays a long game, building recognition, trust, and emotional resonance over time. Performance marketing, by contrast, has traditionally been about immediate, measurable outcomes like clicks, conversions, cost per acquisition, and return on ad spend (ROAS). In 2026, though, the most competitive advertisers have stopped treating these as separate disciplines. Performance marketing in 2026 is defined by a hybrid model: full-funnel strategies that combine precision performance tactics with brand-building storytelling, all measured against tangible business outcomes. Artificial intelligence (AI) is what makes this possible at scale, enabling granular optimization at the conversion level while informing upper-funnel creative decisions based on predictive data. Performance and brand marketing have merged into a single, continuous growth system. ## The Trap of the “Big 2” Walled Gardens Google and Meta remain non-negotiable pillars of most paid media strategies. But, for all the reach they offer, operating exclusively within the Big 2 walled gardens comes with a growing set of friction points that no amount of budget can outspend. Each platform has built its own native automation tools, such as Google’s Smart Bidding and Performance Max (PMax) and Meta’s Advantage+, but these tools come with a catch. They optimize within their own ecosystems, for their own metrics, with minimal transparency into the decisions being made. Data from Google stays in Google, and data from Meta stays in Meta, making cross-platform intelligence nearly impossible to assemble without significant manual effort. These are not niche tools being cautiously piloted by early adopters, either: The same survey found that 91% are currently running Google PMax at scale, and 88% are running Meta Advantage+ at scale. The industry did not trial these platforms — it moved to them wholesale, and fast, because performance outcomes validated the investment early. Again, though, scale of adoption does not mean absence of limitation, and the result is that brands are technically running “automated” campaigns but still spending large amounts of human time managing them. True automation that actually reduces workload and improves outcomes requires something that sits above these platforms, not inside them. The performance data underscores why getting this right matters. The survey found that 76% of marketers using their best-performing AI-powered platform are seeing meaningful improvements in performance, with 29% reporting significant lift and 47% reporting moderate lift. That proof of performance is what drove adoption to near-total penetration in search and social, and it’s the same standard the rest of the channel landscape is now being measured against. ## Pain Point 1: The Daily Grind of Campaign Maintenance Ask any media buyer what their day looks like, and the answer is usually the same: an exhausting rotation of downloading reports, diagnosing performance dips, tweaking bids, moving budgets, and putting out fires, before starting the whole cycle over again tomorrow. This is the campaign maintenance grind, and it’s one of the most persistent and costly inefficiencies in modern advertising. ### The Bottleneck of Media Execution Studies and firsthand accounts from performance teams consistently point to the same reality: that media buyers spend the vast majority of their time (some estimate as much as 90%) on routine execution tasks — pulling search term reports, adjusting negative keyword lists, moving budget between ad groups, responding to quality score fluctuations. These are not high-value strategic decisions but maintenance tasks that happen to require a human because the platforms haven’t made them truly autonomous. The problem isn’t only wasted time, it’s that this constant context-switching prevents teams from doing the work that actually differentiates a brand, like crafting compelling narratives, testing bold new creative directions, identifying emerging audience behaviors, and building the channel strategies that drive long-term growth. The scale of this challenge only grows with the size of the operation. The survey referenced above found that 54% cite difficulty integrating AI solutions into existing workflows as their single biggest internal barrier to broader agentic adoption. That’s a figure that rises sharply with budget size, reaching 74% among organizations spending between $1M and $4.9M per month, and 68% among those spending $5M or more. The maintenance grind persists not because teams lack the will to automate, but because fitting autonomous systems into complex, established operations is itself a significant undertaking. ### The Solution: Agentic AI as Your Copilot This is what autonomous media execution powered by agentic AI is designed to solve. Unlike traditional rule-based automation, which requires humans to define every condition and threshold in advance, an AI marketing copilot operates with real autonomy. It monitors performance in real time, identifies anomalies, executes mid-flight adjustments, and applies optimizations proactively, without waiting for a human to notice and respond. The shift isn’t about replacing media buyers, but redirecting them. When AI handles the maintenance, human expertise can be applied where it creates disproportionate value, in strategy, creative direction, and growth planning. ## Pain Point 2: Creative Production Bottlenecks and Ad Fatigue Creative is the single biggest needle-mover in performance advertising. Targeting and bidding have been largely commoditized by platform automation, but a compelling creative is still a competitive advantage. The problem? Creating enough of it, fast enough, is becoming increasingly untenable for most teams. ### The Insane Workload of Modern Creative Creative production bottlenecks are a growing crisis in performance marketing. Ad fatigue — the rapid decline in performance as an audience sees the same creative repeatedly — now sets in faster than ever. Attention spans are shorter, platform feeds move faster, and consumers have become experts at tuning out anything that feels repetitive or formulaic. On top of that, modern campaigns require creative at a scale that would have been impossible a few years ago. Different aspect ratios for different placements. Localized variations for franchise or regional campaigns. Platform-specific formats. A/B test permutations. What used to require a single designer for a single campaign now demands either a small team, or a fundamentally different approach to production. ### The Solution: The AI Creative Lab Agentic AI reframes creative production entirely. Rather than treating each new creative variation as a bespoke project, an AI creative system functions as a continuous generation engine, analyzing which elements (like headlines, visuals, formats, hooks) are driving performance, and creating new variations informed by that data automatically. When the system detects performance reduction on a creative, it doesn’t wait for a human to notice and notify a designer. It identifies the winning structural elements, generates fresh variations, and routes them into rotation, keeping the creative pipeline perpetually stocked without adding additional work. ## Pain Point 3: Fragmented Data and Reporting Chaos The promise of data-driven marketing has always been in knowing what’s working, cutting what isn’t, and investing with confidence. In practice, most performance teams are living a very different reality. ### The 10-Hour Manual Reporting Headache Fragmented marketing data is one of the most universally cited frustrations among media buyers and marketing leaders. Google has its data. Meta has its data. TikTok has its data. Each platform applies its own attribution windows, uses its own conversion definitions, and reports through its own interface. Putting these data streams into a single, coherent picture of performance typically requires hours of manual export, reconciliation, and interpretation every week — time that is both costly and error-prone. The downstream consequences are significant. Decisions get made on stale data, cross-channel interactions go undetected, and budget allocation is based on siloed platform metrics rather than true business impact. By the time a comprehensive report arrives at a stakeholder meeting, it’s already describing last week’s reality. ### The Solution: Unified Agentic Analytics Automated campaign management powered by agentic AI works at both the execution and intelligence level. A true agentic analytics system doesn’t only aggregate data from across platforms but also interprets it. It surfaces the signal within the noise, identifies cross-channel patterns that wouldn’t be visible in any single platform view, and delivers actionable narratives. The result is a shift from reactive reporting to proactive intelligence. Instead of explaining last week’s results, teams can act on insights in real time. ## Core Channels Driving Modern Performance Campaigns A complete performance strategy in 2026 spans a diverse and growing channel mix, from paid search and paid social to affiliate marketing and native advertising, and an expanding set of AI-led discovery environments where brands can appear in AI-generated recommendations and interfaces before a consumer ever opens a traditional search engine. Each channel has its own bidding logic, creative requirements, audience targeting mechanics, and attribution methodology. Managing even two or three of these channels in parallel is a significant operational undertaking. Managing all of them, with the frequency and responsiveness that modern performance demands, is effectively impossible without autonomous support. This is where agentic AI starts being an operational necessity. Consider, too, that not all channels are at the same stage of agentic maturity. According to the same survey, the channel landscape is effectively tiered. Google and Meta sit at the top as fully at-scale agentic environments, with 91% and 88% adoption respectively. TikTok Smart+ is in wide testing — 73% of marketers are actively piloting it with only 9% currently at scale, making it the next platform most likely to cross the adoption threshold. The Open Web sits at a different inflection point entirely: 44% of marketers are in active pilots and 36% are operating at scale, while 82% of organizations see AI-powered goal-based buying on the Open Web as a meaningful growth opportunity — the widest gap between intent and execution in the current channel mix. ## Essential Performance Metrics and KPIs for 2026 The metrics that define performance marketing success haven’t changed fundamentally. Return on ad spend (ROAS), customer acquisition cost (CAC), lifetime value (LTV), conversion rate (CVR), and cost per click (CPC) remain the core. What has changed is how these metrics are measured, interpreted, and acted on. Manual measurement like pulling numbers, building spreadsheets, and running calculations is increasingly a bottleneck rather than a routine. The speed at which campaigns can change, and the volume of signals being generated across channels, has outpaced what human analysis can process in time to be useful. Smart performance teams are supplementing or replacing manual key performance indicator (KPI) tracking with predictive analytics managed by AI agents, which can model not just current performance but likely future outcomes, flagging underperformance before it becomes a problem and highlighting optimization opportunities before they close. ## Key Takeaways The through-line here that connects every pain point is the same: the tools that made performance marketing manageable in 2022 are no longer adequate for the complexity of 2026. The volume of data is too high, the speed of creative decay is too fast, and the fragmentation across platforms is too severe. The strategic dimension is equally stark. The survey found that 75% of marketing leaders rate finding a performance channel that delivers incremental outcomes beyond search and social as very or extremely important — a figure that rises to 70% rating it “extremely important” among those spending $5M or more per month. The performance marketing model is not just operationally strained; it is strategically over-indexed on two saturated channels, and the people with the most budget and decision-making authority are the most acutely aware of it. Manual workflows, however effective the team executing them, cannot keep up. Realize+, the agentic layer of Realize performance platform, currently in beta, can help solve the pain points that come with walled garden advertising and bring the performance power of Google PMax and Meta Advantage+ to the open web. Rather than simply automating tasks, Realize+ functions as your autonomous campaign agent. That means you’ll benefit from continuous decision-making, execution, and adaptive strategies in real time using first-party data signals. For advertisers who have been constrained by the agentic AI-driven walled garden environment, Realize+ offers a third option, one that delivers closed-loop performance optimization across premium open web publishers, without platform bias. The shift to agentic AI is a fundamental change in operating model, from marketers-as-operators to marketers-as-strategists, with autonomous AI systems handling the execution layer. The teams embracing this model are getting back both the time and cognitive bandwidth to focus on what actually differentiates a brand — sharp strategy, compelling creative vision, and a deep understanding of the customer journey. ## Frequently Asked Questions (FAQs) ### What is modern performance marketing? Performance marketing means spending ad dollars against outcomes you can actually measure, like sales, leads, sign-ups, and revenue. What’s changed is how those outcomes are achieved. In 2026, teams have moved beyond manual bid tweaks and last-click attribution. Today’s performance marketers blend predictive AI, full-funnel creative strategy, and cross-channel orchestration into a single growth engine built for speed and accountability. ### What are the “Big 2” walled gardens? Google and Meta. The walled garden refers to how tightly each platform controls its own data, tools, and reporting. You can run ads inside them, but you can’t easily see across them. Their built-in automation features optimize for in-platform metrics with limited transparency. ### How does agentic AI solve ad fatigue? For those looking for ad fatigue solutions, agentic AI turns ad fatigue from a production problem into a data problem. Instead of briefing designers every time performance dips, the system watches for decay signals, identifies which creative elements are still resonating, and automatically builds and deploys new variations. The pipeline stays fresh without the team needing to keep up. ### Why is manual campaign maintenance a bottleneck? Platforms themselves haven’t made routine tasks hands-off. Someone still has to pull the search term reports, catch the underperforming ad groups, move the budgets, and respond when something breaks. That work isn’t strategic, but it’s ongoing. When media buyers are stuck in daily upkeep, there’s no time left to think about where the brand should be going next. --- ### Mastering Ad Headlines on the Open Web: A Performance Advertiser’s Guide to the Agentic Era URL: https://www.taboola.com/marketing-hub/mastering-ad-headlines-performance-open-web/ Last Modified: 2026-05-28 12:56:28 You’ve probably spent months fine-tuning your returns inside the walled gardens of search and social. Maybe you’ve tweaked every pixel of your Instagram carousels and bid on every long-tail keyword in the Google ecosystem. Those gardens may be well-kept, but the fences are both high and limiting. The open web, on the other hand — the endless universe of news sites, blogs, and niche publishers — remains a massive, untapped frontier for revenue growth. In there, your ad headline isn’t just a label. Instead, it’s the single most critical element determining whether a user dives into your story or scrolls right past it. The rules have changed. We’ve officially entered the Agentic Era of advertising, where AI agents aren’t just writing your copy, they’re researching your audience and negotiating your media spend in real-time. To scale today, you need to master both timeless human psychology, as well as the autonomous machines that now run the show. It’s a whole new way of thinking (both for you and the AI), but the results speak for themselves. ## What Is an Ad Headline and Why Does it Matter on the Open Web? In search advertising, you’re pretty much answering a user’s question: if someone types “best hiking boots,” they want to buy boots. Open web advertising is different, though, and often interruptive. You’re catching someone while they’re reading the morning news or a recipe for pie. Your headline is the hook that earns you a moment of their time, and it has to bridge the gap between what they were doing and what you want them to do. On the open web, a weak headline doesn’t just lower your CTR — it makes your brand invisible. ## Welcome to the Agentic Era of Advertising We’ve moved past the days of simple generative AI where you just ask a chatbot for “five catchy titles” and take a risk on the best ones. We’re now in the Agentic Era, a shift that represents the move from tools that assist to agents that act. An agentic system doesn’t wait for a prompt — it autonomously researches your target audience’s pain points, identifies trending topics on the open web, and executes campaigns with minimal human babysitting. ## Core Principles of a High-Converting Ad Headline Even with all the AI in the world, you’re still trying to convince a biological human to click. That’s where timeless copywriting principles still matter: ### Prioritize Clarity and Relevance Don’t be clever at the expense of being clear. If a user has to guess what you’re selling, they’ve already scrolled past. Make your value proposition immediately obvious. ### Trigger Emotion and Curiosity Use emotional drivers — fear of missing out, the joy of a bargain, or the relief of a solved problem. Pique interest without resorting to low-quality clickbait that burns trust. ### Use Numbers, Specificity, and Social Proof “How to lose weight” is boring. “How 4,500 people lost 10 pounds in 30 days” is a headline that grabs peoples’ curiosity. Specificity builds immediate credibility. ### Include a Strong, Action-Oriented CTA Front-load your verbs. Words like “Discover,” “Get,” or “Stop” tell the user exactly what the next step is. ## How Agentic AI Is Revolutionizing Headline Creation Autonomous advertising agents can now generate thousands of variations of a headline in the time it takes you to write one. But they aren’t just churning out junk: these agents check for brand voice consistency, character limits, and even legal compliance automatically. This allows performance advertising teams to test creative at a scale that was previously impossible. ## Adapting Your Ad Headlines for AI and Machine Discovery Here’s an unexpected ironic plot twist: you aren’t just writing for humans anymore. In the agentic shift, AI assistants are often browsing the web for the consumer. This means that your headlines need to use structured data and semantic keywords so these machine agents can easily categorize your offer as the best solution for their user. If the AI can’t understand your headline, it won’t recommend your product to the humans behind it. ## Best Ad Headline Formats for Native and Display Ads These three frameworks almost always work: ### The Problem-Solution “Struggling with ? This Is the Answer.” ### The Question “Are You Making These 3 Costly Mistakes?” ### The How-To Educational “How to Without .” ## Continuous Testing and Optimization at Scale The era of the A/B test is dying out, because traditional split-testing is just too slow for the Agentic Era. By the time you find a winner, the trend has changed. ### The End of Manual Testing We’re definitely moving toward “Multi-Armed Bandit” testing, where AI reallocates budget to the winning headline in real-time, every millisecond. ### Leveraging Predictive AI Using Natural Language Processing (NLP), AI can predict which sentiment will perform best on a specific publisher site before the first impression is even served. This is exactly where a performance platform like Realize can change the game. Realize isn’t just about placing ads; it’s a high-performance engine designed to bypass the traditional ad-tech taxes of complex exchanges. It uses deep learning to align your headlines with high-intent discovery moments on the open web. By leveraging Realize, advertisers can automate the generation of assets and use predictive bidding to ensure their headlines hit the right person at the exact moment they’re ready to engage. ## How to Choose the Right AI Tools for Ad Copy Optimization Don’t just buy a chatbot. Look for tools that support true agentic workflows. A real AI ad copy tool should: - Integrate directly with your ad networks. - Use real-time performance data to rewrite underperforming ads. - Autonomously manage dynamic creative optimization (DCO) to swap headlines based on the user’s context. ## Common Mistakes to Avoid - Clickbait fatigue: If your headline promises a miracle and your landing page delivers a sales pitch for socks, your bounce rate will kill your campaign. - Ignoring mobile: If your 80-character headline gets cut off at 40 characters on a phone, your message is lost. - Static thinking: In the agentic era, a headline that worked on Monday might be dead by Wednesday. ## Key Takeaways Mastering ad headlines today is a blend of old-school psychology and new-school technology. The open web offers a massive opportunity for those who can stop the scroll and start a conversation. By adopting agentic AI tools and platforms like Realize, you can move from manual guesswork to autonomous scaling, ensuring your brand isn’t just seen, but remembered. ## Frequently Asked Questions (FAQs) ### What is the difference between a search ad headline and an open web ad headline? Search ads are like answering someone who walked into your store and asked for a specific item — they already have the intent to buy. Open web advertising (like native or display ad headlines) is more like a billboard on a highway, where you have to interrupt their journey and spark curiosity or emotion to earn a click. That’s why open web headlines focus more on problem-solving or curiosity gaps to grab attention from someone who wasn’t necessarily looking for you. ### What is agentic AI in digital advertising? Agentic AI refers to smart systems that don’t just talk, but take action. Instead of merely writing a headline, an agentic system analyzes real-time data, identifies that your CTR is dropping on mobile devices, and independently writes and tests 10 new variations to fix it. It’s autonomous advertising that manages the grunt work of ad copy optimization so you can focus on the big-picture strategy. ### How long should an ad headline be for open web platforms? While you might have room for more, don’t forget that brevity is your friend. Aim for 30 to 50 characters. Most open web browsing happens on mobile, and anything longer risks being truncated (cut off), which ruins the impact of your hook and makes your ad look unprofessional. ### How do I select the best tool for ad copy creation in the agentic era? Skip the basic prompts and look for a platform that offers end-to-end automation. The best AI ad copy tools connect directly to your data, understand your brand guidelines, and can autonomously A/B test variations in real-time. You want a system that learns from every click and uses that data to refine the next headline it writes, ensuring constant improvement without you having to constantly correct it. --- ### From Cost Center to Investment Portfolio: Rethinking the Marketing Budget URL: https://www.taboola.com/marketing-hub/marketing-as-investment-function/ Last Modified: 2026-06-04 12:00:09 Every senior marketer knows the moment. The CFO sets down the deck, looks across the table, and asks the question that decides the next 12 months. “What, exactly, is marketing contributing here?” It’s rarely hostile. It’s almost always sincere, in fact. Most importantly, it exposes a problem no dashboard can fix. Classify marketing as a cost, and the budget conversation is lost before it begins. Cost lines are benchmarked against other cost lines, squeezed when revenue tightens, and never expanded on the back of a good brand story or a solid pipeline quarter. The structural problem is not that marketers report poorly or attribute clumsily — it’s that the category itself is wrong. The fix isn’t better reporting, it’s a different conversation, built on a different premise about what marketing is. In his OMR keynote, Tom Inbal, SVP of strategy and corporate marketing at Taboola, put it plainly. He refuses to treat marketing as a cost item and instead runs it like an investment function: capital deployed, returns measured. That single reframe changes the questions a CFO asks, the metrics that matter, and what happens to the surplus when a quarter goes well. This article walks through how to make that shift. ## Why the Cost Center Label Is Costing You More Than Budget A cost is, by definition, something a CFO is paid to minimize. That’s the job. If marketing sits on the expense side of the ledger, every conversation about the function will be shaped by the gravitational pull of cost reduction. Efficiency gains are rarely celebrated as opportunities to reinvest. More often, they’re used as evidence that the same outcome should now require less spend. The pattern shows up in predictable ways across every budget cycle: - A campaign that hits its targets at 80% of planned spend is read as a 20% savings, not as 20% of capital available for redeployment. - A team that automates its reporting and execution gets a smaller line item next year, not a bigger mandate. - An efficiency gain becomes the baseline for next year’s ask, not a reason to expand the team’s scope. The result is a self-undermining trap. The better marketing gets at running efficiently, the stronger the case becomes for cutting the budget that funded the efficiency in the first place. None of this is a Finance problem. Finance is doing exactly what it’s supposed to do. The problem is upstream, in the framing. As long as marketing is evaluated as a cost, efficiency conversations will likely end in budget cuts. The only way out is to move the conversation from cost reduction to capital deployment. From, “How much are we spending?” to, “What is the yield, and where should we deploy more?” That’s more than a change in language. It’s a reclassification of what marketing is inside the business. ### The AI Efficiency Trap The cost center framing was always vulnerable. AI makes it acutely so. In a recent Taboola survey of several hundred marketers, 98% reported actively using or testing agentic AI solutions, and 76% said they were seeing meaningful performance improvements from those tools. Those numbers are real, and they’re good news for marketing teams. Under a cost center model, they are also an existential problem. Here’s the chain of reasoning a CFO will follow, entirely rationally, if marketing is still positioned as overhead. AI is handling more of the briefing, more of the media buying, more of the creative variation, more of the measurement. The work product is similar or better. As Inbal said in his keynote, “If you’re not breaking free of the average, your CFO is thinking, ‘AI is doing 90% of this team’s job. Maybe they’re replaceable. Maybe they can be downsized. Maybe a few agents can do this work.’” In other words, the same output should be achievable with fewer people and a smaller budget. That conclusion isn’t an act of hostility, it’s simply the only conclusion available when marketing sits on the expense side of the ledger. There’s a different way to read those numbers, though. If marketing is run like an investment portfolio — with an Exploit bucket of proven, reliable activity and an Explore bucket of deliberate bets on new opportunities — then AI-driven efficiency in the Exploit bucket is not a saving. It’s capital freed up to fund Explore, which is where the outsized returns come from. The threat to marketing is not AI itself. It’s the framing that lets Finance read efficiency as a reason to cut the budget instead of redirecting it. Inbal made a related observation in his keynote. AI has compressed the variance in performance across the industry. The distribution of cost per action on the Taboola network narrowed by roughly 50% between 2023 and 2025. The average has gotten better, which means being average has gotten cheaper and easier, and being above average has gotten harder. If the response to that compression is to keep optimizing the proven layer, the team becomes more efficient and less differentiated at the same time. That’s the exact profile a CFO can replace with a few agents. None of this is an argument against AI. Inbal is clear on this point in his keynote. “Anybody not deep into AI should be looking into their resume pretty quickly,” he said. The marketers who win in this environment are the ones racing to adopt AI, not bracing against it. The question isn’t whether to use the tools, it’s whether you’re using them to optimize a cost line or to fund the bets that generate alpha. https://youtu.be/iMZB4gSqls4 ## What It Actually Means to Run Marketing Like an Investment Portfolio As Inbal emphasized in his keynote, being courageous without a system is recklessness. Being courageous with a system is how you outperform. That’s the purpose of a portfolio: making bravery operational rather than accidental. A portfolio is a very specific concept, and it’s worth being clear about what it is and what it isn’t. - A portfolio is not a mix of activities that all carry an expectation of positive return. That’s a diversified spend plan. - A portfolio is a structured allocation of capital across assets with different risk and return profiles, where some allocations are explicitly expected to lose money. A portfolio without losses isn’t a cautious portfolio. It isn’t a portfolio at all. This is the conceptual leap most marketing organizations haven’t yet made. Many marketing budgets are still built as if every line is expected to perform. Every campaign has a target. Every test is expected to validate. Every quarter is supposed to be green. That construction is what makes the marketing function look responsible from the outside, while also preventing it from generating the disproportionate returns that justify expansion. Running marketing like a portfolio means budgeting from the start with two distinct buckets and two distinct sets of expectations. ### The Two Buckets: Exploit and Explore A portfolio works because its two buckets are doing fundamentally different jobs. The Exploit bucket is the proven layer. These are the channels, creatives, audiences, and tactics where the team has historical data, predictable return profiles, and confidence in the unit economics. If the team can say, “We know what this returns,” it lives here. This is where most of the budget sits, and where AI-driven automation delivers the most value. Efficiency is where the Exploit bucket shines. Let AI run the things it can demonstrably run better than a human operator. The role of the Exploit layer is to produce stable, predictable yield, not breakthrough performance. The Explore layer, on the other hand, is where you put money behind things you haven’t proven yet. New channels, new formats, new audiences, and new creative territories. A few years ago, connected TV sat in Explore for most advertisers. Today, large language model platforms sit there. The point of Explore is not innovation theater layered over a steady Exploit operation. The point of Explore is to get there first. The bets only pay off if the rest of the market hasn’t already crowded in, because once everyone is doing it, the edge disappears. Some of these bets will fail. That’s the model working as designed. The ones that succeed generate the alpha — the return above the market average — that the Exploit layer can never deliver on its own. ### Why a Real Portfolio Expects Some Losses Most marketing leaders can handle the math of a portfolio model. What trips them up is the discomfort of planning for failures they know are coming. This discomfort isn’t a character flaw. Performance reviews, quarterly check-ins, board updates, and CFO conversations are all built around the assumption that every line item should be working. A failed test is a thing to be explained, not a thing to be expected. Under that operating logic, Explore is impossible. The first time a bet doesn’t pay off, the budget is at risk. “You expect only some of it to work out,” Inbal said in his keynote. “The part that pays off, that’s where your alpha is going to come from. That’s where you’re going to outperform, not by optimizing your Exploit.” The CFO conversation has to include this expectation explicitly, in advance. The loss rate inside Explore is not a defect of the strategy, it’s the price of access to outsized returns. If that isn’t settled at the front end of the budget cycle, every Explore failure becomes a threat to the entire allocation. ## Changing the CFO Conversation: From Spend to Yield The budget conversation is where this reframe gets real. When marketing is treated as a cost, the conversation centers on, “How much are we spending on marketing?” When it’s treated as a portfolio, it centers on, “What’s the yield on our marketing capital, and how is it allocated?” That shift requires a different reporting structure, a different set of metrics, and a different relationship between the marketing team and Finance. It’s more demanding for marketing leaders, not less. It’s also the only version of the conversation that ends with marketing being treated as a value driver. ### Replacing the Budget Request With a Capital Deployment Plan A budget request is a cost justification. A capital deployment plan is an investment thesis. The two documents look similar on the surface, but do very different work. A capital deployment plan spells out three things: - How much capital goes into Exploit, with the expected return range and the confidence level behind that range. - How much goes into Explore, with the expected loss rate, the criteria for what counts as a viable bet, and the upside if those bets land. - What the reinvestment mechanism looks like when the Exploit layer overachieves — and that mechanism needs to exist before the overachievement happens, not after. That puts the marketing leader in roughly the same conversation a fund manager has with a board. Capital’s being deployed against a stated risk-and-return profile. Some allocations are expected to outperform, some to underperform, and the portfolio as a whole is evaluated against a benchmark. That’s a fundamentally different posture than defending a cost line. ### Reinvesting Surplus: The Mechanism That Changes the Dynamic The most concrete piece of the model is what happens when things go well, as Inbal described in his keynote. “Last year, my teams overachieved on their goals,” he said. “We didn’t just kick back that extra surplus to the CFO. We came in with a business plan, and we funded additional exploration.” That single behavior is the test of whether the portfolio framing has actually taken hold. If marketing is still classified as a cost, surplus becomes a cut. The logic is often unavoidable: if the team hits its goals with less, it can hit them with less next year, too. If marketing is classified as a capital deployment function, surplus becomes fuel for the next round of bets. The same dollars, framed differently, produce opposite outcomes. This mechanism is only available to marketing leaders who’ve already done the upstream work with Finance. You can’t retrofit it after a good quarter: it has to be the default assumption built into the budget cycle from the start. ## How Realize Enables the Portfolio Approach in Practice The portfolio framework isn’t just a conceptual reframe. It needs to be auditable and reportable, because CFO buy-in doesn’t come from a good argument, it comes from your ability to show that the allocation is intentional, measurable, and tracked over time. That’s where Realize earns its place in the model. Realize+ runs as the autopilot for the Exploit layer, handling proven campaigns automatically and protecting the efficiency that generates the stable returns funding Explore. You aren’t babysitting the proven layer. The machine runs it, and you get the capacity back for the bets that actually move the needle. The Realize dashboard and autonomous budgeting tools handle the structural side. Teams can split spend cleanly between Exploit and Explore from the start of the planning cycle, track the performance of each bucket separately, and show Finance that the allocation is deliberate. The Exploit layer gets reported against efficiency and stability metrics like cost per action (CPA), return on ad spend (ROAS), and contribution margin. The Explore layer gets reported against learning velocity, bets initiated, signal strength, and the rate at which Explore bets graduate into Exploit programs. The point isn’t that Realize makes the portfolio strategy possible in theory. It makes it auditable and reportable, which is what CFO buy-in actually requires. ## Key Takeaways The CFO sets down the deck again. The same question, the same table, the same decision about the next 12 months. “What, exactly, is marketing contributing here?” This time, though, the marketing leader doesn’t defend a cost line. Instead, they present a capital allocation review. The Exploit layer is performing. The Explore pipeline is healthy. Losses are inside expected parameters. Two programs are signaling strongly enough to graduate into Exploit next quarter, and the surplus from Exploit overperformance is going into a third Explore allocation. That conversation doesn’t get cut. It gets expanded. The shift from cost center to investment portfolio is partly a financial reframe, but it’s also a career strategy. The marketers who become irreplaceable in the AI era are the ones who take ownership of outcomes, not just activities — who are willing to be measured on yield, comfortable with planned failure, and able to defend their portfolio in the language Finance already speaks. ## Frequently Asked Questions (FAQs) ### How do I decide what percentage of budget should go to Exploit vs. Explore? There’s no universal ratio. The right split depends on the maturity of the business, the competitive intensity of the category, and how much risk the organization is willing to take. The starting principle is that the Explore allocation needs to be big enough to generate meaningful signal across multiple bets, but small enough that the expected loss rate doesn’t breach a threshold Finance won’t tolerate. The Exploit layer also has to be robust enough that the team can take risks elsewhere without the whole budget feeling threatened. For teams new to the portfolio, an 80/20 split between Exploit and Explore is a defensible starting point, adjusted as the team learns its actual hit rate. ### How do we report on Explore investments without the CFO treating every failure as a red flag? This is a conversation that has to happen before the budget is allocated, not after a loss. Sit down with the CFO upfront and agree on three things: how often Explore bets are expected to miss; how much any single bet can lose before it gets pulled; and what actually counts as a win, a wash, or a failure. Once those numbers are on the table, losses that fall inside them aren’t bad news, they’re proof that the model is doing what it’s supposed to do. Try having that conversation after a bet fails, and the framing will land very differently. ### Should CFOs reclassify marketing spend as a capital investment on the balance sheet? Formal reclassification is a finance and accounting decision that varies by jurisdiction, business model, and audit treatment, and it isn’t something that a marketing leader can drive unilaterally. The more immediately actionable point is that the mental model can shift even when the accounting treatment doesn’t. Treating marketing spend as capital deployment changes the questions Finance asks, the metrics the team tracks, and the default assumption about what happens to surplus. Marketing leaders who put in the upfront work with Finance can shift the conversation immediately. The balance sheet can catch up later, or not at all. --- ### Data Transparency on the Open Web: How AI is Reshaping Performance Advertising URL: https://www.taboola.com/marketing-hub/data-transparency-open-web/ Last Modified: 2026-06-24 14:32:44 Performance advertisers moving beyond search and social face a real tradeoff when expanding to the open web: they gain massive reach, but lose the data guardrails that walled gardens provide. On search and social, each platform controls its own data end-to-end, with collection, modeling, targeting, and attribution all happening within a closed system. On the open web, data accountability shifts to you. Beyond regulatory compliance, data transparency helps you earn consumer trust and gives your artificial intelligence (AI) models the clean, consented signals they need to drive campaign return on investment (ROI). This guide explains how transparent data practices and AI-driven technology work together to help you scale automated media buying across the open web with confidence. https://www.youtube.com/watch?v=QUYxkx8KFfg ## What Is Data Transparency in Performance Advertising? Data transparency means having documented visibility into how consumer data is collected, where it originates, what permissions are attached to it, and how it’s used in automated targeting, optimization, and attribution. On the open web, you need to know which publishers and partners are contributing audience signals, what consent mechanisms are in place, and if the data meets your brand’s standards. That requires taking active ownership of your data flows, rather than relying on platform black boxes. ## Why the Open Web Demands a New Approach to Data The open web is inherently fragmented. Inventory is spread across thousands of independent publishers, data flows through a stack of third-party vendors, and attribution depends on connecting signals across environments that weren’t built to work together. That complexity creates more room for signal loss, inconsistent consent handling, brand safety blind spots, and measurement gaps. Without a proactive data strategy, those issues compound quickly and become difficult to control at scale, making transparent data governance a baseline requirement for sustainable open web buying. ## The Difference Between Data Privacy and Data Transparency Data privacy is about protecting consumer information by limiting access and preventing misuse. Data transparency is about communication: being clear with consumers about what you collect, how you use it, and why. Privacy governs how data is handled; transparency governs how data practices are disclosed and understood. AI can support both by classifying data, enforcing access controls, and monitoring consent signals in real time, helping advertisers maintain data privacy compliance across complex open web environments. ## How Data Transparency Fuels High-Performing AI Models Your AI is only as good as the data you feed it. Bidding algorithms predict conversion likelihood by processing historical signals, including which users engaged, in what context, and what they did next. When that data is consented and traceable, the model can learn from reliable patterns. When the data is opaque, stale, or poorly governed, the model learns from noise instead. That challenge is showing up across the industry: IAB's State of Data 2025 found that nearly two thirds of agencies, brands, and publishers cite data quality, data protection, and fragmentation across AI tools as their top barriers to effective AI adoption. Data lineage — the documented record of where your data came from, how it was transformed, and how it informed downstream decisions — helps mitigate those problems. That visibility gives advertisers more confidence in the signals shaping optimization, supports stronger model performance, and reduces the risk of flawed AI-driven decisions. ## Algorithmic Accountability and Data Lineage Algorithmic accountability means being able to explain why your AI made the targeting and bidding decisions it did. On the open web, where automated media-buying happens in milliseconds across thousands of sites, visibility has to be built into your architecture. Data lineage tracks audience signals from first-party data collection through consent validation, segmentation, and activation. When a performance anomaly or compliance question surfaces, lineage is how you trace it. It may not be glamorous infrastructure, but it’s what makes independent ad buying more accountable and defensible. ## Building Consumer Trust Through Clear Data Collection Consumer trust affects performance. When users understand what data you’re collecting and why, they're more likely to opt in, giving your AI better signals to work with. IAB research found that 82% of advertising executives believe Gen Z and Millennial consumers feel positively about AI-generated ads, but only 45% actually do. Clearer disclosure helps narrow that gap, and brands that do it well earn trust and end up with better data. ### Designing User-Centric Consent Experiences Consent user experience (UX) directly impacts campaign scale. Clear, easy-to-navigate preference centers that give users control can drive higher opt-in rates, resulting in broader consented audiences and stronger signals for your AI models. Regulators are also scrutinizing consent interfaces that make refusal harder than acceptance, so good UX supports compliance, too. ### The Value Exchange: Personalization for Data Consumers are willing to share their data when they feel they're getting value in return. When you’re transparent about the exchange, users who opt in are signaling genuine interest. That creates a higher-quality audience than one built from unconsented data, and your AI marketing models will reflect that in performance. ## Navigating Privacy Regulations on the Open Web General Data Protection Regulation (GDPR) and the California Consumer Protection Act (CCPA) have raised the stakes for independent ad buying. In 2025 alone, data protection authorities issued more than 330 GDPR fines totaling over 1.15 billion euros, underscoring the cost of weak data governance. For open web advertisers, the challenge is in applying privacy rules consistently across a fragmented publisher ecosystem. Your data privacy compliance framework needs to travel with your campaigns. Manually verifying consent signals across thousands of publisher sites isn’t feasible. AI makes that work more manageable by scanning publisher-level consent data, flagging non-compliant inventory before bids are placed, and triggering revalidation when user preferences change. A well-built compliance workflow keeps campaigns running continuously and eliminates the need for reactive audits whenever regulations shift. ## Implementing a Data-Transparent Architecture for Your Campaigns Start by auditing your data pipeline: map every source feeding your campaigns, identify where consent is collected, and confirm that consent status is accessible to your bidding systems in real time. Tag data at ingestion by source, consent status, and collection date so lineage is accurate from the start. Then connect that lineage to your bidding logic so your AI knows what data it’s using and where it came from. That’s what makes transparent data governance operational. Platforms like Realize are built to support that approach, combining AI-driven optimization with the controls and transparency performance advertisers need to scale on the open web. ## The Future of Open Web Advertising: Agentic AI and Data Standards The next phase of open web advertising is agentic AI: autonomous systems that can increasingly plan, execute, and adjust campaigns with minimal human involvement. IAB’s State of Data 2025 points to a near future in which AI supports the full media campaign lifecycle, from audience segmentation and media partner selection to performance forecasting. However, those systems can only operate responsibly on a transparent data foundation, so if an AI agent relies on unconsented or untraceable data, it amplifies compliance and brand safety risks at the same speed it amplifies performance. The data infrastructure you build now will shape how effectively you can adopt the next generation of advertising technology. ## Key Takeaways Data transparency is both a performance strategy and a compliance requirement. Clean, consented, well-documented data is what makes AI-driven advertising effective. Without it, bidding algorithms optimize against noise. Performance advertisers that scale on the open web are building transparent data infrastructure now: consent experiences that earn opt-ins, lineage systems that support algorithmic accountability, and governance workflows that stand up to regulatory pressure. Transparency doesn’t limit performance. It makes performance sustainable. ## Frequently Asked Questions (FAQs) ### What is data transparency in open web advertising? Data transparency means handling consumer data used for targeting and attribution in a clear, accountable way so advertisers know where it came from and consumers understand how it’s being used. ### How does AI improve data transparency for performance advertisers? AI helps automate data lineage tracking, classify sensitive information, and monitor consent signals at scale, making it easier to maintain compliance without slowing performance. ### What is the difference between data privacy and data transparency? Data privacy protects consumer information from misuse or unauthorized access. Data transparency explains what data is being collected, how it’s used, and why. For performance advertisers, transparency builds the audience trust needed to ethically collect the data that privacy laws protect. ### Why is data transparency essential for scaling AI-driven ad models? AI models perform best when the data behind them is clean, consented, and well understood. Opaque or unconsented data weakens predictions, wastes spend, and undermines ROI. --- ### How Do Advertisers Overcome Analysis Paralysis to Test More Ad Creatives? URL: https://www.taboola.com/marketing-hub/advertiser-paralysis-analysis/ Last Modified: 2026-06-10 08:08:58 If you’re working in advertising or marketing these days, you can probably remember a time — maybe even recently — when you worked on an urgent project. It might have been a creative brief that still hasn’t been approved, or an A/B test scoped but not launched. Or, the campaign you’d been focused on shipped with only two creative variants instead of the recommended 10, because production couldn’t keep up. Whatever the project at your particular company, the common theme is uncertainty. These activities needed to be completed, launched, or approved, but they all became victims of analysis paralysis. It’s a recurring problem, particularly for creative decisions, because those decisions feel high stakes, and it happens to the teams who most need to test, and test frequently. Humans often create a bottleneck in creative production, because there’s an instinct to overanalyze before acting. This isn’t a personality flaw to overcome for advertising teams, but rather a structural problem that emerges when the cost of being wrong is perceived as higher than the cost of being slow. That human bottleneck then results in less testing velocity, which leads to less optimization and worse performance results. This isn’t an unfixable problem, though. You can implement a system that scales creative output and ships more tests to market without losing creative focus. It’s even possible to make the cost of a failed test low enough that speed becomes easier and operationalized. Overcoming paralysis is possible. Read on for a blueprint. ## What Is Analysis Paralysis in the Context of Creative Testing? Decision-making is hard. In creative testing, it can be even harder, because creative work always contains some element of subjectivity. Within advertising specifically, analysis paralysis is a response to a system where every creative test involves production time, budget, and approval cycles. Every team member likely knows how hard the design team worked on a new concept or last-minute campaign addition. Failure then becomes more visible to senior stakeholders. So, advertisers fall back on the instinct to gather more data before acting. That feels safer, but it’s counterproductive: by the time the team achieves certainty about a creative direction, the cultural moment or audience signal has already passed. A competitor may have gained the upper hand or share of voice, or ads didn’t run far enough in advance of a sporting event to capture good testing data and deploy the winner. Taboola’s SVP of strategic and corporate marketing, Tom Inbal, spoke at the OMR Festival recently about this kind of stalled action. “When you encounter uncertainty, the instinct is to go do more homework. Let’s get more data, let’s do more testing,” he said. “But, you can’t analyze your way out of uncertainty.” https://youtu.be/iMZB4gSqls4 The solution to this very human problem is creating a new system to work within. “You have to come to grips with the fact that there’s no way to get to certainty fast enough to have impact,” Inbal said. “By the time you’re sure, it’s often too late. So, you have to create the system that allows you to operate despite the uncertainty, and not tell yourself that you’re going to analyze your way out of it.” ### The Real Bottleneck Is Not Creativity — It’s System Design When this uncertainty happens consistently in your organization, it isn’t because of a lack of ideas or creative talent. The problem lies with the system around the creative process, including approval chains, production costs, localization time, and the organizational pressure that backs only what is already proven. Systemic issues also spread to team size — a team with five average ideas and a two-day turnaround will always outpace a team with 10 great creative ideas but a two-week production cycle and four-person approval chain. Fixing analysis paralysis in advertising needs an operational redesign, not mindset changes. ## Why More Data Will Not Solve Your Testing Problem When creative work is underperforming for marketing and advertising teams, a common response is to do more research, gather more audience data, and run another competitive analysis. When every marketing team is running the same AI-powered tools, though, they’re performing the same analyses, arriving at the same conclusions, and running similar creative. Rather than more research, teams have to differentiate with quick action. Acting on an early signal in the data before it’s confirmed, rather than waiting for certainty, can propel your ads to success. Analysis is an important tool to evaluate the bets you’ve made quickly, but it isn’t a substitute for making those bets in the first place. ### The Certainty Trap: How Waiting for Confidence Kills Momentum You’ll likely never find the certainty you want in creative advertising work. Even worse, delaying ad testing because of uncertainty can cost the business money. For example, say a team spent four weeks refining a creative brief and testing audience hypotheses, then discovered that a competitor had already launched, tested, and iterated on a similar concept. That competitor captured the advantage because they moved quickly, not because they had superior creative. Successful teams are “very big on momentum,” said Inbal. “Whenever they see momentum, they’re putting a lot behind it quickly. They would rather lose money on a failed pilot than lose an opportunity that they think has a 51% probability of success.” ### Speed of Judgment vs. Depth of Analysis Deep analysis isn’t the only action that performance marketers can take. They can also use speed of judgment as a lever. These have different applications: deep analysis is useful for strategy, campaign architecture, and budget allocation. Speed of judgment is essential for signal detection and creative iteration. Most marketing teams overinvest in the deep analysis at the creative execution layer, but that’s where fast cycles and fast reads are more valuable than comprehensive, pre-campaign research. Start thinking about creative testing differently — the goal of a creative test isn’t to prove a hypothesis, but to generate a new signal to act on. The more of those signals you capture, the more testing data you’ve just gathered. ## Building a System That Makes Testing Safe How can performance marketers and advertisers shift to a testing- and risk-safe system, then? Analysis paralysis can be cured by courage, but not in an abstract sense. Instead, it’s ensuring that the system in place can structurally reduce the perceived risk of each test. That type of system creates a permission structure for movement and action. When the team has agreed-upon rules about how to evaluate a test, what counts as a signal, and when to kill or scale a tested item, then the individual launch decision is much lower-stakes. This type of effective creative testing system has three key components. ### 1. Treat Creative as Data, Not Art Creative advertising work brings together human imagination and business guardrails in a way that can’t be replicated. Your design team’s creative quality will ideally meet testing in a way that generates signals and further iteration. In a high-velocity testing environment, each creative variant serves as a single hypothesis that either generates a useful signal or it doesn’t. Either of those outcomes is valuable. If it overperforms, you can double down for continued success. A variant that underperforms provides valuable insight as well. Reframing quick testing in this way makes it easier for your team to launch imperfect creative quickly, learn from it, then iterate. It’s much more effective than polishing indefinitely before going live. ### 2. Set a Testing Cadence and Hold It In Inbal’s keynote speech, he noted the pattern he’s seen among top performance advertisers: they set a firm quarterly target for the number of new creative concepts, formats, or channels tested, and organizational accountability for hitting that target. “They’ve found a way to have a mindset of risk-taking built into how they operate,” Inbal said, later adding that, “Being courageous with a system is how you outperform.” With regular targets and a regular cadence in place, there’s now no need for each test to be individually justified. Testing is the default mode, not the exception. Try this as a starting framework for your own organization: set a minimum number of new variants to test per campaign cycle, and set a maximum evaluation window before a kill-or-scale decision is made. ### 3. Agree on a Master Metric Before You Launch Inbal also found that top-performing advertising teams choose a master metric. “They compare everything in a very simplified way, all the different interactions,” he said. “They know it’s not perfect, but it helps them cycle through a lot of testing fast, and make up their minds and move.” Conflicting interpretations of which metric matters can stop testing velocity in its tracks. When analysis paralysis happens at the point of decision, having a simplified, directionally correct master metric in place provides a focus point for everyone involved. When a team has a pre-agreed, imperfect-but-fast metric for evaluating performance, the post-launch evaluation becomes fast and unambiguous. Teams can cycle through creative variations quickly and make up their minds without a lot of debate. Master metrics can vary between industries, but include customer acquisition cost (CAC) for e-commerce or retail; daily active users or total time spent for media and content platforms; and gross merchandise value (GMV) for marketplaces. Generally, look for the specific action where a user realizes the core value of your product as a starting point to establish your master metric. ## How Realize Lowers the Cost of Every Creative Test Realize takes into account all these structural barriers to creative testing. Two key platform capabilities make it easier to avoid analysis paralysis: ### Agentic AI creative generation and localization These capabilities dramatically reduce production time and cost per variant, so even small teams can test significantly more without increasing production overhead. ### Autonomous, real-time A/B testing and creative rotation These features remove the need for manual monitoring and intervention during a test, which can save many hours and work faster than a human team is capable of. Together, the effect of these Realize features is that the cost of failure for any individual creative test drops significantly. Then, the risk calculus changes, making high-velocity testing rational, not reckless. This takes the pressure off performance marketers and helps reduce the chance of analysis paralysis within the system. It operationalizes the entire system as one that encourages frequent testing. ## Key Takeaways With a testing system in place, day-to-day advertising work looks different. A scenario like one where the creative took six weeks to ship now looks like six weeks where that team captured zero signals. How many useful, informative signals from the data might have shown up and informed the creative strategy with the new system in place? That one missed opportunity may have ultimately cost the business a lot. The choice isn’t between being careful and careless, or no-risk vs. high-risk testing. It’s between a slow system and a fast one. Realize makes fast possible without increasing risk or overhead, leading to more frequent testing for better creative success. Creative velocity isn’t a nice-to-have, but an essential, strategic part of your business’ long-term competitive position. ## Frequently Asked Questions (FAQs) ### How many creative variations should we be testing per campaign? To get the number of creative variations that works for your business, start by identifying your current velocity and then setting a target that’s 2-3x higher. The right number is the one that generates a statistically useful signal within your evaluation window. Most teams should probably be testing more than they are now, and agentic AI makes it possible to test 10x more ad variations without increasing production overhead. A recent Taboola survey found that 76% of marketers see meaningful improvements in performance using agentic solutions. ### How do we decide when to kill a creative test vs. give it more time? The most common source of testing inertia is post-launch deliberation about whether a variant had enough time in market. In an operationalized testing environment, it’s important to set guardrails for when to kill a creative test or give it more time. Get to know the concepts of both a pre-agreed evaluation window and a kill threshold, and make sure they’re defined and communicated to teams before the test launches, not after. It’s also useful to consider the master metric framework: when there’s one agreed signal to watch, the kill/not-kill decision is much easier to make. ### Won’t testing more creative variations dilute our brand consistency? This is a common concern from brand teams — that testing multiple creative variations will appear to users as inconsistency. However, creative testing at the performance layer doesn’t require testing brand identity. Typically, creative testing variables might include format, message sequencing, visual treatment, and the calls to action (CTAs). They don’t usually include logo placement, tone of voice, or other brand treatment details. A well-designed testing system should include guardrails that define what is in scope for iteration and what is always the same. --- ### How Advertisers Use Agentic AI Goal-Based Optimization to Scale Budgets Profitably URL: https://www.taboola.com/marketing-hub/goal-based-marketing/ Last Modified: 2026-06-24 08:16:12 Performance advertisers can no longer rely on vanity metrics like clicks, impressions, and engagement rates to measure ad campaign success. While these metrics may look strong inside ad platforms, they often fail to reflect actual profitability. Custom goal-based ad optimization helps advertisers align campaigns with real business outcomes by focusing on lower marketing funnel metrics such as purchase CPA, ROAS, and revenue. Combined with rule-based performance systems, advertisers can automate campaign decisions in real time. Using automated campaign rules for stop loss ad optimization, target CPA automation, publisher blocking, and automated budget scaling, brands can protect spend, scale winners faster, and create a continuous optimization loop that improves efficiency around the clock. ## What Is Custom Goal-Based Ad Optimization? Custom goal-based ad optimization is the process of optimizing campaigns around business-defined profitability goals, instead of relying entirely on platform algorithms. Native ad platforms often optimize toward broad conversion signals that may increase volume, but fail to generate profitable customers. By implementing custom optimization rules tied to actual business economics, advertisers can optimize campaigns based on margins, revenue, customer value, and target acquisition costs. This creates a more controlled and sustainable approach to performance advertising that prioritizes profitability over vanity metrics like click or “like” tallies. ## Why Performance Advertisers Need Rule-Based Performance Manual optimization is too slow for modern advertising environments, where performance can shift hourly (or faster) across multiple campaigns and placements. For advertisers focused on lower funnel metrics, delayed action often leads to wasted spend and missed scaling opportunities. Rule-based performance solves this by automating decisions based on predefined business logic. Through agentic ai ad optimization, advertisers can instantly pause underperforming campaigns, prioritize and scale up profitable ones, and protect budgets without constant manual oversight, allowing media buyers to focus on strategy and testing (and, maybe, even step out for coffee now and then). ## Setting Benchmarks for Lower-Funnel Metrics Before turning on automation features, advertisers need to define what success actually looks like. This starts with understanding core unit economics, including margins, average order value, fulfillment costs, and customer acquisition limits. From there, advertisers can establish target CPA automation thresholds, minimum conversion rates, and profitability benchmarks that guide every automated decision. These benchmarks become the foundation for all automated campaign rules and optimization logic. ## 4 Custom Rules to Automate Your Ad Campaigns The most effective ad campaign automation strategies rely on simple “if/then” logic that continuously manages budgets, placements, and creatives based on performance outcomes. These custom optimization rules allow advertisers to react instantly to changes in campaign performance while maintaining strict profitability controls across every stage of delivery. ### 1. Implementing Stop/Loss Rules to Prevent Wasted Spend Stop loss ad optimization rules act as automated safety nets that pause campaigns once spending exceeds acceptable limits without generating conversions. For example, a rule may state: “If spend exceeds $250 with zero purchases, pause the campaign.” These systems provide critical ad spend protection by stopping inefficient campaigns before wasted spend escalates. ### 2. Automating Budget Scaling for Winners Automated budget scaling helps advertisers increase spend on profitable campaigns without waiting for manual approval. This allows winning campaigns to grow faster while maintaining controlled risk. A common example is: “If purchase CPA is below $5 for three consecutive days, increase budget by 20% up to a set cap.” This approach supports sustainable scaling while protecting overall profitability. ### 3. Real-Time Publisher and Site Blocking Not all placements generate quality traffic. Some publishers may drive high impressions and clicks while producing little to no conversion value. Using custom optimization rules, advertisers can automatically block sites or apps that exceed spend thresholds without meeting conversion goals. This improves overall campaign efficiency and reduces wasted budget across low-quality inventory sources. ### 4. Pausing Underperforming Creatives Automatically Creative fatigue can quickly reduce campaign performance if weak ads continue receiving spend. Automated rules help advertisers identify and pause declining creatives in real time. For example, advertisers can pause ads when CPA rises above target thresholds or conversion rates fall below acceptable benchmarks. This ensures algorithms continue prioritizing the strongest-performing creative assets. ## Building a Continuous Optimization Loop Automation should not be treated as a one-time setup. Successful advertisers continuously review rule performance, analyze execution logs, and refine optimization logic over time. A strong continuous optimization loop includes testing new creatives, adjusting scaling thresholds, refining placement exclusions, and updating rules as campaign economics evolve. This keeps automation aligned with long-term business goals and changing market conditions. ## Key Takeaways Custom goal-based optimization allows advertisers to focus on profitability instead of surface-level platform metrics. By aligning campaigns with real business economics, brands can make smarter decisions based on hard performance data, rather than guesswork. Through rule-based performance systems, advertisers can automate stop/loss protection, publisher blocking, budget scaling, and more, to improve efficiency 24/7. Combined with a strong continuous optimization loop, these strategies create a more scalable and profitable performance advertising framework. ## Frequently Asked Questions (FAQs) ### What is a custom rule in ad optimization? A custom rule is an automated condition that triggers a predefined action when specific campaign criteria are met. For example, a rule may pause a campaign if CPA exceeds a target threshold or increase budgets when ROAS improves. These automated campaign rules help advertisers manage campaigns continuously without relying on manual monitoring, improving both efficiency and consistency across ad accounts. ### How does rule-based performance differ from platform algorithms? Platform algorithms generally optimize for broad delivery goals like clicks, engagement, or conversion volume. While useful, they may not always align with real profitability targets. Rule-based performance uses advertiser-defined constraints based on actual business economics and lower funnel metrics. This gives advertisers more control over campaign profitability and budget management. ### What is a stop/loss rule in digital advertising? A stop/loss rule is an automated safeguard that pauses campaigns, ad sets, or placements when spending exceeds a defined limit without producing conversions. These rules are commonly used for ad spend protection because they react instantly to poor performance and prevent campaigns from wasting budget. ### How often should automated optimization rules run? Automated optimization rules should ideally run continuously so campaigns can react immediately to performance changes throughout the day. Real-time monitoring helps advertisers maintain target metrics, protect budgets, and scale profitable campaigns faster by eliminating delays caused by manual oversight. --- ### AI Governance in Marketing: Real-Time Decision Logging Explained URL: https://www.taboola.com/marketing-hub/ai-governance-in-marketing/ Last Modified: 2026-05-28 11:18:28 There’s never a dull moment for those working in marketing, and figuring out how to safely use AI is the current top challenge for a lot of teams. After enterprise AI marketing adoption surged in the past few years, marketers are now seeing what can happen without AI guardrails. In response, enterprise advertisers and agency buyers are slowing down on their use of unchecked, black box AI. They’re disabling automated AI features to retain operational control, create audit trails, and generally shift toward a robust governance layer. That AI governance layer includes real-time decision logging and confidence scoring to make sure AI tools are justifying their reasoning every step of the way. ## The Agency Buyer’s Dilemma: Why Marketers Turn Off Unchecked AI The rise of agency buyer AI skepticism is well-founded. Businesses have been fined, lost crucial brand trust, and been taunted online for their AI usage. AI can increase speed and volume dramatically, but without careful human control, ungoverned AI can cause massive problems. Unchecked AI risks are a big liability for enterprises across a few different areas of concern: ### Compliance Legal standards around the world include the GDPR, the CCPA, HIPAA, and more, all of which govern data protection and privacy for companies operating on the internet. AI companies have been fined under the GDPR because they couldn’t show documentation of how data was used — a common issue when using AI tools that can’t create an AI audit trail. New AI-specific compliance regulations include the EU AI Act, which takes full effect in August 2026, and mandates documented controls for high-risk AI systems. ### Credibility Trust is a cornerstone of successful AI adoption. While AI has quickly become commonplace, users want to know that brands will connect with them as humans. Transparent and accountable AI governance can maintain or build confidence for users, stakeholders, and regulators. This is particularly important for sectors like finance and healthcare, but all verticals have to bring trusted products and ads to market for long-term success. ### Risk Mitigation While compliance standards can enact financial and legal consequences, there are plenty of uses of AI that don’t violate standards, but still pose business risks. A partner might ask how the marketing team created joint ads, for example, but if they used AI without governance or guardrails in place, there’s zero visibility or history available. ## Shadow AI vs. the Governance Layer: Why Static Policies Aren't Enough A written AI policy might already be in place for a lot of enterprises — something the legal team asked for in the earlier days of AI tools. But, marketing teams have likely already explored further use cases, various AI tools and models, and created work with AI far outside of that policy. Those static rules and documentation can easily lead to shadow AI marketing. AI governance in marketing has to be embedded into the actual workflow for it to work. Marketing teams move quickly, with dynamic work that changes day to day or week to week depending on ad performance, audience needs, open web trends, and more. To be able to show AI history and maintain continuous visibility, true AI governance should be an embedded operational layer inside of marketing tools. ## Real-Time Decision Logging: The 2026 Baseline for Enterprise AI So, how are marketing and advertising teams and agency buyers tackling the need for compliance? Many are turning to books-and-records compliance for AI, which refers to a type of regulatory compliance. For financial services firms, books-and-records compliance now requires them to track all AI-generated content as official records. That has created a high bar that treats AI outputs like emails or memos, meaning that they have to be available during audits and securely stored for five or more years. Financial services might be the first to require this compliance, but every AI action in 2026 should have an immutable audit trail that tracks the prompt, the data used, and the model’s rationale, including real-time decision logging, which records the choices made by a system or algorithm at the moment they occur. AI audit trails show to internal stakeholders and regulators why a decision was made the way it was, and what the resulting actions were. ## Demystifying the Black Box With Confidence Scores AI audit trails should include AI confidence scores, too. These scores allow an AI model to report its own uncertainty, generally as a numerical value or percentage from 0% to 100%. It’s a metric for AI reliability, with higher scores reflecting stronger evidence to support AI output, and lower scores showing uncertainty that might need human review. AI confidence scores are a self-reporting mechanism that can serve as a safety net for agencies and businesses to triage and avoid risk. It also removes any concerns around black box technology platforms, since human operators get involved whenever the AI’s recommendation is low-confidence. ## Applying Human-in-the-Loop (HITL) for Safe Scaling Human review of AI outputs, or human-in-the-loop (HITL) marketing, brings confidence scores into daily reality for advertisers and marketers. A team might set specific thresholds for confidence scores, e.g., if the AI confidence score drops below 85%, then the decision automatically routes to an agency buyer. For agencies moving super quickly, this is a way to automate execution with AI while also providing essential human oversight. ## How Realize+ Embeds Governance Into Marketing Automation Using AI responsibly in marketing automation is an essential part of trustworthy, explainable AI marketing. Agencies and performance marketing teams know that AI can be a strong partner for scaling quickly, but they need a transparent audit trail, too. Too many performance marketing platforms have added AI features in response to trends, but haven’t yet developed the capabilities to document actions and create visibility. Realize+ has built on its modern foundation to provide an explicit audit trail of its reasoning. It’s an antidote to unchecked AI and black box platforms, bringing radical transparency to skeptical agency buyers along with other teams and partners that need to see AI documentation. The end result: Buyers can take full advantage of AI’s speed and scale while maintaining the control they need. ## Key Takeaways Agency buyers and marketing teams can move fast, create volumes of ad variations, and target audiences like never before with AI. But, the era of blind trust has ended, with verifiable proof now essential for enterprise AI use to ensure brand trust, credibility, regulatory compliance, and risk mitigation. Performance platforms like Realize+ can now provide real-time decision logging, AI confidence scores, and AI audit trails to allow users to scale AI securely and transform governance into a competitive advantage. ## Frequently Asked Questions (FAQs) ### What is AI governance in marketing? AI governance in marketing is the entire framework of technical controls, processes, and policies in place to make sure AI is used responsibly and in alignment with brand and regulatory standards. This includes decision logging, confidence scoring, and the creation of audit trails, all of which bring transparency to a company or agency’s AI usage. Incorporating AI governance into marketing workflows can set rules for data usage, brand guidelines, and other guardrails, to prevent risk. ### Why are agency buyers skeptical of unchecked AI? Agency buyers hold a lot of responsibility, with accountability for brand safety and ROI. The use of unchecked or “black box” AI makes it impossible to perform audits, discover how decisions were made, or see how errors occurred. Agencies frequently disable the use of unchecked AI tools to avoid reputational damage or costly compliance violations. ### How do confidence scores improve AI governance? Confidence scoring reveals exactly how certain an AI model is about its specific output. High scores lead directly to automated actions, while low scores send the decision to a human for review. Confidence scores improve AI governance significantly, since human users of AI can set thresholds to mitigate operational risk while still taking advantage of AI’s speed and scale. ### What is real-time decision logging? Real-time decision logging refers to the practice of keeping an audit trail for every AI-generated action made by a system or algorithm. This decision logging records the inputs, the model’s logic, and the final output, all of which create an immutable audit trail for organizations to retrace and use to justify AI decisions after they occur. --- ### The Opportunity Arbitrage: How Marketers Provide the Alpha AI Can’t Find URL: https://www.taboola.com/marketing-hub/how-marketers-stand-out-in-ai-era/ Last Modified: 2026-06-04 09:37:58 Every performance marketer is asking the same question right now, even if most won’t say it out loud: Is AI going to take my job? The fear isn’t unreasonable. Every quarter brings another agentic capability, another piece of the workflow that once required a human and now doesn’t. Briefs, bids, creative variations, performance measurement — all of it is being handled by tools that didn’t exist three years ago. That said, it’s still not the right question. AI is already doing parts of the job. The real question is whether AI can replace the specific kind of value the best marketers bring to the table. As Tom Inbal, SVP of strategic and corporate marketing at Taboola, put it in his recent OMR keynote, AI doesn’t have style. It doesn’t have accountability. It doesn’t have a team it trusts. That might sound like a mere pep talk, but it isn’t. Those are the three things still separating the marketers who outperform from those who don’t. This article covers what AI can’t replace, who’s still generating real returns, and how you can land on the right side of the shift. ## What AI Can Do — and What It Cannot There’s a clean line between where AI adds value and where it starts to fall short. AI is excellent inside known parameters. It optimizes well when the variables are defined and the success criteria are clear. It scales proven playbooks faster than any human team, and it reduces variance, which is why even average operators are getting decent results now. In his OMR keynote, Inbal shared a stat that should stop every performance marketer in their tracks. Across Taboola’s network, the variance in cost per action (CPA) between campaigns has compressed by roughly 50% over the last two years. The average improved. The gap between average and below-average shrank. That’s AI doing what it does well. What AI doesn’t do well: - Identify opportunities before the data confirms them. - Make a bet that depends on trusting a specific team member’s instinct. - Read a cultural or market shift that isn’t yet in the dataset. - Decide how much risk is appropriate given the business context, the politics inside the company, or the CFO’s mood this quarter. Those decisions require judgment, and judgment is the part of the job that compresses the slowest. ## AI as Autopilot: The Exploit Function Every healthy marketing portfolio has an Exploit layer. It’s the channels, tactics, and creative approaches that are tested, proven, and reliably profitable. This is where AI shines. When the signals are known and the variables sit within a defined range, an automated system is a better operator than a human. It’s faster, more consistent, and never gets tired around hour three of dashboard review. Inbal’s framing in the keynote splits any healthy marketing budget into two layers: Exploit and Explore. The Exploit layer is where automated bidding, budget pacing, performance monitoring, and creative rotation should run on autopilot. It’s also the layer where Taboola’s Realize+ operates with full autonomy — not because humans are being demoted, but because humans were never going to win that race. The point isn’t that machines are replacing people in the Exploit layer. The point is that people shouldn’t have been spending their hours there in the first place. If your team’s identity and time are still anchored to operational management of the Exploit layer, that’s the part of the job that’s most exposed. It’s also the part where your marginal contribution is smallest. ## Where AI Reaches Its Limit: The Explore Function The Explore layer is different. This is where new channels get tested, new creative directions are bet on, and new audience theses are tried before the data is conclusive. It’s the domain where human judgment is irreplaceable, not because AI lacks processing power, but because the relevant signals aren’t in the data yet. In his keynote, Inbal told a story about a recent campaign his team launched. “A couple of years ago, this campaign would never have been made,” he said. “It was too risky. We’re a public company, billions on the line. We wouldn’t have taken the chance.” What changed, he continued, wasn’t the appetite for risk, but rather, the cost of testing, adding that, “AI pushed us and enabled us to be braver.” The idea still came from a human. AI just made it cheap enough to try. The Explore layer, then, is where alpha lives. It’s not an optimization problem. It’s a pattern-matching problem that requires: - Lived experience in the category. - Cultural fluency that isn’t yet in training data. - The willingness to act on incomplete information. AI can help validate or kill a hypothesis once a human puts it on the table. It can’t form a hypothesis to begin with. https://youtu.be/iMZB4gSqls4 ## The Three Things That Make a Marketer's Contribution Irreplaceable In Inbal’s keynote, he named three specifically human attributes that AI doesn’t replicate. These aren’t aspirational, they’re observable patterns in the marketers who consistently outperform. ### Style: The Distinct Point of View That Cannot Be Prompted Style in marketing isn’t aesthetic preference, it’s the accumulated sense of what a brand’s audience will find compelling, built from category knowledge, cultural fluency, and direct experience with what has and what hasn’t worked before. A well-prompted AI can produce creative that’s technically competent and audience-appropriate. What it can’t produce is creative that reflects a real point of view about where the culture is going next. That conviction is what makes some creative memorable and most of it forgettable. There’s no shortcut to it. It comes from years of paying attention to what actually gets a response. ### Accountability: Owning the Outcome, Not the Process There’s a difference between executing a decision and owning one. AI does the former. It can’t do the latter. Owning a decision means facing the consequences when things go wrong, and getting credit when they go right. Both of those things compound over time into something useful: a track record. The marketers who are trusted with the next big bet are the ones who have been right or wrong about previous bets, and can tell you why. That’s not a soft skill, it’s the basis of every budget conversation, every promotion, every shot at running a bigger team. ### Team Trust: Backing the Right Instinct at the Right Moment Team trust is probably the most overlooked piece of this. A lot of the best marketing decisions don’t come from a clean dataset — they come from a leader saying yes to a team member’s read on something before the numbers are in. That happens because the team member has been right before, or because they know the category in a way no one else on the team does. In some cases, the leader has simply learned over time that this person’s instincts are usually worth the risk. None of that lives in a tool. It lives in working relationships that are built over years. The person who called something correctly last quarter gets more room to call the next one. That’s what makes a good team so powerful. ## From Platform Operator to Growth Architect: Redefining the Role The job description that’s emerging from all of this has a name: Growth Architect. A platform operator works inside the system. They optimize what’s running, watch the metrics, and keep the campaigns moving. A Growth Architect builds the system itself. This professional decides where capital goes, what level of risk each part of the budget carries, where the team should be hunting for opportunities the dashboards aren’t yet showing, and what counts as a win at the portfolio level. The Growth Architect runs marketing the way a fund manager runs a portfolio — with a mix of safe, predictable returns and higher-risk bets that aren’t all expected to land. The marketing budget stops being a spend plan and starts being an investment thesis. A few things shift when you start working this way: - Finance conversations look different: Instead of walking through last week’s CPA fluctuations, you’re explaining why a slice of the budget is funding tests that won’t all pay off — and why that’s the point. - The definition of success looks different: No one’s asking whether every line item returned positive return on ad spend (ROAS). They’re asking whether the portfolio, as a whole, beat what your competitors are pulling off with the same tools. - Your calendar looks different: Mondays aren’t for tweaking bids anymore. They’re for spotting the next channel or audience nobody has priced in yet. ### What a Growth Architect Does With Their Time When automation is doing the work in the Exploit layer, and a real testing program is feeding the Explore layer with fresh signals, your week opens up. The hours go to decisions only a person can make: - Finding the next channel or audience that’s still undervalued. - Building the internal pitch for a new bet — the slides, the numbers, the story. - Staying close to the CFO and CEO, so when it’s time to ask for funding, you already have the relationship. - Coaching your team to make better calls under uncertainty, not just better-optimized campaigns. Now, picture the platform operator’s week alongside that: tweaking bids, rebalancing budgets, scanning creative reports, pushing approvals through Slack, and fixing broken UTMs. Those two weeks aren’t producing the same value. One generates alpha. The other gets you a little closer to the middle of the pack. ## How Realize Amplifies the Human Edge This is the role Taboola’s Realize platform is built to support. The pitch is straightforward: Realize isn’t here to replace the marketer, it’s here to take the operational grind off the marketer’s plate so they can spend time where their contribution actually matters. Realize runs the Exploit layer for you with autonomous bidding, creative rotation inside proven parameters, localization across markets, and ongoing performance tracking. Realize+ extends that further, taking full ownership of the optimizable, repeatable parts of the workflow. You don’t give up control. You give up the busywork. The bandwidth, the hours, the team’s attention — all of it is redirected to the Explore layer, which is where alpha actually comes from. A marketer using Realize isn’t a platform operator. They’re a Growth Architect who finally got their week back. ## Key Takeaways Now we circle back to the initial question: Is AI going to do your job? Parts of it, yes. The execution, the optimization, the scaling of what’s already working. That’s being absorbed, and the absorption will continue accelerating. What isn’t being absorbed is the part of the job that requires you to make a call under uncertainty, hold conviction when the data isn’t conclusive, back an idea before it’s safe to back it, and own how it turns out. The marketers who define themselves by their command of the execution layer are the ones who should be worried. The ones who define themselves by the quality of their bets — by the alpha they pull out of a market where everyone is using similar tools — are the ones who are about to become harder to replace, not easier. As Inbal said in his keynote, this isn’t a forecast. It’s the current state of play. The only thing left to decide is which side of it you’re on. ## Frequently Asked Questions (FAQs) ### If AI is getting so capable so quickly, how long will the human edge in marketing last? This is a fair concern, and there’s no point in pretending otherwise. AI will keep getting better at more things, and some of the work that requires a person today won’t require one in two years. But, the underlying need for a human layer above the algorithm isn’t disappearing. Markets will keep being uncertain. Culture will keep moving faster than the training data. Real accountability will keep belonging to people who bear consequences. The specific tasks shift, but the structural role — making bets, owning outcomes, reading what the data hasn’t caught yet — sticks around. ### How do I make the case internally that my team should be expanded, not reduced, as AI automates more tasks? Start by changing what your team gets measured on. If your value to the company comes from how much execution you crank out, you’re going to lose that argument, because the company can run that volume with fewer people now. If your value comes from the quality of the bets you’ve made and the returns those bets generated, that’s a different conversation. Track Explore outcomes separately from Exploit performance. Attribute revenue to the strategic calls your team made, not just the campaigns it ran. Once the numbers tell that story, the headcount question stops being about cost. ### What skills should performance marketers be developing to stay relevant as AI expands? Three human skills are doing the heaviest lifting in performance marketing now. The first is strategic risk-taking — building a thesis about something promising and putting money behind it before the data has fully proven it out. The second is portfolio thinking — running your budget as a mix of bets, some safe and some speculative, instead of a flat plan to be optimized. The third is talking to Finance in their language — capital, yield, return on investment, business growth — so the work is funded properly. None of these are accidentally AI-resistant. They’re what the Growth Architect role requires. --- ### Is Alpha Decay Inevitable in Performance Marketing? URL: https://www.taboola.com/marketing-hub/alpha-decay-performance-advertising/ Last Modified: 2026-06-08 12:22:06 For senior marketers and growth leaders, AI has been an incredible tool for scale. The only problem: it’s been an incredible tool for every other marketing and growth team to scale, too. Data shows that cost per action (CPA) variance shrank by about 50% between 2023 and 2025 on Taboola’s network, according to Tom Inbal, SVP of strategic and corporate marketing at Taboola, who spoke at the OMR Festival recently. His observation points to the uncomfortable truth that the more everyone adopts AI, the harder it becomes to stand out. Realize data also found that 98% of marketers are actively using or testing agentic AI to run parts of their workflow. So, while AI is the best tool you can use to grow your performance campaign success, it’s likely also the reason you feel like performance is plateauing. Your particular AI stack on its own won’t make you stand out, because everybody is using the same tools. In this article, I’ll explore the concept of alpha decay, why it’s accelerating for performance advertisers, and what the outperformers are doing differently. ## What Is Alpha Decay and Why Does It Matter to Marketers? In this context, alpha decay is a term taken from quantitative finance. It refers to the measurable gap between your performance and the market average — essentially, how good you are as compared to the average. When thinking of that CPA variance, for example, if your advertising platform’s average CPA is $25 and yours is $18, that gap is your alpha. This concept has come into play as more advertisers use the same AI tools to optimize toward the same signals, which shrinks the gap between the top and average performers. “If you’re thinking that your AI stack is your edge, you’re probably wrong,” Inbal said during the keynote. “It’s not really going to make you stand out. Everybody has the tools. Everybody has access to the same answers.” ### Why Alpha Decay Applies to Advertisers Now In quantitative finance, algorithmic trading eroded the edge of early quantitative funds. The alpha strategy is one that can outperform the market by taking calculated risks. Today, as AI democratizes bidding, targeting, and creative optimization, the same dynamic plays out. Alpha decay takes place when performance marketing teams lose their edge, because their competitors have all gained that exact same edge. ### What the Data Shows Is Actually Happening on the Open Web The Taboola data Inbal cites shows how alpha decay is entering the performance marketing arena. Engagement campaigns show a 52% variance reduction, while reach campaigns show a 50% variance reduction. For purchase objective campaigns, it’s a 40% variance reduction. “AI makes being decent much easier,” Inbal said. “You can do the average easier, but it works both ways. It is that much harder to be an overperformer.” On top of this, 76% of marketers report meaningful performance improvements from using agentic AI solutions. Looking at the data points in combination, it reflects the trend of AI raising the floor while compressing the ceiling. AI makes poor performance less likely, and exceptional performance harder to achieve at the same time. https://youtu.be/iMZB4gSqls4 ## Why Standard AI Optimization Accelerates the Problem Alpha decay is a structural challenge. Information asymmetry, which is what gives most marketing teams their edge, has already been eroded by shared AI infrastructure. As per Inbal’s example in his speech, if every chef at the fish market has a sophisticated AI scanning tool to scan all available fish and recipes, the price discovery becomes close to perfect. The opportunity for sellers to differentiate on pricing or other advantages disappears. Optimizing harder within the same system isn’t a way to escape the alpha decay issue. “Alpha decay is happening in our market,” Inbal said. “It means that the average is the trap. It means that if you’re not breaking free of the average, your CFO is thinking this team is replaceable. So, how do you break free of that trap of the average?” ### When Everyone Runs the Same Playbook This may sound familiar to many performance marketers working in advertising firms: your competitors are running similar creatives and using similar targeting logic, and seeing similar CPAs. If 98% of marketers are actively using or testing agentic AI, the tools stop being a differentiator. Near-universal adoption of the technology means it simply becomes table stakes. You’ve likely been asked whether and how you’re using AI by your leadership team, but the new question now becomes: “What are you doing that AI cannot replicate?” ### Why the Instinct to Do More Analysis Makes Things Worse Performance marketers are data-driven. When uncertainty starts ramping up, it’s common for marketing teams to gather more data and perform even more testing before taking action. It’s a rational response, but it will backfire. If every team runs the same analysis with the same AI tools, they’ll come to the same conclusions and bid on the same opportunities. When a team does finally determine what action to take, the window has often closed, so what’s the new way to tackle performance marketing goals when all your competitors are doing the same thing? It isn’t doing more and more analysis — it’s creating a better system for taking action despite uncertain conditions. As Inbal puts it, “As marketers, if we don’t exercise the option to be braver, we will be replaced.” ## What the Outperformers Are Doing Differently From Inbal’s work with top-performing advertisers across insurance, health, home and garden, and credit cards, there’s a common thread that drives their success: operationalized courage. Beyond building a culture of risk tolerance, these high performers have built a system for being bold. They’ve developed ways to find new value. “Being courageous without a system is recklessness,” said Inbal. “Being courageous with a system is how you outperform… For me and my team, we had to come to grips with the fact that we couldn’t analyze our way out of uncertainty.” This operationalized courage, or systematized risk-taking, is a repeatable process for finding and backing high-potential opportunities. It isn’t the recklessness of throwing money at random ideas, or the excessive diligence of extra testing or data-gathering that delays decisions and actions. “AI is going to favor the bold,” said Inbal, “and the question is: Are you going to be one of the bold and one of the brave or not?” ### Developing a Portfolio Mindset of Exploit and Explore These bold outperformers often approach their marketing and advertising work with the Exploit vs. Explore framework. It’s a structural backbone that allots space for both of these activities. Exploit includes the proven, high-confidence activities that fund exploration without putting the business at risk. Explore includes the deliberate allocation of budget to unproven channels, formats, and strategies, with the expectation that some will fail. This isn’t a random or loose mix of activities, or just a spending plan, but rather a different way of thinking about testing and learning. If you take a portfolio mindset with Exploit and Explore, you would anticipate losses in the Explore column. “If you’re expecting everything you do to have a positive return, that’s not a portfolio,” said Inbal. “In a portfolio, you have a mix that exposes you to different risks and opportunities. You expect only some of it to work out.” ### The Top Three Patterns the Best Advertisers Use Outperformers also have some common behaviors that show up in Taboola’s platform data, per Inbal. These are a few operational patterns seen among top performers in advertising: - Determining a simplified master metric: Choosing a single metric lets teams cycle through testing quickly, without getting lost in conflicting key performance indicators (KPIs), and keeps teams unified. - Setting a firm testing target: Pick a firm target each quarter for testing new channels, platforms, or strategies, and set accountability for hitting it. - Incorporating a bias toward momentum: Teams put weight behind early signals quickly, rather than waiting for statistical confidence. Choosing to take action, prioritizing boldness in a structured way, and using data plus instincts can all help performance teams find new and different success. “Make your move. AI will give you a lot of clarity that you never had before, but it won’t make the move,” Inbal said. “AI does not have style. It does not have a sense of belonging or accountability. It does not have a team that it trusts. So, you can make that move because you saw some promising early data, and because there’s someone on your team who is passionate and you believe in their instincts. These are valid reasons to dare and to explore.” ## How Realize Helps You Escape the Alpha Decay Trap Realize and Realize+ each play a role in the Exploit vs. Explore framework. Realize+ serves as the autopilot for the Exploit function. It can handle proven, high-confidence campaigns autonomously to protect efficiency and free the team’s time and budget. Realize serves as the copilot for the Explore function, where alpha is still available. It lowers the cost per creative test, making it possible to run more experiments without proportionally increasing production overhead. Together, they create the practical infrastructure that makes a portfolio strategy possible. “If it’s very tried-and-tested for you, it’s probably tried-and-tested for others,” said Inbal. “So, the trick is to use Exploit to give you the confidence that you can afford to explore. AI lets you do a lot more exploration.” ## Key Takeaways If you’re doing everything right with performance marketing optimization, and still see any advantage slipping away, you’re not alone. You can trust your instincts — and the data to back them up. At this stage in an alpha decay construct, seize the opportunity to do something different. Most teams will respond by optimizing harder within the same system, but marketers who build a parallel exploration system will be much more rare, and more valuable. Now is the time to choose to be bold. ## Frequently Asked Questions (FAQs) ### If AI is causing alpha decay, why should I invest more in AI tools like Realize? There’s a difference between using an AI as a commodity, where everyone does the same thing and loses any advantage they had, and using AI strategically to open up capacity for exploration and gain new advantages. AI doesn’t universally cause alpha decay. It erodes alpha when it’s used for exploitation, but if AI is used to automate the baseline and lower the cost of creative testing, it can become the engine of exploration, instead of dragging down performance. Advanced AI tools can help increase speed and identify opportunities to capture returns before your competitors do. ### How do I know if my marketing is suffering from alpha decay right now? When marketing is suffering from alpha decay, your competitive edge — like messaging, channel, or targeting — has become commoditized or stale. To tell if your marketing is suffering from alpha decay, there are three observable signals to track. First, you can see CPA variance compressing quarter over quarter. Next, creative testing velocity has slowed while time required for analysis has increased. And third, the team is spending more time optimizing existing campaigns than launching new ones. As a starting diagnostic, benchmark your performance distribution internally over 12 to 24 months. ### What does operationalized courage actually look like in practice? Within the context of performance marketing, practicing operational courage falls into three areas. In master metric simplification, the team picks a unified metric to test quickly, such as overall equipment effectiveness in manufacturing. Testing targets should be chosen per quarter to test new channels, platforms, or strategies with accountability, such as launching a product in one city and evaluating before a national rollout. Third, high performers have a momentum bias, putting weight behind earlier signals quickly rather than waiting for statistical confidence, such as shifting budget to a promising publisher site as soon as the numbers start rising there. The abstract idea of being bold can transform into specific process changes that you can implement and evaluate in the next quarter. --- ### Agentic AI vs. Traditional Automation: The New Era of Performance Campaigns URL: https://www.taboola.com/marketing-hub/agentic-ai-vs-automation/ Last Modified: 2026-05-28 11:49:52 We hear a lot about AI automation these days, but automation has been used in marketing for a long time. Traditional automation in marketing campaigns already performs tasks like triggering scheduled emails or managing rigid workflows, but it isn’t able to adapt on the fly as market conditions change. As digital marketing moves ever faster and AI technology matures, agentic AI can go beyond traditional automation to achieve marketing goals. With its predictive capabilities, agentic AI serves as an autonomous digital teammate for marketers to make real-time decisions, optimize budgets, and orchestrate cross-channel strategies. This guide will explain the main differences between agentic AI vs. traditional automation, and how both marketing platforms and human marketers’ daily work can benefit from agentic AI capabilities. ## Agentic AI vs. Traditional Automation: What Is the Core Difference? Traditional rule-based automation uses if/then logic, relying on deterministic, predefined rules to perform repetitive tasks without human intervention. Rule-based automation, also known as robotic process automation (RPA), might include a trigger to send an out-of-office email, a macro run in a spreadsheet, or a welding robot used on an assembly line. Agentic AI, on the other hand, is goal-oriented and context-aware, so it can adapt its strategy without a hard-coded script. Agentic AI refers to the coordination of multiple agents that can plan and execute complex workflows. These systems can learn over time and interpret context to make autonomous decisions in pursuit of a specific goal. Humans set the guardrails for AI agents to operate, but they’re more flexible than traditional automation systems or processes. Goal-oriented AI agents can manage tasks like parsing customer emails, analyzing chat sentiment, or routing help desk support tickets. Research firm Gartner anticipates that by 2029, AI agents will resolve 80% of common customer service issues without any human intervention. That’s projected to save 30% on operational costs. ## Platform Impact: From Static Workflows to Dynamic Orchestration Busy performance marketing teams probably have plenty of ideas already on how AI, and particularly agentic AI, can help save time and meet big goals faster. That’s why marketing and advertising technology platforms have already begun to incorporate agentic AI to support marketers’ goals and help develop dynamic marketing workflows. While marketing platforms have typically relied on static automation, with triggers set up for straightforward tasks, modern agentic platforms take automation much further, as they can continuously analyze live data and adjust ad variations without manual prompting. Clearly, agentic AI automation in marketing offers a lot of promise — here are the top ways agentic AI platforms can help teams meet their goals. ### Proactive Budget and Bid Optimization AI marketing agents can detect even subtle shifts in channel performance, find the underperforming segments, and reallocate budget to the better-performing channels. Performance marketers and advertisers can apply agentic AI to marketing tactics like reallocating spend or adjusting bids to maximize return on ad spend (ROAS). Agentic AI can perform real-time budget reallocation — a task that even the most dedicated human team can’t do that quickly. Agents also learn continuously (from competitor activity or A/B testing) to adapt and make changes without any manual intervention needed. ### Cross-Channel Campaign Execution Disparate channels, workflows, and platforms can cause headaches for performance advertisers trying to orchestrate all their campaigns. Agentic AI serves as an autonomous campaign manager, coordinating actions and complex workflows across search, social, and email platforms simultaneously. This performance campaign optimization ultimately benefits the user, who gets a cohesive journey, as well as marketers, who get improved accuracy and added efficiency without more hours of manual work. ## The Human Impact: From Task Executor to Strategic Director Agentic AI can bring together many disparate tactics and cut down on the thousands of details that a marketing team has to handle. With agentic AI now embedded in advertising platforms, the control layer is shifting. Human marketers can reduce the time they spend monitoring dashboards and looking for the source of underperforming campaigns. Instead, as agentic AI automates optimization tasks, marketers can focus on high-level strategy and guardrail management. ### Eliminating Dashboard Fatigue When agentic AI is applied to marketing and embedded in marketing platform technology, it’s designed to help human teams work better and more proactively. AI agents help to offload dashboard checking and other tasks aligned with marketing goals, as well as proactively notifying advertising and marketing teams about any anomalies that arise and suggesting corrective actions. This always-on, automated alerting means that marketers can stop constantly monitoring performance data. ### Focusing on Strategy and Guardrails With AI agents handling strategy implementation, workflow orchestration, and campaign optimization, the human role in marketing changes. With daily manual work offloaded, human marketers can do the important work that can easily get lost on long to-do lists. This includes setting the high-level key performance indicators (KPIs) and metrics for campaigns; providing creative direction; and setting, managing, and maintaining brand boundaries for the AI workforce. ## How Agentic AI Redefines Campaign ROI Measuring agentic AI ROI compared to traditional automation requires a shift in thinking about value to marketing teams. Traditional automation is best measured in hours saved on repetitive tasks, such as automatic email replies that a human didn’t have to write and send. Agentic AI’s value is best measured in improved decision velocity and adaptation to market shifts. The sheer scale that agentic AI makes possible can change the campaign ROI calculation. When AI agents adjust campaign tactics and reallocate budgets in real time, marketing teams can see huge returns. ## The Hybrid Approach: Building the Ultimate Revenue Engine Both traditional automation and agentic AI have a role to play in modern marketing automation platforms. Traditional automation, or RPA, can still work well for scheduled reports or lead routing from actions like a form fill. Agentic AI likely has countless use cases that marketers are still exploring — any dynamic, multi-variable performance environment can likely benefit from its use. When tasks involve judgment calls, like resolving customer support tickets, or multi-step workflows requiring coordination between systems, agentic AI can probably help. ## Key Takeaways Traditional automation has been a part of marketing and advertising roles for decades, saving time with rule-based execution. As agentic AI becomes more sophisticated, its outcome-driven autonomy can help performance marketers to become more goal-oriented and focused on strategic oversight. For ideal agility and ROI, a hybrid approach helps marketers direct campaigns strategically and meet goals faster with AI support. ## Frequently Asked Questions (FAQs) ### Is agentic AI the same as traditional marketing automation? Agentic AI and traditional marketing are both technically types of automation, but they perform very differently. Traditional marketing automation uses if/then rules to execute predefined tasks based on certain rules or conditions. Agentic AI, in contrast, is goal-oriented and aware of its context. So, it can plan, adapt, and make decisions in real time to meet a specific KPI or other pre-set goal without human intervention. ### How does agentic AI impact a marketer’s daily role? Agentic AI can quickly take lots of manual tasks off a marketer’s plate. AI agents can handle common tasks like campaign optimization or data analysis. That means human marketers can move from working on tactical execution to strategic governance and goal-setting. Teams can stop monitoring dashboards and making manual bid adjustments, and instead work on defining campaign goals, setting budget guardrails, and providing creative direction. ### Will agentic AI completely replace rule-based automation? No, the two logically coexist in marketing technology platforms. Traditional automation makes sense for predictable, stable tasks, and most enterprises use it in some form with strong results. Agentic AI usage continues to develop in dynamic environments where variables change frequently, such as in live ad optimization. This hybrid tech stack offers lots of support for busy performance marketers. ### How do you measure the ROI of agentic AI in marketing? Agentic AI ROI is best measured by the quality of decisions made, the speed of adaptation to market shifts, and tangible improvements in campaign performance. The net financial gain from these improvements can be compared to the total AI deployment costs to measure the total ROI. Consider the financial gain from using agentic AI, like better ROAS or increased conversion rates, or other benefits specific to your business. --- ### How Performance Budgets Are About to Shift to the Open Web URL: https://www.taboola.com/marketing-hub/performance-budgets-shift-on-open-web/ Last Modified: 2026-06-10 10:45:04 If you look at the current data, you’ll see that it both describes where budgets sit today, and predicts where they’re going. That data-informed roadmap has a few new features, too, including agentic AI. What was once a nice-to-have experiment is becoming the primary engine driving the reallocation of performance spend on open web advertising. If marketers want to position their companies at the head of the pack, they need a copy of that map as the shift happens. ## The Gap Between Now and Next: Why the Open Web Is Underinvested Relative to Its Potential As a recent survey conducted by Taboola shows, there’s a pretty big disconnect between current investment and future intent. Currently, the open web — the expanse of premium publishers, news sites, and niche content hubs that comprise the internet — is a second thought for many organizations. Per the survey, only 4% invest significantly (think: 25% of their budget) in the open web, which currently accounts for about 13% of performance spread. Despite this, 82% of those same organizations describe AI-powered, goal-based buying on the open web as a meaningful growth opportunity. The gap between those numbers (13% actual allocation versus 82% belief in that potential) forms the premise of the future. The open web is the world’s largest advertising environment by inventory, reaching audiences at a scale dwarfing individual walled gardens. Yet, it commands only a fraction of objectively smaller platforms. So, why the 13% average allocation? It’s not because the audience isn’t there — you can actually credit (or blame) an infrastructure lag. Search and social won the budget wars because they built the best machines — black-box, goal-based tools that made spending money easy and predictable. The open web’s always been there, but the tools for getting it to function well as a managed performance channel have been fragmented and manual. Marketers know the open web has value, but operationalizing the open web, with its multiple interfaces and supply paths, presents a major headache for the programmatic advertising future. ### The Channel the Budgets Haven’t Caught Up With (Yet) The open web offers something search and social can’t: genuine incrementality. When you buy there, you reach someone consuming high-intent, vertical content. Interestingly, when you ask those same marketing teams about the future, the numbers flip. Nearly 100% of organizations say they’d shift budgets to the open web if an agentic AI open web solution existed to manage it. That’s an average expected allocation of 24%, with nearly 40% of marketers ready to shift a quarter of their total spend. Talk about an opportunity — but also a missing product. ## The Demand Signal: What 81% of Marketers Are Telling Us About Readiness In performance marketing, where leaders scrutinize every dollar and ROI remains the number one metric, achieving 81% agreement on a hypothetical shift is rare. When you talk about the open web, though, over 80% of marketers say they’d increase their investment if it offered the same automated, AI-powered campaign solutions they currently use in search and social. That’s a seriously strong signal. It’s nothing new, either. When agentic solutions became available and proved their value, marketing dedicated more budget to them. We saw this dynamic play out with tools like PMax and Advantage+. Once automation made the channel accessible and results defensible, budgets followed the path of least resistance. Optimizing your performance budget allocation for the open web is the next destination. ### Strong Agreement vs. Somewhat Agree (and Why Intensity Matters) If you examine that 81% more closely, you’ll see that nearly half (49%) “strongly agree.” These organizations have already decided in principle: they don’t need convincing that the open web works, they’re just waiting for the infrastructure to be finished. Those 32% who “somewhat agree” represent the second wave — directionally aligned, but more cautious. They most likely want to see the first movers succeed before they commit and jump. This scenario creates a pretty familiar wave dynamic: - The early adopters move. - The results validate the model. - The rest of the market follows in a surge. ## Who Leads the Shift: Why the Biggest Spenders and Most Senior Leaders Are Ready The driving force toward the open web? The biggest players in the room. The data shows a sharp increase in readiness as you climb the ladder of spend and seniority. Those who agree or strongly agree with shifting spend include: - 35% of senior managers. - 46% of directors. - 67% of VPs. If you break it down by spend, those who strongly agree include: - 3% of those spending $300K-$499K. - 21% of those spending $500K-$999K. - 67% of those spending $1M-$4.9M. - 74% of those spending over $5M. This disparity exists because the biggest spenders hit the ceiling first. If you’re spending $10M monthly on social, you’ve likely found the point of diminishing returns already. With plateaued growth and CPAs creeping up, your CFO is asking why that extra million didn’t produce the same results. Those leaders are motivated to find a third leg of their performance stool and are the ones most likely to move the budget when they find it. ### Why VPs Are the Catalyst and the Most Aligned VPs often own the high-level strategy and hold the purse strings. They don’t need a committee to authorize a channel shift, but they do need a solution that works. When two thirds (67%) express this level of conviction, the conversation shifts from, “Should we commit?” to, “How fast can we execute?” They’re looking for ways to de-risk their portfolios and see the open web as a logical next step for scaling outside the currently dominant platforms. The TL;DR: Those with the most authority to act are also those most ready to act. ## The Three Barriers: Vendor Complexity, Measurement, and Brand Safety If the intent is so high, then, what’s holding them back? The survey identifies a few specific operational hurdles: - Too many vendors/complexity of managing multiple partners (74%). - Lack of unified attribution and measurement (71%). - Brand safety concerns (54%). - Insufficient resources to manage additional channels (42%). Here’s the thing: No one in the industry doubts the efficacy of the open web, but the operational requirements of managing it at scale feel insurmountable. Fortunately, it’s a solvable problem. Let’s look at the top three primary barriers identified by survey respondents in a bit more detail. ### Vendor Complexity: The Fragmentation Problem Nearly three quarters (74%) of marketers cite the sheer number of vendors as a primary barrier. To understand why it’s such a deterrent, look at the performance team’s daily reality. On one side, you have walled gardens with a single interface where you set a goal and a budget. On the other side, you have the open web, which is a maze of thousands of publishers, dozens of intermediaries, varying creative specs, and separate billing cycles. Managing a campaign across even ten different publisher relationships creates a time-crushing amount of operational overhead. Each has its own trafficking system and reporting format. For lean teams that have built their workflows around automated tools like PMax or Advantage+, managing a fragmented open web buy is nightmare fuel — and a huge resource drain. Bring on the consolidation. Performance marketers are asking for a single interface that uses agentic AI open web technology to handle the heavy lifting across all those fragmented sources. Automating the optimization and unifying the workflow banishes the operational nightmare and leaves the performance. ### Unified Attribution: The Measurement Confidence Gap Measurement is the technical barrier sitting under everything else. 71% of survey respondents are stuck here, and it’s a rational hesitation. Performance marketing, at its core, is a numbers game. If you can’t compare your open web CPA to your search CPA with 1:1 accuracy, how do you justify a $2M shift in spend? This apples-to-apples gap (which is more like comparing apples to oranges) prevents confident allocation. The industry has plenty of data; what it really needs is a way to close the loop on attribution within a decentralized environment. Marketers want the same level of certainty in open web advertising that they’ve grown accustomed to in closed ecosystems. It’s worth remembering that a lack of certainty (or data) isn’t unique to the open web. Every major channel faced this same measurement lag in its infancy. Before the right tools were built, search and social were experimental buckets. The budget moved at scale once measurement caught up to the opportunity. The open web is at that same inflection point now. ### Brand Safety: The Governance Requirement Over half (54%) cite brand safety concerns, and that’s a legitimate worry. In walled gardens, you trust the platform to police the content (rightly or wrongly). The open web inventory, however, is decentralized, so that safety net isn’t a given. The main thing marketers considering the open web want to know is whether they can be certain that their ads won’t appear next to fake news or low-quality clickbait, without manually whitelisting every single URL. For most, the answer is no — at least, not yet. This lack of continuous, hands-off oversight is a dealbreaker for brands that can’t afford a PR crisis in exchange for incremental growth. Automated tools and AI-driven sentiment analysis are a solution, but these tools haven’t become standard issue. It’s a solved problem in a technical sense, but it’s not a solved problem in a confidence sense. Until the automated governance is as robust as the buying tools, organizations will likely keep their largest budgets behind the walls where oversight feels baked in. ## The Budget Roadmap: From 4% to 24% and Beyond Let’s calculate the math of reallocation. Currently, 4% of organizations dedicate 25%+ of their marketing spend to the open web, but when agentic AI catches up to intent, the landscape will shift almost instantly. A full 99% of marketers say they’d move budget to the open web, increasing the average allocation to 24%. If you put it into perspective, a 24% share would put the open web on equal footing with paid search (22%) and paid social (21%). In other words, it becomes the third leg of our stool and a pillar of the performance mix, not a side experiment. Moving from a 13% average to 24% nearly doubles the channel’s share of wallet. The internal breakdown is even more telling: - 50% of respondents expect to allocate 11% to 25% of their budget. - 37% would go further, allocating 26% to 50% of their budget. - A tiny fraction (about 2%) would anticipate allocating over 50% of their budget in this way. There’s a clear asymmetry in who’s likely to jump first. The intent to increase open web investment is much stronger among those spending $1M or more per month. When these organizations move, they’ll change percentage points while moving huge absolute dollar values. This level of spend has the potential to reshape the programmatic advertising future. ### The 39% Who Will Redefine the Market The most significant number in this dataset may well be the 39% of marketers saying they’d allocate 26% or more of their budget to the open web. This group is testing the channel while making it a central component of their growth strategy. When nearly 40% of the market moves from 4% to 26%+ spend, the entire advertising ecosystem will evolve. ## The Roadmap Is Drawn: Timing Is the Only Variable We often talk about the future of marketing as if it’s a mystery we’re trying to solve, but the data suggests it’s not speculative — it’s decided. Budgets are over-concentrated, current channels are saturated, and the demand for a third option is overwhelming, particularly among those who control the most money. Agentic AI is already working to solve the engineering barriers of complexity, measurement, and safety. The performance marketing roadmap is clear. What separates the leaders from the laggards now is timing. Organizations that have already begun developing their open web muscles, testing measurement frameworks, understanding publisher value, and getting comfortable with AI-driven buying, are giving themselves a huge head start. By the time infrastructure is fully standardized, the early movers will have already optimized their workflows and claimed the best-performing inventory. The shift is coming, and you have options. You can take a leap of faith now, joining those who are already moving to incorporate the open web as a bigger piece of their marketing strategy. Or, you can wait to see how it goes and risk the channel becoming as crowded and expensive as all the others. You hold the roadmap; how fast are you willing to drive? --- ### Working Smarter, Not Harder: How Ad Managers Can Reclaim Strategic Focus URL: https://www.taboola.com/marketing-hub/human-limitation-ad-management/ Last Modified: 2026-06-30 07:27:51 As seasoned performance marketers know, the day-to-day is a relentless cycle of pressure, fluctuating metrics, and the expectation that the numbers will keep climbing regardless. When accounts plateau despite more optimizations, most media buyers assume they’re doing something wrong, but the fact is that there’s a ceiling on what any human brain can process before seeing a reduction in quality. This guide highlights how ad managers can work more efficiently, exploring how shifting your strategy once you hit your human cognitive limits can restore both your return on ad spend (ROAS) and your team’s strategic focus. ## The Cognitive Ceiling: The True Human Limitation in Ad Management Modern ad platforms generate a large volume of signals in real time. Bids, audience behavior, quality scores, creative performance, and budget dynamics are all shifting. The human limitation in ad management is not a question of skill, but simply the structural reality of how the brain processes information. Cognitive load theory establishes that working memory has a finite capacity, and when the volume of data exceeds it, processing declines. Research published in Frontiers in Cognition confirms that under high cognitive demand, individuals shift from careful, deliberate decision-making toward faster, simplified choices. A separate study from The Journal of Neuroscience found that as cognitive output accumulates, people become less willing to expend effort on higher-reward tasks. For campaign performance, that shift has a real, direct cost. Hitting a plateau is often a sign that the operating model has outgrown what human bandwidth can reliably support. ## How Manager Fatigue Directly Caps Your Campaign Performance The link between the cognitive load on managers and declining account results follows directly from how tired decision-makers behave. Research reviewed by The Decision Lab shows consistently that as mental resources deplete, people favor familiar and low-effort options to avoid complex decisions. In ad management, decision fatigue can result in a preference for safe, incremental changes over bolder creative testing, more reactive responses to performance dips, and a growing reluctance to restructure campaigns that are underperforming. This dynamic doesn’t erode ROAS overnight, but might cause a slow plateau as the team narrows what it’s willing to try. ## The Illusion of Control: Why Micro-Managing the Algorithm Fails When it feels like ad campaign performance is capped, many ad managers start implementing more manual control. The desire to do this is understandable, but machine learning platforms require volume, consistency, and stability to optimize. Frequent manual interventions disrupt the signal patterns they rely on to improve. Compulsive reactive, short-term changes based on short-term variance can fully reset the learning phases. ## The Human-in-the-Loop Solution: Balancing Automation and Expertise The solution here isn’t to hand everything over to platform automation and walk away. Native tools can lack brand judgment, competitive context, and long-term business understanding. The real solution is to divide the workload, with artificial intelligence (AI) handling data-intensive execution, and human expertise being reserved for decisions that actually require it. This is the principle behind human-in-the-loop AI. AI-powered systems reach their potential when humans remain in a strategic and supervisory role, directing the system rather than being replaced by it. In reality, that means AI manages bids, pacing, and creative rotation at the signal level, while humans focus on strategy, creative direction, and interpreting what performance means. This is the model behind Realize+ (currently in Beta), the agentic engine for the open web from Realize. Realize+ continuously decides, executes, and adapts campaign strategies in real time using first-party data signals, bringing the performance power of Google Performance Max (PMax) and Meta Advantage+ to premium publishers, without the platform bias. It handles the execution complexity that limits manager capacity, freeing teams to focus on the strategic work where their expertise actually matters. ## Actionable Strategies to Break the Plateau and Reclaim Strategic Focus ### Simplify Your Account Structures Highly segmented accounts were built for manual optimization. In a machine learning environment, though, they fragment data, slow signal accumulation, and increase what a manager has to monitor. Consolidating campaigns gives algorithms access to broader data pools and reduces overall workload for managers. ### Reframe the Client Relationship Much of the pressure on media buyers comes from accepting responsibility for outcomes that are not fully within anyone’s control. Algorithms change, markets shift. Repositioning the client relationship as a strategic partnership navigating an unpredictable environment, rather than a guaranteed-outcome arrangement, is a more accurate description of how digital advertising works. ### Let Agentic Tools Carry the Execution Load Agentic AI tools are designed to take the high-volume, high-frequency execution work off your plate. By autonomously managing budget allocation, creative rotation, and campaign element generation in real time, these platforms can handle the complexity that limits manager capacity, so teams can stay focused on the strategic decisions that actually move the needle. ## Key Takeaways The performance ceiling that plagues many media buyers is often a cognitive ceiling, not a strategic one. The human cost of trying to manually manage more data than any brain can process at scale trickles down into account results as much as it impacts individual efficiency. The most effective way to overcome this is to change the operating model, letting AI handle the execution that creates overload, and redirecting human expertise toward strategy, creative, and client relationships. ## Frequently Asked Questions (FAQs) ### What is the human limitation in ad management? This is the finite capacity of human memory and cognitive processing when faced with the volume of real-time signals that modern ad platforms generate. Attempting to monitor and respond to all of it manually creates pressure and degrades the quality of decision-making. ### Can AI completely replace human ad managers to solve this? Simply put, no. AI excels at processing data and executing optimizations at scale, but it often lacks business context, brand understanding, and competitive intuition. The effective model is partnership, where AI manages the execution that creates cognitive overload, and human expertise directs strategy, creative, and client relationships to add value. ### How does manager fatigue affect ad performance? As mental resources deplete under sustained high-volume decision-making, choices shift toward familiar, lower-effort options. In an ad account, that produces stale creative, defensive optimization strategies, and reluctance to make the structural changes that would move performance forward. --- ### Agentic AI in Marketing: 3rd Generation of Marketing Automation URL: https://www.taboola.com/marketing-hub/agentic-ai-in-marketing/ Last Modified: 2026-05-28 11:52:51 Modern digital marketers are trapped. Despite a decade of digital transformation — and the acquisition of dozens of martech platforms — marketing teams still spend 90% of their day drowning in the endless, mind-numbing busywork of manual ad maintenance and data aggregation. We were promised a future of high-level strategy and creative breakthroughs. What do we have? A daily grind of spreadsheet pivoting and bid-adjustment hell. Enter agentic artificial intelligence (AI): the 3rd generation of marketing automation. No, it’s not another chatbot or tool that can suggest a better headline: it’s a paradigm shift, where AI still helps with content but also autonomously executes, adjusts, and optimizes campaigns from start to finish. Stop tweaking that dashboard and embrace AI for the heavy lifting so you can reclaim your role as a strategist. ## What Is Agentic AI in Marketing? You could say that agentic AI in marketing represents the pinnacle of current martech evolution. At its core, an agentic system exists as an autonomous entity. It can reason through a set of instructions and take independent action to achieve a goal. Unlike standard automation, which follows a rigid, linear script, agentic AI writes the script. If your goal is to lower customer acquisition cost (CAC), the agent doesn’t wait for a human to change the budget. Instead, it: - Analyzes the performance. - Identifies the underperforming creative. - Generates a new variation. - Redeploys the capital. Welcome to the transition from copilot to autopilot AI. As author and social media influencer Pascal Bornet points out, “When AI is grounded in core business processes and rich enterprise context, it becomes far more than a chatbot. It becomes a real business capability.” ## The Evolution of MarTech: Gen 1 vs. Gen 3 Marketing Automation To understand where we’re going, we need to analyze where we’ve been. The journey of marketing automation includes a history of rigid rules that have evolved into fluid agency. ### Gen 1: Rule-based Logic The era of If/Then. We start at the dawn of marketing automation, like early Mailchimp and basic HubSpot workflows. If a user downloads a whitepaper, send email A. These static systems require a human to map out every single branch of the decision tree. When a variable that wasn’t programmed into the logic changes, the system breaks or, worse, keeps executing a strategy that no longer works. ### Gen 2: AI Copilots/Generative AI Large language models (LLMs) entered the scene, empowering us to generate 50 ad headlines in a minute, or ask a chatbot to summarize immense amounts of data. While helpful, the Gen 2 tool still needs a human to hold the handle. It solves the blank-page problem, but not the manual task-management problem. You still have to copy-paste the AI’s copy into your content management system (CMS) or ad manager. ### Gen 3: Agentic AI, with Realize+ Welcome to the 3rd generation of marketing automation. In this era, we introduce Realize+ (currently in BETA). The name is intentional: Realize+ means not just understanding or analyzing data, but also realizing revenue. That’s because the “+” represents the autonomous AI layer — Gen 2 might alert you that your cost per click (CPC) is too high, but the agentic systems of this third generation also act on that information, transitioning the focus from proxy metrics like clicks and impressions to bottom-of-the-funnel outcomes. Realize+ closes the loop between insight and execution — just another dedicated team member working 24/7 to hit your revenue targets. ## The Complexity Trap and Human Limitations Marketing has officially outscaled human cognitive bandwidth. Welcome to the complexity trap. Teams manage omnichannel campaigns across Google, Meta, TikTok, Instagram, and LinkedIn, and each requires hyper-personalized creative for dozens of micro-segments. When growth slows, the natural human instinct kicks in to do more: more campaigns, more creative, more channels. But, if the underlying system is disconnected, scaling only amplifies the mess, and right now, human marketing teams have reached a breaking point. Data suggests a combination of productivity issues and financial leaking. According to the 2026 State of Performance Marketing report by DemandScience, human teams, overwhelmed by the sheer volume of manual maintenance, inadvertently allow non-performing impressions to leech away about 29% of budgets. That daily grind is both boring and expensive. We humans just can’t move fast enough to reroute bad spend in real time — but Gen 3 agentic AI can. ## Autonomous Campaign Copilots: Eliminating the Daily Grind There’s a solution to this conundrum, and it lies in autonomous campaign copilots. These agents are designed to live inside your ad platforms and handle the granular implementation that usually erodes a media buyer’s hours. As Manu Mehra, head of APJ, puts it, “Outcome-based pricing (OBP) is a game-changer for building trust and generating AI adoption.” By leveraging OBP models, Realize+ agents do more than adjust bids. A standard agent might try to get you more clicks for $2. A Realize+ agent recognizes that those clicks aren’t converting and shifts the focus toward verified sales leads and confirmed bookings. Agentic AI automates negative keyword management, bid scaling, and budget rebalancing to protect your return on investment (ROI) and empower your team to focus on the big picture. Realize+ is an autonomous growth engine for the open web. Rather than you having to tweak dials or pull levers, the system uses real-time data to: - Decide where your money goes. - Build the necessary assets. - Pivot as needed to find the highest ROI. ## AI-Driven Creative Agents: Overcoming Ad Fatigue at Scale To avoid ad fatigue, algorithms need a constant stream of fresh creative to perform well, but human creative teams can’t produce assets quickly enough to keep pace. The struggle is real. Often, as Chris Boggs, founder of Moira AI, says, “Every competitor in the space is running the same structure. Same hook pattern, same text overlay, same testimonial clip. Your audience scrolls past all of it on autopilot.” AI-driven creative agents solve this challenge by operating as automated creative labs. These agents use closed-loop optimization to bridge the gap between performance and production: - The agent sees an ad’s performance dip. - It analyzes which elements (the hook, color, call to action ) are failing. - It brute-forces the testing phase by generating dozens of new variations. - It deploys them — instantly. This process creates a loop where conversion data is fed back into the perception layer. The AI can then anticipate future trends and increase customer lifetime value (CLV) by serving exactly what the audience wants to see next. ## End-to-End Reporting Agents: Aggregating Fragmented Data Data silos are the enemy of growth. Most marketers spend hours each Monday stitching together reports from five different platforms to analyze what happened last week. This process is, in short, inefficient — and we humans may miss certain signals or nuances hiding in the data. End-to-end reporting agents act as your tech stack’s unified nervous system. They: - Autonomously aggregate fragmented data. - Resolve identities across channels. - Provide a single source of truth. Even better, while they streamline reporting, they also make recommendations. A multi-agent system will notice that Meta is driving cheaper leads than Google today and suggest — or, if given the autonomy, execute — a budget reallocation to maximize that day’s return. ## Technology Deep Dive: How Marketing Agents Think and Act Trusting an agentic system requires understanding its brain. Most agentic workflows are built on a three-tier architecture. ### 1. Perception Layer During this input phase, the agent ingests real-time behavioral data, customer relationship management system (CRM) updates, and unified customer profiles. It senses the market pulse even as it processes info from the spreadsheet. ### 2. Reasoning Layer The fun begins in this layer, which prioritizes decisions based on actual business goals. The agent uses LLMs to analyze context. It asks, Based on the goal of maximizing revenue, and given that Facebook’s cost per mille (CPM) just spiked, what’s the best move? ### 3. Action Layer The final step. The agent uses its API integrations with CRMs like Salesforce, ad platforms like Google Ads, and email systems to execute the task. It logs in, changes the bid, uploads the new creative, or triggers the email. The act is the differentiator. ## Guardrails and Governance: The New Role of the Human Marketer There’s a fear among many marketers that agentic marketing will replace the human role. In reality, agentic marketing will elevate humans, because while it can “manage more complex ad campaigns, optimize spend, and handle lead flows without constant oversight, it cannot replace the human touch,” per Grace Ukonu-Onuoha, sales consultant at Kayla Technology Advisors. The human role is shifting from execution to governance and strategy. Consider this analogy: If you were captain of a steamship, you wouldn’t spend time in the engine room shoveling coal (the manual ad adjustments). Your place is on the bridge and maproom: - Setting the destination (strategy). - Defining the path (brand voice). - Keeping the ship within the safety lines (ethical guardrails). You provide the why and the who. Agentic AI takes care of the how and the when. ## How to Prepare Your Tech Stack for the Agentic Era You cannot build a Gen 3 strategy on Gen 1 data. To prepare for this new era: - Clean your data. Agents are only as good as the information they ingest. Dismantle silos and make your CRM the source of truth. - Enable APIs. Make sure your tools can talk to each other. Agentic AI needs robust, bi-directional API access to act. - Focus on identity. Invest in identity resolution so your agents know that User A on TikTok is the same person as your CRM’s Lead B. ## Key Takeaways Marketing teams that have incorporated agentic AI into their workflows are positioning themselves well in the digital landscape. Digital marketing will continue to favor those who connect the deepest (not those who shout the loudest). The gap between those reliant on manual labor and those who’ve embraced autonomous systems may become an unbridgeable chasm. Adopting third generation marketing automation moves you away from the daily grind and closer to a future where your technology is as invested in your revenue goals as you are. The era of the Realize+ marketer has arrived. It’s time to let the agents take the wheel and drive while you strategize. ## Frequently Asked Questions (FAQs) ### What is the difference between generative AI and agentic AI? Several features define generative AI vs. agentic AI. Generative AI (Gen 2) creates content like text or images based on a human prompt. Agentic AI (Gen 3) is an autonomous system. It can reason through a goal, identify and generate the content needed, and deploy it autonomously across channels — no continuous human prompting needed. ### How does Gen 3 marketing differ from traditional automation? Gen 1 traditional automation relies on rigid, human-programmed if/then rules. Gen 3 agentic automation acts autonomously within defined guardrails. It determines the optimal time, channel, and message for each user based on the real-time data it analyzes, rather than relying on a predefined (and potentially outdated) sequence. Realize+ optimizes for the final outcome, not only to save time, but to provide the closed-loop efficiency of a search or social platform across the entire web. Since the system never sleeps, it can search constantly for high-performing paths and instantly switch gears, stopping strategies that aren’t pulling their weight. ### Can AI marketing agents replace my marketing team? No. AI agents free marketers from the drudgery of tactical, lower-level execution so they can focus on strategic leadership. Humans must still define: - Brand strategy. - Creative vision. - The ethical guardrails guiding the autonomous agent’s actions. ### What is the complexity trap in marketing? The complexity trap happens when the channel volume, audience segments, and necessary personalization exceed a human team’s operational bandwidth. Teams stuck in this trap face a grind of manual ad maintenance and data entry, leaving little time for high-level strategy. ### How does Realize+ use agentic AI to guarantee marketing outcomes? Instead of relying on copilot AI that can only make suggestions, Realize+ uses Gen 3 agentic AI to autonomously navigate the action layer, adjusting bids, budgets, and creatives in real time. Optimizing Realize+ for bottom-of-the-funnel goals like verified purchases eliminates the risk of paying for vanity metrics like non-converting clicks. Most platforms guess because they’re looking out of a foggy window. Because Realize+ has direct code on the pages of thousands of publishers, it sees the data firsthand. It eliminates the middlemen (SSPs and Exchanges) to solve two problems: - It finds your audience with first-party accuracy that generic DSPs can’t match. - It skips the ad tech tax. Most of your budget goes toward buying ads, while platform fees consume less. Welcome to a shorter, cleaner path to your customer. ### What makes the Realize+ reasoning layer different from standard automation? Standard automation follows a static script. The Realize+ reasoning layer uses predictive analytics to analyze complex attribution paths. Instead of asking Did they click? It asks Will this action result in a qualified lead? Based on the answer it receives, it shifts resources autonomously to capitalize on the highest-probability outcome. ### How do outcome-based goals change how AI-driven creative agents work? Rather than testing the images getting the most clicks (CTRs), Realize+ creative agents brute-force dozens of variations at super-human speeds to see which drive the most verified sales. The AI aligns creative generation with bottom-of-the-funnel data to provide ad fatigue solutions and lower your CAC simultaneously. --- ### 10 Best Practices for Effective Mail Domain Targeting URL: https://www.taboola.com/marketing-hub/mail-domain-targeting-best-practices/ Last Modified: 2026-05-24 09:25:53 Mail domain targeting can be an effective, consistent part of modern performance marketing or advertising campaigns. That’s especially true now that we know a lot more about this approach, based on years of experience and the capabilities of modern technology platforms. Email data receipt information can go beyond the inbox and help inform further targeting tactics for broader user reach and conversion opportunities. Our Realize experts have spent years of hands-on time exploring what works — and what doesn’t — when using mail domain targeting for modern digital audiences. I’ve gathered them here, with tips on everything from the best creative to use to incorporating search keyword retargeting correctly. ## 10 Best Practices for Effective Mail Domain Targeting, by Realize Experts ### 1. Master the Three-by-Three Creative Rule When you’re developing a new campaign, make sure never to launch with just a single ad. The standard for success is the 3x3 rule: You should use at least three different images and three distinct headlines per campaign. This variety provides the algorithm with enough fuel to find the winning combination that resonates with your specific audience. “You definitely want to have a good mix,” says Jason Poulos, advertising account manager at Realize. “Publishers will block you, sometimes, if there isn’t enough creative diversity. I think that’s probably what’s happening with many advertisers today.” For email domain targeting in particular, consider the three inboxes per domain rule, designed to scale volume while protecting sender reputation. Create a maximum of three email accounts per domain to avoid spam filters and get more emails delivered. ### 2. Embrace Aspirational Scenery Over Stock Imagery There are a lot of data points to use when you’re choosing imagery for email domain targeting and performance marketing campaigns. Importantly, data shows that corporate-looking stock photos (like a random guy in a business suit walking through a city) often underperform. Instead, lean into images like these: - Dreamlike landscapes and imagery. - Uplifting, motivational scenery that makes the user feel good about the partnership. - Active shots: For home services, images of professionals actually working on a roof outperformed AI-generated or static poses. “Scenery images were the strongest performers,” confirms associate advertising account manager Isabel Saltzman. “We recommend leaning into aspirational landscapes, metaphorical imagery, and phasing out some of these stock images, like a business person.” ### 3. Lean Into Mail Placements for High Intent One of the most effective tactics for advertisers currently is mail domain and inbox targeting. Users checking their mail are often in a tactical, to-do list mindset. “People in their inbox are in that utility-driven headspace: money management, financial planning, and organizational mindset,” says client success director Sam Rothberg. “They’re already in that flow. If it’s a trusted brand, a name that they know, the user intent there is strong.” Target users who are receiving emails from specific competitors, or place your ads directly at the top of the inbox (think Yahoo, Outlook, AOL). These ads should be shorter and use higher-contrast imagery, since they’ll appear smaller in an inbox environment. ### 4. Let the Algorithm Breathe (The 14-Day Rule) Modern performance advertising relies on machine learning, so it’s imperative to let new campaigns run for 10-14 days without major manual adjustments. You aren’t doing high-volume blasting with mail domain targeting, but rather using authenticated domains to get relevant content to users without spamming them. Keep that in mind as you’re waiting to see results. ### 5. Utilize Search Keyword Retargeting Bridge the gap between search and native advertising by utilizing search keyword targeting. “You could reach high-intent audiences by search signals,” says Saltzman. “Those will be high-intent users that would be searching for those keywords and really converting.” To do this, upload a list of high-intent keywords (branded or competitor). Then, you can retarget users who have recently searched for those terms across premium publisher sites. ### 6. Prioritize Mobile for Lead Gen Across almost every lead generation vertical — from insurance to home services — mobile continues to take the largest piece of the cake when users are browsing, shopping, or researching. Realize experts recommend weighting your budget 60/40 in favor of mobile during the initial testing phase to gather signals faster. ### 7. Test Motion Ads and Gen AI Variations Static images are the foundation of an ad campaign, but motion ads (subtle movement in a static image) are driving significant engagement. Use built-in generative AI tools to create variations on an image, such as changing a background or adding a pointing gesture, to see which version captures the most attention. ### 8. Avoid Creative Decay With Frequent Overhauls Even winning ads eventually fatigue. If you notice your conversion rate (CVR) dropping, or your cost per acquisition (CPA) creeping up, it’s likely time for a creative refresh. Try a strategy where you keep your top performers, but rotate in at least one or two new visual concepts every few weeks. “Start looking at the campaigns you’re running that have low conversion rates and introduce some new creative,” adds Poulos. “I would definitely start with the lower conversion rate ones.” ### 9. Implement Smart Bidding (Maximize Conversions) Instead of setting and forgetting your bidding strategy, it’s time to move away from fixed cost per clicks (CPCs). Use automatic bidding through your performance marketing platform to adjust your bid, and secure placements on sites at times that are most likely to result in a lead. Once you hit 100 conversions, you can transition to a target return on ad spend (ROAS) model for better efficiency. ### 10. Use Predictive Audience Capabilities Instead of guessing your demographics, let your data do the work. There’s no longer a need to make educated guesses about who’s seeing your ads or interested in your products: With modern performance marketing platforms that use AI, you can get great data to incorporate into your planning. Once a campaign reaches a statistically significant number of conversions (typically 100+), use a predictive audiences feature to create a behavioral lookalike of your actual converters to find new, untapped users. Within your email domains, you can then ensure you’re crafting the right messages and getting them to the most relevant audiences, to engage and convert users. ## Key Takeaways Mail domain targeting remains an effective part of a broader performance marketing campaign, but modern digital environments have made precision and data-driven targeting more important. The expert tips above can help marketers and advertisers create better tailored mail domain targeting plans, including using aspirational imagery, targeting users while they’re checking email, and carefully timing your campaigns and testing to understand what’s really working and what’s not. ## Frequently Asked Questions (FAQs) ### Why is mail domain targeting considered a conquesting tool? When conquesting is the goal in your advertising campaign, mail domain targeting can play a key role. It’s considered a conquesting tool because brands can use it to identify users getting email from competitor domains, then display relevant ads specifically to them. You can intercept a rival’s customer base this way to capture that market share and drive attention to your brand. Knowing who these prospects are is incredibly valuable information, letting you develop specific offers and messages to those audiences. It adds precision targeting and creates a better conversion opportunity. ### Does mail domain targeting reach users outside of their actual inbox? Mail domain targeting these days is sophisticated and data-driven, so it can reach users outside of their actual email inbox. This works by using data from email receipts to then serve up display, video, or social media ads across the internet where those users are spending time. This can be useful when doing competitive conquesting or to target subscribers of specific brands, and it can serve as a form of retargeting based on a user’s interaction with a specific brand’s emails. ### How do I optimize creative assets specifically for mail domain targeting? When you’re thinking about how to best optimize creative assets for mail domain targeting, it’s important to balance high-quality, personalized, and relevant visuals and copy along with technical guidelines. This ensures that emails get delivered and capture attention. So, make sure creative assets are optimized for mobile-first and use responsive design. In addition, use a 60/40 text-to-image ratio, always use alt text, and avoid spam-triggering elements like all caps or heavy HTML. --- ### Best Meta Advantage+ Alternatives for Performance Advertisers URL: https://www.taboola.com/marketing-hub/meta-advantage-plus-alternatives/ Last Modified: 2026-05-28 12:46:14 Things are changing fast in the performance marketing universe, with a long-time leader — Meta Advantage+ — showing signs that it’s not invincible. Advertisers today face rising customer acquisition costs, lack of transparency, and fluctuating lead quality when using search and social platforms like Meta and Google. When you’re seeing diminishing returns and user fatigue from Meta Advantage+ and similar platforms, it’s time to try something different. Lots of performance marketing teams have turned to the open web, scaling beyond walled gardens and finding new ways to connect with audiences and drive conversions. This guide explores the best Meta Advantage+ alternatives beyond search and social, including open-web, agentic AI-driven solutions like Realize+ and top programmatic demand-side platforms (DSPs) that take advantage of the huge opportunity of the open web. Read on to see how you can take back control of your campaigns, access high-intent audiences, and diversify your marketing mix for sustainable, profitable growth. ## Top 5 Meta Advantage+ Alternatives for Performance Advertisers There are already some AI-driven programmatic ad platforms available to performance advertisers. The top five platforms below all made the list because they help brands capture high-intent users across the open internet, rather than the siloed walled-garden approaches that restrict advertising to social feeds Platform Best for Special features Pricing 1. Realize+ Data-driven marketers, strategic agencies, direct enterprise advertisers. Predictive AI with conversion focus: agentic AI handles real-time ad decisions, automated campaign testing, and execution. Performance-based model; campaigns billed on CPC basis, or CPM for programmatic. 2. The Trade Desk Those with experience in ad bidding, enterprise-level brands, omnichannel marketers. Proprietary AI and identity solutions, user ID beyond cookies, real-time bidding, and budget allocation. Percentage of ad spend, high minimums. 3. StackAdapt Mid/large agencies, verticals with specific targeting needs like finance, government, and healthcare, B2B lead generation. Easy to get started if new to programmatic buying, creative tools for asset repurposing, multi-channel reach. Cost per mille (CPM), no monthly minimums. 4. Criteo E-commerce, D2C retailers, retail media networks, brand manufacturers, and those with high-volume inventory. Large product catalog, real time analysis, and 20 years of shopper data help the tech serve up dynamic ads to cart abandoners. CPC. 5. Outbrain Content-heavy marketers, performance marketers, D2C brands, and education-heavy products. Leading recommendation and discovery platform; many popular, reputable news site placements from a single platform can engage high-intent users. CPC. ### 1. Realize+ The Realize+ advertising platform takes automation further with a new, third generation agentic engine that helps performance advertisers, direct enterprise advertisers, and strategic agencies become outcomes-led. It’s built for scale, so that teams can test the huge range of campaign elements automatically, and brings the performance strengths of tools like Meta Advantage+ to open web advertising strategies. Features:  - Performance AI tools focus on maximizing conversions, while targeting technology places ads globally outside of walled gardens, showing them to intent-driven users. - Provides cookieless options, using first-party data and unique data signals to target users with the Predictive Audiences feature. - The new Realize+ agent can make real-time decisions that align with specific performance goals, bridging the gap between human limitations and unique data signals. Pricing model: Performance-based model; campaigns billed on CPC basis, or CPM for programmatic. Pros:  - Realize+ can decide, execute, and adapt ad strategies in real time. - Focuses on performance across premium and brand-safe environments on the open web. - Vertical ad options to mimic social success. - A real-time reporting dashboard that optimizes for conversions. - Unique placements that pre-install content to bypass mobile browsers. - A Social Importer tool to repurpose high-performing creative. Cons: - Strict ad policies can lead to slow approval processes. - Users have to do more frequent headline and image testing. ### 2. The Trade Desk The Trade Desk offers omnichannel reach with access beyond Meta’s ecosystem, a huge premium inventory, and an AI engine that optimizes real-time bidding and budget allocation based on performance data. It’s best for data-driven marketers with some experience in ad bidding, including those at ad agencies, enterprise-level brands, businesses focused on omnichannel advertising, and companies with a large set of existing data. Features:  - Koa, The Trade Desk’s proprietary AI engine, and its advanced identity solution, UID2, provide frequency capping for advertisers to ensure no wasted budget on the same user or audience. - UID 2 also addresses the need for cookieless advertising to identify users across devices with encrypted email addresses. Pricing model: Managed or self-service fees based on a percentage of ad spend; can require high monthly minimums starting at $20,000 per month. Pros:  - Identity solutions framework for cookieless targeting. - First-party data integration to find lookalike audiences. - Unified dashboard to easily shift budget between channels. - Great transparency on where money was spent. - Cross-device attribution. Cons:  - High spending minimums and platform complexity. - Steep learning curve and some manual oversight needed. - Reporting data volumes can overwhelm smaller teams. ### 3. StackAdapt StackAdapt is a self-serve DSP that’s an accessible option for those newer to programmatic buying, and useful for mid- to large-sized agencies, specific verticals like healthcare and government, and advertisers working in B2B lead generation. Features:  - Includes proprietary Page Context AI that runs campaigns across multiple channels to target users based on consumed content vs. browsing history. - Provides specialized B2B targeting tools, which is a common pain point with Meta+. - Offers creative tools for marketers to quickly adapt assets for different formats without a large design team. Pricing model: CPM-based pricing with no monthly minimums. Pros:  - Easy to implement and spin up new campaigns quickly. - Multi-channel reach across native, display, video, and more. - Easy to adapt assets for reuse. - Cost-effective for smaller businesses, without strict contracts. - Intuitive, modern UI. Cons:  - Requires a lot of high-quality creative assets. - Smaller global reach compared with The Trade Desk and other options. - Some reporting lag for real-time metrics. ### 4. Criteo Criteo is a commerce-focused platform for performance advertisers looking for retargeting beyond Meta, with 2.5 billion active users in its database and access to on-site inventory, and AI capabilities that can predict what product a user will buy next. It’s also useful for retail media networks monetizing their own sites, brand manufacturers, and high-volume inventory businesses such as travel or classifieds. Features: - Its dynamic ad technology serves e-commerce and D2C retailers looking to improve cart abandonment metrics. - Provides first-party media network data that knows what users bought across the web. Pricing model: CPC pricing; new GO platform designed to be more affordable for smaller teams. Pros:  - Dynamic retargeting leader. - Unique retail media access for brands that want access to advertising space on major global retailers. - Easy to use, with automation handling the bidding and creative generation. - Access to real-time shopping data from billions of users and a large product catalog to help bring shoppers back to complete their purchase. - Predictive analytics based on 20 years of shopping behavior data. Cons: - Ads are highly templated, with less creativity possible. - Cost-per-click may be higher because of high-intent shopper focus. - Attribution overlap potential requires careful testing. ### 5. Outbrain Outbrain’s content-first approach reaches users through storytelling, and works well for industries that need education before making a sale and those looking to lower acquisition costs away from social channels. Its “Recommended for You” placements appear on major news sites to bring high-intent users onto blogs, whitepapers, or other editorial content. The platform is best for content-heavy marketers, D2C brands, performance advertisers prioritizing measurable actions, and media and publisher sites. Features:  - Helps combat ad fatigue and banner blindness, since its ads look like organic content that blend well with user experiences. - Offers one point of access across thousands of premium publishers to save time negotiating individual deals. Pricing model: CPC-based; low daily minimum starting at $20 per day for self-serve option. Pros:  - Leading recommendation and native discovery platform. - Bid strategy automation that works in real time. - Lower cost-per-click opportunities. - First-party data focus. - Dynamic ad formats. - Offers high scale on the open web, especially reputable, brand-safe sites. Cons: - Traffic and click quality can vary. - Images and headlines require frequent rotation. - Rigorous editorial standards and guidelines may take more time. ## The Problem with Walled Gardens (Why Move Away from Meta Advantage+?) Meta Advantage+ tools have provided benefits to performance advertisers: they can simplify ad setup, optimize performance using AI, and save time through automation. Results can include lower cost per acquisition (CPA), simpler targeting and budgeting through help from the algorithm, and time saved on setting up and executing campaigns on Meta social channels, especially when considering Meta Advantage+ vs. manual options. That said, performance marketing teams quickly see steep drawbacks, which is when they start to look for Meta Advantage+ alternatives. The Meta Advantage+ tools rely on automation to find high-value users and ignore the specific targeting set by advertisers. This leads to spending on the wrong audiences, which advertisers can’t control. Ultimately, these tools aren’t transparent or able to combat the impact of ad fatigue and rising CPA. Because Meta Advantage+ uses broad targeting, advertisers will quickly see increased costs and face a harder time when trying to reduce CPA on the open web. ## What to Look for in an Open-Web Advertising Platform So, what should marketers prioritize when choosing an alternative platform to Meta Advantage+? Keep these criteria in mind when evaluating open web advertising platforms: - Robust AI targeting: Ensure robust contextual targeting so that you get both automation capabilities and targeting based on your brand’s audience demographics and your business goals. - Performance focus: As technology like agentic AI evolves, look for platforms that can bring a performance focus to your advertising strategy, rather than only reach or ease of use. - Omnichannel reach: The platform should include the range of channels and ad formats you want to use across native, video, display, programmatic, social, web, and any other potential options. - Deep reporting transparency: Look for a platform that shows all the details on where your advertising budget was used and how each allocation performed. - First-party data activation: Integrating platform data and taking full advantage of your first-party data is essential for performance advertising success — make sure it’s being put to use within an open-web advertising platform. ## Meta Advantage+ vs. Programmatic Demand-Side Platforms (DSPs): A Quick Comparison The primary differences between Meta Advantage+ and programmatic DSPs are largely around transparency and inventory. Teams looking for AI-powered performance marketing strategies can consider these key differences between social feed environments and the open web, and between algorithmic and granular targeting control. Feature or benefit Meta Advantage+ Programmatic DSPs Why use Maximum automation and ease of use. More control and transparency. Formats and channels High-engagement social feeds (Facebook, Instagram, Reels), the walled gardens. Premium news sites, connected TV (CTV), and mobile apps across the open web. Transparency Difficult to see where ads appeared. Granular reporting shows where ads appeared. Targeting options Finds the audiences based on guardrails you set. Manual control over keywords and publishers. Control required Very little. More hands-on, with some expertise necessary to manage bidding and placements. Creative options Social format options include vertical videos, static images, and carousels. Many formats, including banner ads, native content, display ads, pre-roll video, and CTV commercials. ## How to Transition Ad Spend Safely Without Losing Momentum Moving off of Meta Advantage+ can seem daunting, especially if you’ve been relying on it for a long time. Here are some tips to transition your ad spend off of Meta without losing any of the audience momentum you’ve built. - Use a 15% to 20% test budget allocation when you start the move. - Run A/B tests against your existing, high-performing Meta ads to compare CPA and return on ad spend (ROAS) for each. - Duplicate the winners into a new campaign to safely move budget, so the original keeps working for you while you’re establishing stability in the new channels. - Use the pause function to keep the algorithm warm and retain data, rather than stopping an entire campaign abruptly. ## Key Takeaways Performance marketers relying heavily on Meta Advantage+ should start diversifying away and exploring more transparent ad tech solutions. The walled gardens of search and social have served an important purpose, but ad fatigue and diminishing returns no longer deliver for many advertisers. Brands have to test open-web alternatives to find and engage high-intent audiences, reclaim transparency and control, and reach for measurable growth with advanced AI and first-party data. ## Frequently Asked Questions (FAQs) ### Why are advertisers moving away from Meta Advantage+? Meta Advantage+ is easy to use and familiar to many users, but advertisers are moving away from it because it’s a black box without the transparency or control that many teams need. For example, advertisers can’t exclude specific low-performing placements or get granular data, which often leads to wasted spend and too many unqualified leads as budgets scale. The broad targeting of Meta Advantage+ can result in low performance for niche brands and unpredictable ad costs. ### What is an open web advertising platform? An open web advertising platform is a technology solution that allows its users to buy and serve ads to audiences on the open web — meaning a variety of independent websites, news outlets, streaming services, and apps outside of closed ecosystems like Meta or Google. These platforms offer massive reach and brand-safe environments, using programmatic technology and AI features to deliver ads to specific audiences with higher transparency than social media platforms. ### How does Realize+ compare to Meta Advantage+? Realize+ includes agentic AI that’s designed to bring the performance strengths of Meta Advantage+ to the reach and scale of the open web. It can execute and adapt campaign strategies in real time, going beyond what’s possible for a human advertising team to accomplish. Realize+ incorporates unique data to drive campaign decisions based on performance goals, as well as handling the many decisions included in a campaign: the targeting, the creative elements, the budget available, and the bidding strategy. This is a big contrast to Meta Advantage+, which restricts ads to its own closed properties and often targets too broadly, without any transparency available to users. ### Do I need a massive budget to use these Meta alternatives? Meta Advantage+ alternatives have a wide range of pricing options, so you don’t need a massive budget to use another platform. Multiple self-serve DSPs offer a low or zero minimum spend. Or, depending on your brand and business goals, you might explore CTV platforms or other options where you can test ads with small or moderate budgets. --- ### The Best Platforms for AI-Generated Ad Creatives: Creation, Testing, and Placement URL: https://www.taboola.com/marketing-hub/best-platforms-for-ai-ad-creatives/ Last Modified: 2026-06-18 08:27:48 In today’s performance marketing landscape, algorithms on major platforms cycle at machine speed through ad variations, rewarding the most engaging creatives with lower costs and better distribution. This has created a critical need for solutions that manage the entire creative lifecycle, from initial creation and rigorous testing to automated placement and real-time optimization. Still, the “perfect platform” is in the eye of the beholder, and the right choice depends entirely on your advertiser type, budget, and strategic objective. With that in mind, this guide cuts through the noise to help you build your ideal ad creative stack, analyzing the market's leading players, from integrated, conversion-focused full-stack solutions to specialized tools for testing, dynamic creative optimization (DCO), and fast AI-powered production, to reveal the features, price points (when available), and best-fit use cases for each. ## 10 Top Platforms to Generate, Test, and Place Ad Creatives Platform Best For  Features Industries Notable Clients Price 1. Celtra Enterprise creative operations teams needing cross-channel creative governance and large-scale creative production. Creative build and preview, template management, collaboration, analytics, ad validation. Retail, CPG, telecom, agencies. Adidas, Nike, Spotify, Unilever, NBCUniversal (NBCU). Enterprise (contact sales). 2. Marpipe Advertisers who need exhaustive multivariate creative testing and clear component-level answers. Modular creative combinatorics, automated multivariate testing, insights on creative elements. E-commerce, direct-to-consumer (DTC), gaming, apps. Growth marketers, conversion rate optimization (CRO) teams. Pricing tiers by SKU, from $1,499/month to $3,499/month for up to 75,000 SKUs. 3. Realize Advertisers seeking an integrated creation, testing, and placement stack for conversion outcomes beyond search and social. Gen AI motion ads, social importer, performance AI, direct publisher integrations, multiple display formats including vertical, carousel, and video. E-commerce, travel, finance, auto, telecom, enterprise, lead gen. Philips, PortAventura World, Unilever, Anantara, Ashiana. Performance-based model; campaigns billed on CPC basis, or CPM for programmatic. 4. AdCreative.ai Small to mid-sized teams that want fast AI-generated ad variants and predicted performance. AI creative generation, variant scoring, export to platforms. Small and medium-sized businesses (SMBs), e-commerce, agencies. Fast-growing DTC brands. Three introductory tiers (Starter, $39/month; Professional, $249/month; Ultimate, $599/month) and Enterprise. 5. Pencil Performance teams focused on fast video/motion creative powered by predictive AI. Short-form video generation, performance prediction. E-commerce, apps, DTC. Direct response brands. Three tiers: Core, $14/month; Growth, $55/month; Pro, custom pricing. 6. Hunch Advertisers requiring dynamic creative optimization (DCO) and highly personalized creative at scale. Dynamic creative, feed-driven personalization, variant decisioning. Retail, travel, e-commerce. Catalog-led advertisers. AI Workspace between $25 and $50 per user; Enterprise custom starting around $2,800/month. 7. Canva Pro Small teams needing fast static and light motion creative production. Templates, exports, team collaboration. SMB, agencies, creators. Mass market. $15 per user/month; $120 per user/year. 8. The Trade Desk (DSP) Programmatic buyers scaling access to global inventory and omnichannel targeting. Custom bidding, data integrations, private marketplace (PMP), cross-device. Enterprise, agencies. Fortune 500 brands. Platform fee: 20% of the total media spend and cost-per-mille (CPM). 9. Outbrain Native/content-led performance advertisers seeking premium open-web placement. Native image/title units, editorial-style contextual placements, recommendation widgets. Travel, finance, retail, publisher monetization. Major publishers, DTC, agencies. CPC/CPA (cost per acquisition) model. 10. TripleLift Visual-first native and CTV inventory — advertisers who want richer in-feed and video experiences. Native display, in-feed video, CTV, layout optimization. CPG, auto, entertainment, retail. Global brands and publishers. Programmatic CPM. ### 1. Celtra Celtra is an AI-powered creative automation platform built for brands, agencies, and media owners that need to produce a lot of on-brand ads quickly across formats and channels. It centralizes template-based production for static, HTML, video, and rich media ads, then layers on automation and performance insights so creative teams can scale variants without sacrificing design quality. With integrations for existing tools like Photoshop, Figma, ad servers, and analytics stacks, plus collaborative features for review and approvals, Celtra is designed to remove bottlenecks between creative, marketing, and media teams while keeping every asset on brand. It’s usually a better fit for mid-market and enterprise organizations or agencies than for small teams looking for a simple, low-cost design tool. Price Celtra uses a custom, subscription-style pricing model based on seats, usage, and export volume, with published listings typically showing “pricing available upon request.” It’s generally positioned as a mid-market to enterprise solution and is often described by reviewers and competitors as relatively expensive for small teams, so most buyers will need to speak with sales for an exact quote. Special Features - AI and automation for production: Automates repetitive design work and adapts templates into many sizes and formats. - Template-based ad builder: Builder for static, HTML, video, and rich media ads with reusable modular templates for different channels, markets, and audiences. - Design tool integrations: Direct copy-paste/import from Photoshop and Figma so designers can turn existing layouts into scalable templates. - Creative insights and optimization: AI-powered insights and creative performance analytics highlight winning variants. - Collaboration and workflow management: Centralized workspace for comments, approvals, and updates, plus multi-client account management for agencies and media operators. - Omnichannel and multi-screen support: Supports ads across display, video, native advertising, and other formats. - Enterprise integrations: Connectors into ad servers, DSPs, DCO, and analytics/CRM tools like Google Ad Manager, The Trade Desk, Adobe/Google Analytics, Salesforce, and more. Pros - Strong fit for agencies and large brands that need sophisticated workflows and high-volume creative adaptation across markets and channels. - Speeds up creative production and trafficking so teams can ship more campaigns faster while keeping assets on brand. - Robust collaboration tools and approval flows to reduce back-and-forth and keep creative, marketing, and media aligned. - Reporting and creative analytics help marketers understand which variants work best. Cons - Custom, enterprise-oriented pricing can be expensive for small teams or early-stage advertisers. - Feature depth and workflow complexity mean there’s a learning curve and typically a more involved implementation. - Optimized more for creative adaptation and workflow than for full-stack GenAI creative ideation, which means teams may still need to rely on other tools for heavy concepting or long-form video editing. - Best suited to organizations with defined creative operations. ### 2. Marpipe Marpipe is a multivariate creative testing and catalog-ad platform built for performance marketers who live inside Meta, Google, and other dynamic product ad ecosystems. Marpipe lets you generate large numbers of ad variants from existing assets, launch structured tests to real audiences, and see results broken down at the level of individual elements like headline, background, call to action (CTA), and image, to get a better view of why some creatives win — and others don’t. The tradeoffs are that it’s more specialized (and potentially pricier) than generic design tools, and it rewards teams willing to commit to structured testing rather than one-off experiments. Price Marpipe uses a flat monthly SaaS pricing model rather than taking a percentage of ad spend or revenue, which is attractive for brands that scale spend aggressively. Entry-level plans start in the low hundreds per month, with higher tiers for multivariate testing and large catalogs. Exact pricing requires checking the current self-serve tiers or talking to sales. Special Features - Multivariate creative testing: Automatically builds and launches structured multivariate tests to Meta audiences, measuring the performance of every combination of creative variables, rather than just A/B winners. - Element-level creative intelligence: Breaks down performance by individual elements so teams can build a reusable creative playbook of what consistently works. - Enriched catalogs and design control: Overlays brand design, messaging, pricing, and urgency treatments across product feeds to make dynamic product ads look like real on-brand ads instead of plain catalog tiles. - Product-level video at scale: Automatically turns individual SKUs into short videos for social and dynamic product ad (DPA) placements, improving thumb-stop rates without manual editing for each product. - Generative AI for catalogs: Uses GenAI to generate or enrich product copy (titles, descriptions, overlays) for each SKU, tailored to performance and brand voice. - Built-in stat-sig “Confidence Meter”: Live statistical significance calculator that shows when test results are reliable, reducing guesswork for performance teams. Pros - Purpose-built for multivariate testing and catalog ads. - Element-level reporting turns creative testing into a long-term knowledge asset. - Strong fit for e-commerce and Shopify brands running DPAs at scale. - Flat-fee pricing with no cut of ad spend can be more predictable and economical for brands with large budgets. - Self-serve UX and onboarding content (blog, DPA Academy, examples library) make it easier for in-house teams to adopt without heavy service retainers. Cons - Optimized primarily for catalog and social performance ads (especially DPAs on Meta and similar channels). - Very small advertisers or low-budget campaigns may not fully benefit from multivariate testing overhead. - Compared with lightweight creative tools, Marpipe can feel complex. - Some pricing is above entry-level catalog tools, and some features (like advanced testing or very large catalogs) may be locked behind higher tiers. ### 3. Realize Realize is an integrated performance advertising platform that combines creative generation (including static to motion, GenAI, and more), creative testing (AI bidding, performance simulation), and direct placement on premium publisher sites, allowing advertisers to run conversion-focused campaigns on the open web, outside the walled gardens of search and social. Embedded publisher integrations, proprietary data signals, and specialist performance AI are all included, enabling optimization far beyond simple vanity metrics. Price Performance-based model; campaigns billed on CPC basis, or CPM for programmatic. Special Features - GenAI Motion Ads: Converts static creative to motion formats at scale, better capturing audience attention. - GenAI Ad Maker: Creative suite that incorporates AI image generation, background replacement, image extender, and more. - Social Importer: Allows you to repurpose your best-performing social creative into various forms of high-impact display ads. - Video/Vertical/Carousel: Supports display ads across a range of formats, allowing you greater customization over your narrative. - Performance AI: Maximizes conversions, optimizes for engagement, and incorporates SpendGuard and Performance Simulator. - ABBY: Dedicated AI assistant designed to speed up onboarding, flag creative issues, and recommend optimizations. - Direct publisher integrations: Access to more than 11,000 premium publishers across the open web, with brand-safe inventory and no made-for-advertising (MFA) websites. Pros - GenAI AdMaker is trained on billions of real-world performance signals to automatically produce creative assets that are statistically more likely to drive conversions. - The platform enables high-velocity A/B testing by instantly transforming static images into high-engagement motion ads and varied copy, effectively eliminating creative fatigue and production bottlenecks. - Realize unlocks access to premium, high-intent audiences on top-tier publisher sites, providing a powerful and often more cost-effective alternative to the saturated and rising costs of search and social walled gardens. Cons - Enterprise-first orientation means that smaller teams may face setup/time costs. - As it’s a full-service performance engine, some advertisers will prefer to use specialist point solutions for very specific creative workflows (e.g., complex video editing beyond platform GenAI capabilities). ### 4. AdCreative.ai AdCreative.ai is an AI-driven creative generation and optimization platform focused on quickly producing high-CTR, conversion-oriented ad assets. It can generate display and social banners, short videos, text ads, and product “photoshoots” from simple inputs like URLs or product shots, then score each creative for predicted performance so marketers can choose the most promising variants. AdCreative.ai is best understood as an “AI ad factory” for marketers who need lots of social and display concepts fast, and want a performance-oriented scoring layer to sort them. The tradeoffs are credit limits on lower plans, variable reviews in terms of customer support, and a degree of human curation still being required. Price AdCreative.ai follows a credit-based SaaS model with multiple tiers and a free trial. Entry plans start around $40 per month with limited downloads. Higher “Professional” or “Pro” tiers offer more credits, brands, and users at significantly higher monthly fees. Enterprise plans have custom pricing. Special Features - AI creative generation for ads and social: Generates conversion-focused ad creatives across platforms like Meta, Google, TikTok, and LinkedIn from simple prompts or URLs. - Creative Scoring AI: Assigns a “conversion score” to each generated creative based on models trained on historic ad performance. - AdLLM text generator: A proprietary LLM trained on high-performing ad copy. - Product photoshoots and image generation: Turns basic product photos into polished “photoshoot” creatives and can generate custom, royalty-free images at scale to replace stock. - Brand profiles and custom templates: Lets users define brand colors, logos, and guidelines, then create reusable templates with dynamic fields to generate on-brand variations across sizes and channels. - Ad account integrations and exports: Connects to ad accounts so the AI can learn from historical performance, and supports direct export or easy download for Meta, Google, and other major platforms. - AI Compliance Checker: An AI-based checker that reviews creatives for platform and legal compliance, including promotion terms, misleading claims, and brand usage issues. - SMB, agency, and enterprise workflows: Use-case flows and plans tailored to small businesses, agencies, and enterprises, including dedicated onboarding, data governance, and account management at the higher tiers. Pros - Extremely fast way to imagine and produce many ad variations without a designer. - Multi-platform support (Meta, Google, TikTok, LinkedIn, etc.) and export options make it useful as a central creative utility. - Creative Scoring AI and AdLLM help performance marketers prioritize concepts and copy based on predicted outcomes. - Template and brand systems enable on-brand scaling of creatives across sizes and campaigns. - Flexible plan structure (from low-cost starter to enterprise), plus free trial. Cons - Limited download credits on entry plans are quickly exhausted if you’re running serious tests. - Because it’s a generalized AI creative engine, it doesn’t natively handle advanced multivariate testing or deep, platform-specific workflow automation. ### 5. Pencil Pencil is an end-to-end GenAI ad creation and optimization platform built specifically for marketers, agencies, and enterprise brands. You can use its chat to go from brief to finished ads, generating images, video, and copy before you spend your first dollars. Pencil pulls in data from connected ad accounts, and applies machine learning to forecast outcomes via its Media Performance Score. Pencil is best thought of as a “creative co-pilot” for performance marketers, standing out from simpler AI design tools by covering the full workflow and offering serious enterprise controls. Price Pencil uses a tiered SaaS model based on the number of outputs you create (“generations”) as well as enterprise options. Core plans start around $11 per month for 50 generations. Growth plans are around $44–$55/month (250 generations), billed annually or monthly, and the Pro/Enterprise tiers unlock unlimited generations and advanced governance features. Special Features - Chat-to-Ads workflow: A unified chat interface that guides you from idea to finished ads. - Media Performance Score/predictive AI: Machine-learning models analyze large datasets of past ad performance to predict how new creatives will perform, assigning a score so you can prioritize high-potential ads and cut wasted spend. - AI Agents catalog: Prebuilt and custom “Agents” for specific channels and tasks, plus an Agent Catalog to help you extend your own prompts and rules. - Full-funnel creative coverage: More than 50 channel-specific templates and workflows to create assets for social ads, video, display, emails, and landing pages. - Automated adaptation and localization: Spreadsheet-style bulk generation to adapt and personalize creatives for different markets. - Multi-model GenAI stack: An aggregator for top foundation models for image, video, and text. - Brand controls and enterprise governance: Role-based access, data ringfencing, “no-train” policies, and IP assignment terms to ensure privacy and data safety compliance. - Ad platform integrations: Direct ad platform connections to Meta, Google, YouTube, TikTok, DV360, and LinkedIn. Pros - Purpose-built for performance advertising. - Predictive scoring is a genuine differentiator. - Strong multi-channel support and templates across the full customer journey. - Deep integrations with leading AI and creative ecosystems. Cons - There’s a learning curve. - Free and lowest-tier plans are limited in generations and features. - While it supports a wide range of marketing assets, highly bespoke video production or complex brand storytelling may still require traditional creative teams and tools. ### 6. Hunch Hunch is a creative automation and media-buying platform built for performance marketers. It connects feeds, creative templates, and campaign setups in one workspace to let marketers generate thousands of personalized image and video ads, localize them, and sync them into campaigns automatically. It’s positioned for e-commerce and travel marketplaces, and other verticals that need to turn large catalogs into revenue-focused paid social programs. For performance-driven brands spending meaningful budget on Meta and other social platforms, its catalog automation, dynamic video, and performance-linked creative workflow can improve your efficiency and ROAS. Price Hunch runs on an enterprise-leaning SaaS model with pricing based on campaign volume, media spend, and service scope. Most customers are given a demo and custom quote. Publicly available data place Hunch’s entry point in the $2,000 to $2,500+ per month range, often including a minimum included ad spend (for example, around $50,000/month). Hunch is an enterprise-level investment that requires contacting sales for an accurate number. Special Features - Creative automation for catalog and DPAs: Turns any product or content feed into on-brand ad templates, automatically generating and updating image and video creatives for each SKU. - Dynamic creative and localized messaging: Lets you build dynamic templates with overlays for price, discounts, location, or other feed signals, and deliver highly localized, personalized ads at scale. - Dynamic video ads/Catalog Product Video (CPV): Automatically transforms catalog items into product-level videos tailored for Meta placements like Reels and Stories, using a template-driven video studio. - Media workflow and campaign automation: Builds and manages complex campaign structures on Meta and other paid-social platforms. - Performance data hub: Combines Meta Ads data with Google Analytics into one view, with product-level and template-level breakdowns to show which creatives and items actually drive revenue and return on ad spend (ROAS). - Creative Studio and examples library: A template editor for static and video creatives, plus a gallery of high-performing catalog/DPA examples to shortcut strategy and design. - Meta-first but multi-platform: Deep Meta integration (including the “New Meta experience” UI inside Hunch) with support for TikTok and Google, giving social teams a central hub for performance creatives. Pros - Purpose-built for paid social performance, especially catalog and dynamic product ads. - Strong automation for both creative and media workflows. - Deep product- and template-level reporting. - Well-suited to enterprise e-commerce and agencies managing multi-market, multi-catalog setups. Cons - Pricing is pitched to enterprise clients , rather than small operations. - Because it bundles creative automation and campaign orchestration, some teams end up using only a subset of features they’re paying for. - Focus is heavily on Meta, TikTok, and Google paid search, not as much on search, programmatic, or brand-heavy video. - Complexity and implementation overhead mean it shines most for teams with established paid social operations. ### 7. Canva Pro Canva Pro is an all-in-one visual communication platform that layers premium content, brand controls, and AI-powered tools on top of Canva’s familiar drag-and-drop editor. Aimed at creators, small businesses, and in-house marketing teams, it unlocks millions of stock assets, advanced editing features, brand management, and workflow tools. Pro also gives full access to Canva’s Magic Studio AI, turning quick prompts and uploads into polished, on-brand designs across channels. The tradeoffs are that it’s still not a pro motion or layout tool for the most demanding design teams, and rising prices mean you’ll want to actively leverage its brand kits, collaboration, and Magic Studio features to fully justify the spend. Price Canva Pro is sold as a subscription for individuals and small teams. Current rates price Canva Pro at about $12.99 to $15 per month, or $119 to $120 per year per user, depending on region and billing. Canva for Teams starts at around $14.99/month for up to five users and roughly $100 per user per year at scale. Education and nonprofit programs can access many Pro/Teams capabilities for free if they qualify. For large organizations, Canva offers Enterprise with custom pricing and expanded governance. Special Features - Magic Studio AI toolkit: Includes Magic Design for generating full layouts from prompts or uploads, Magic Write for AI copy, and Magic Media (Text-to-Image/AI video) for creating custom visuals and clips inside the editor. - Premium content library: Access to more than 100 million premium photos, videos, audio tracks, icons, and templates. - Brand kit and Brand Controls: Store logos, brand colors, fonts, and design guidelines, as well as the ability to keep the team on point with Brand Controls. - Magic Resize and advanced editing: One-click resizing of designs for multiple platforms, background remover, Magic Expand/Erase/Grab and other editing tools. - 1TB cloud storage and asset library: Centralized storage for brand assets, templates, and design files with version history and shared folders for teams. - Apps and integrations marketplace: Integrations with tools like Google Drive, Dropbox, social platforms, and productivity apps. - Collaboration and publishing tools: Real-time collaboration, commenting, task assignment, approval flows, and direct publishing or scheduling. Pros - Extremely low learning curve and broad template library make it easy for non-designers to produce professional-looking creatives quickly. - Magic Studio AI saves time on ideation, layout, and asset creation. - Strong value for money. - Robust collaboration, storage, and brand-kit features. Cons - While powerful for general marketing design, Canva Pro isn’t a full replacement for high-end motion graphics or complex, pixel-perfect work that still demands tools like After Effects or native Affinity desktop apps. - Teams pricing and add-ons have increased as AI features expanded. - Advanced brand governance and security features sit behind Teams/Enterprise tiers, so freelancers on individual Pro plans may not get full governance controls. ### 8. The Trade Desk (DSP) The Trade Desk is an independent demand-side platform that lets advertisers buy media across the open internet from one data-driven, self-serve interface. It’s built for sophisticated media buyers who want omnichannel reach, granular controls, and transparent reporting, rather than a walled garden. The platform’s new Kokai interface and Koa AI engine analyze millions of bid opportunities per second, while its identity, retail data, and measurement marketplaces help brands connect impressions to real-world results. Best understood as a high-end media-buying operating system for the open web, it’s a better fit as the backbone of a scaled programmatic program than as a casual tool for small teams dabbling in display. Price The Trade Desk charges a platform fee as a percentage of media spend, rather than a flat SaaS license, typically in the mid-teens to ~20% of spend. Most advertisers negotiate pricing via The Trade Desk or through an agency/reseller, and the DSP generally expects meaningful monthly budgets, which is why smaller brands often access it through partners that aggregate spend and lower minimums. Special Features - Premium omnichannel inventory at scale: Access to CTV/OTT (over-the-top), video, display, audio, native, mobile, and digital out-of-home (DOOH) across the open internet, with strong positioning as a preferred DSP for CTV and premium publishers. - Kokai and Koa AI decisioning: Kokai is the AI-powered interface layered on top of the Koa machine-learning engine, which adjusts bids, audiences, and supply in real time. - Audience-first, identity-driven targeting: Deep support for first-party and third-party data, plus industry identity frameworks like UID2 and EUID. - Retail data and retail media integrations: Marketplace of retailer partners that lets you use retail and transaction data for targeting. - Measurement and optimization marketplace: Plug-and-play integrations with third-party measurement, attribution, brand lift, and incremental reach providers. - APIs and custom solutions: Extensive API access, log-level data, and custom solution support. Pros - Independent, open-web focus means it’s not conflicted by owning media, and buyers get broad access to premium inventory. - Strong CTV and omnichannel capabilities make it a go-to DSP for brands that want to unify CTV, video, display, and audio in one place. - Powerful AI optimization, while buyers still retain a high level of manual control. - Transparency and detailed reporting around fees, placements, and performance. - A safe, scalable choice for enterprise advertisers and agencies. Cons - It's a sophisticated trading platform that can be difficult for new or lightly staffed teams. - The percentage-of-spend model and typical budget expectations mean it may be overkill for small advertisers. - While it can optimize and report on creatives, you still need separate workflows (or partners) for asset creation, dynamic templates, and rich creative production. ### 9. Outbrain Outbrain is an open-web native advertising and content-discovery platform that places your ads as “recommendations” across a network of premium news and lifestyle publishers. Its AI-based platform uses an Interest Graph and Smartfeed/Smartlogic recommendation engine to serve personalized sponsored content, video, and display units in-feed and in-article. Although the corporate brand has merged into Teads, the Outbrain technology and advertiser workflows (Amplify, Smartads, Smartfeed) are still widely used across media partner sites. The tradeoffs are uneven customer reviews and a platform that’s now part of a larger Teads ecosystem. Price Outbrain primarily runs on a CPC model in its self-serve Amplify dashboard: You set a bid per click and a daily or campaign budget, and you’re charged only when users click your native ads. Typical CPCs vary by geography, vertical, and competition, but industry and network benchmarks suggest native CPCs often fall roughly in the 10-cent to 50-cent range. Special Features - Smartfeed and Smartlogic personalization: Outbrain’s Smartfeed and Smartlogic engine uses AI and machine learning to personalize recommendation feeds. - Interest Graph targeting: An Interest Graph built from browsing behavior across thousands of sites powers interest-based targeting. - Smartads format suite: A portfolio of performance-focused native formats, including Standard Smartads, Carousel Smartads, App Install Smartads, Outstream Video Smartads, and Click-to-Watch units, designed for everything from awareness to app installs and conversions. - Broad format coverage: In addition to classic native tiles, Outbrain supports Clip (short animated units), App Install and Carousel natives, pre-roll and outstream video, high-impact display, and standard banners. - Conversion Bid Strategies and performance AI: Multiple automated bidding options adjust CPCs and serving patterns based on pixel data and on-site behavior. - Measurement, pixel, and analytics integrations: Native conversion pixel, support for codeless conversion, and integrations with Google Analytics (GA/GA4), Voluum, and other analytics platforms track down-funnel outcomes from native campaigns. - Programmatic access via DSPs: Outbrain inventory can be bought directly through the Amplify UI or programmatically via major DSPs. Pros - Access to premium, brand-safe environments on the open web. - Strong personalization and targeting via Interest Graph, Smartfeed, and Smartlogic. - Broad format mix and Smartads options mean you can support full-funnel objectives within one platform. - Self-serve UI is relatively easy to learn. - For content marketers and performance advertisers, Outbrain often delivers cost-effective CPCs and incremental traffic outside walled gardens. Cons - Targeting depth and controls, while improved, are sometimes seen as less granular than major social platforms or top DSPs, particularly in lower-touch self-serve accounts. - Minimum daily budgets, bid recommendations, and priority for larger spenders can make it feel less friendly to very small advertisers. ### 10. TripleLift TripleLift turns standard ad assets into high-impact native, display, and CTV experiences across the open web, streaming, and retail media. It sits on the supply side, but its core pitch to advertisers is creative innovation at scale with native tiles that match publisher layouts. TripleLift combines creative tech, data, targeting, and a premium SSP to deliver more engaging formats while still transacting programmatically through your existing DSPs. You will still need DSP access and trading chops, and while its formats are differentiated today, you should view TripleLift as part of a broader, competitive creative-media ecosystem rather than a one-stop solution. Price TripleLift doesn’t charge advertisers a separate SaaS license. Instead, its inventory is bought on a CPM basis via integrated DSPs, with TripleLift earning an SSP “take rate” baked into the media cost. It has no platform licensing fees when you access it through your DSP, so your effective cost is your DSP fee plus TripleLift’s margin on the media. Special Features - Creative SSP positioning: Marketed as “the creative SSP,” TripleLift specializes in formats that blend into content and elevate brand metrics. - Native and display formats: A broad portfolio of native tiles, scroll/carousel units, collection/flipbook format, and standard display ads that match each publisher’s look. - High-impact CTV and Pause Ads: CTV Spots, Pause Ads, Split Screen, dynamic overlays, and In-Show placements that insert brands into streaming content. - Advanced creative technology: Tech that automatically adapts standard assets into native, display, and CTV formats, which optimizes in real time for engagement and performance across screens. - Data and targeting stack: TripleLift Audience and identity partnerships combine advertiser first-party data with publisher data for privacy-conscious, cookieless targeting. - Premium inventory and SSP: Access to high-quality publishers through TripleLift’s SSP with real-time bidding, supply-path optimization, and integrations. Pros - Native, CTV, and in-show formats often outperform standard banners on attention, awareness, and purchase intent. - Easy to add to existing workflows. - Premium publisher and CTV supply with brand-safe environments and growing coverage across web, streaming, and retail media. - Generally no extra platform fee to buyers beyond CPMs and DSP costs. - Rapid innovation in Pause Ads, In-Show, and commerce formats. Cons - You still need a DSP (and trading expertise) to plan, buy, and optimize campaigns. - Competitors in native, CTV, and retail media are rolling out high-impact formats, so TripleLift’s current tech advantages may narrow over time. - Agencies and brands may need to push for detailed SPO analysis via their DSP partners. - Brands with very small spend or no CTV/native strategy may not fully benefit from the advanced formats. ## Key Takeaways Which platform fits best for you depends on where AI sits in your workflow, but with prediction and testing becoming as important as generation, platforms that pair GenAI with performance signals will help teams decide which creatives to run, not just create more of them. Remember, too, that AI creative works best as part of a broader media ecosystem, and generation tools deliver the most impact when paired with strong distribution and optimization layers. The point is to turn AI creatives into measurable performance, not just assets. ## Frequently Asked Questions (FAQs) ### How should I test whether motion or static creatives will drive higher CVR for my brand? The most reliable way to test creatives is with a clean, controlled A/B test. Run motion and static creatives side by side while keeping everything else the same — audience, placement, bidding, and campaign settings — so the type of creative is the only variable. Launch them both at the same time, making sure each gets enough traffic to reach statistical significance, and track the same conversion event for both. If your platform supports it, you can layer in multivariate testing to see how motion or static formats interact with different hero images, headlines, and CTAs. When you analyze results, prioritize CVR and CPA over CTR, and look at post-click signals like time on site and bounce rate to confirm you’re not just driving clicks, but quality conversions. ### I have limited creative resources. What’s the fastest way to scale motion ads from existing static assets? Start by identifying your best-performing static creatives with the strongest CTR or CVR and use them as the foundation for motion, rather than trying to animate everything at once. Templates and DCO-style tools make this efficient by letting you add lightweight motion to key elements like the hero image, product, logo, or CTA in short three- to 15-second loops that perform well without heavy production. You can also save time by repurposing social motion assets with minor format and size adjustments for open-web placements. Before rolling motion out broadly, put a small budget behind a live test to confirm lift versus static, then scale only the formats and messages that prove incremental impact. ### How do I ensure my creative tests are not biased by placement or inventory differences? The key is to control everything except the creative itself. Run all creative variants at the same time, in the same placements, and against identical audience segments so no version benefits from better inventory or timing. Set equal budgets, bids, and pacing rules for each test cell, and use randomized assignment so impressions are distributed fairly. Make sure every variant is measured against the exact same conversion event, then sanity-check the results by reviewing viewability, invalid traffic, and fraud metrics to confirm that one creative wasn’t favored by higher-quality supply. You can also run your static vs. motion tests on Realize with the same site lists and targeting. Realize’s direct publisher integrations provide consistent, brand-safe supply and richer user signals so the platform can surface valid creative winners. Use Double Verify/Integral Ad Science (DV/IAS) pre-bid filters and allow/block lists inside the platform to keep supply consistent across test cells. --- ### How to Set Up Your Performance Campaigns in 7 Steps URL: https://www.taboola.com/marketing-hub/how-do-advertisers-campaign-set-up/ Last Modified: 2026-06-25 09:54:01 The walled gardens of search and social advertising channels have served an important purpose for performance marketers for many years. Meta’s and Google’s owned channels have been the default for brands to drive conversions, but cost increases and stiff competition are changing that. Open web advertising has emerged as a vastly scalable option, with high-intent audiences visiting publishers and other media sites. Advertising on the open web is incredibly different from those search and social channels. In particular, campaign set up and optimization can be manual and disjointed, taking up way too much time for performance marketing teams. AI-driven technology has matured to the point where it can serve as a workflow copilot, helping teams launch open web campaigns more quickly and easily than with traditional channels. AI capabilities now inhabit every step of the campaign setup process, from generating creative to doing predictive audience discovery and targeting. ## The Evolution of Campaign Set Up: Enter AI and Automation Newer AI-driven campaign setup tools can’t come soon enough to help boost performance marketing scale. The era of tedious, manual campaign configuration is finally ending as AI automation platforms mature. These modern platforms use conversational AI and LLMs, so marketers can type goals in plain text or simply speak instructions to start the AI automatically building media plans and targeting parameters. The power of AI for campaign set up is not only that it saves time, but that users get results. Realize’s Abby generative AI tool, for example, works to increase advertiser productivity and pull in trusted data to support campaigns. These types of performance platforms can walk an advertiser through every step of setting up and managing a campaign, including AI ad optimization, targeting the right audience, testing and optimizing creative, and setting budgets correctly. When you’re creating a new campaign on the open web, the scale is massive compared to walled gardens, so you have to make sure you get foundational tracking and signals set up correctly and do robust AI training for a solid start. ## How Do Advertisers Execute a Full Campaign Set Up in 7 Steps ### 1. Defining Your Performance Objectives and KPIs AI can do a lot of the execution and analysis, but the human directing the performance marketing campaign has to define goals up front, to ensure the AI’s success. Make it clear whether the campaign goals are related to lead generation, sales, ROAS, brand awareness, or other options. This first step trains the AI models on the specific conversion actions they should optimize for during the learning phase. ### 2. Configuring Conversion Tracking and Data Signals The conversion tracking configuration plays a key role in successful AI campaign usage. This step involves setting up server-to-server (S2S) tracking and/or conversion pixels as well as post-back URLs. These data signals serve as the fuel for AI algorithms to optimize ad delivery and identify winning patterns, then continue learning from that information going forward. ### 3. AI-Powered Audience Discovery and Targeting Static demographic segments have become stale, as predictive audience targeting gains ground and becomes easier with help from AI tools. Machine learning and AI can analyze contextual signals, semantic data, and predictive purchase behaviors on the open web. This works even in privacy-first, cookie-less environments, where so many marketers are operating. ### 4. Structuring Campaigns for Machine Learning Success Ad campaign architecture needs to provide a solid foundation for AI and ML advertising campaigns. Make sure to consolidate ad groups and avoid over-segmenting — these tactics allow AI algorithms to gather sufficient data density and shorten the learning phase. Broad Targeting vs. Niche Segmentation AI algorithms for advertising campaigns perform better with more data, so setting up wider initial targeting parameters makes the most sense. This trade-off between narrow targeting and broader reach generally pays off down the road, once the AI has learned from the data. Funding the AI Learning Phase While it may seem counterintuitive to allocate budget for this initial AI launch and learn phase, it’s absolutely necessary to give the AI algorithm a sufficient base of data. Make sure to allocate enough budget in this stage so there are enough conversion events feeding data to the algorithm, so it can optimize effectively. ### 5. Leveraging Generative AI for Creative Asset Production Generative AI creatives can bring the power of a video studio and design team to even the smallest business. Many modern ad platforms offer built-in gen AI tools to create hundreds of ad variations, headlines, and images, all based on as little as a site URL. Beyond saving teams time and money, these tools make it easy to quickly do the A/B testing that performance marketing needs to succeed. ### 6. Automated Bidding and Predictive Budget Allocation AI-driven media buying is another emerging feature of modern performance platforms that can save tons of time. Automated bidding strategies powered by AI focus on a target metric, like cost per acquisition (CPA) or return on ad spend (ROAS). With that goal, predictive analytics can adjust bids in real time across thousands of publisher sites, based on the likelihood of a user converting. This adds precision to bidding, so you can spend the right amount of money in the right place at the right time. ### 7. Launch, Monitor, and Optimize Before you click “launch” on this new campaign, always do a final quality assurance process. AI is ideal for micro-optimizations, real-time bidding, and creating new ad variations, but the human advertising team has to monitor high-level metrics, work against creative fatigue, and guide the overall strategy. ## Overcoming Measurement Challenges on the Open Web Tracking user journeys outside the closed walled-garden ecosystems brings more complexity to an advertising campaign. The right tools can simplify and automate a large part of the tracking process to take full advantage of the open web’s conversion potential. Performance marketers have to move beyond last-click attribution and incorporate AI-based predictive modeling to get full-funnel visibility and cross-channel attribution and tracking. ## Key Takeaways Open web success needs a solid campaign foundation and best practices of steering AI toward business goals, instead of manual micromanagement. Getting an advantage in the digital ecosystem today requires speed and the right tools to help you automate the bulk of digital campaign execution. Make sure to prioritize data signals, generative tools, and automated bidding to stay ahead of competitors on the open web. ## Frequently Asked Questions (FAQs) ### How does AI improve the campaign setup process on the open web? AI can improve the entire campaign setup process by making it much simpler, replacing repetitive, manual tasks with automation. AI tools bring hyper-personalized targeting and can immediately generate ad creatives from a URL, as well as suggesting optimal budget allocations, automating bidding strategies, and using ML to build high-intent audience segments. Many modern tools can ingest plain text or natural language goals, so advertising teams can focus on strategy rather than manual set up tasks. AI tools can reduce production time for campaigns dramatically. ### Can I achieve the same ROAS on the open web as I do on search and social? Yes, it’s possible to achieve the same or better ROAS with open web campaigns, since AI bidding algorithms and predictive targeting can identify users with high purchase intent in real time. This can save budget as well, since marketers are accessing premium ad inventory that’s less saturated than search and social channels may be. ### What is the best bidding strategy for a newly launched open web campaign? For a newly launched open web campaign, start with an automated strategy that focuses on maximizing conversions. This feeds initial data to the algorithm, which it needs to produce top-notch, accurate results later. Once there is sufficient conversion density, you can switch to target CPA or target ROAS so the AI can optimize for profitability. ### Do I still need third-party cookies to target audiences effectively? No, you don’t need third-party cookies for targeting any longer. Now, advanced ad platforms use AI to analyze real-time behavioral signals, such as contextual and semantic. This allows for precise, persona-based targeting that’s independent of third-party cookies. --- ### Three Key Variables to Test in Ad Headlines for Enhanced Performance URL: https://www.taboola.com/marketing-hub/what-to-test-top-optimize-ad-headlines/ Last Modified: 2026-05-24 07:54:49 Advertising on the open web requires precise copy and eye-catching creative to get the attention of audiences. The most effective ad campaigns on the open web — whether for leads, e-commerce, or reach — rely on a headline that performs two jobs: filtering the right audience and hooking their curiosity. Creating a great headline that works for your performance marketing goals requires audience knowledge and testing. Every detail matters when you’re crafting this piece of short copy to make sure it will perform for your campaigns. To that end, we consulted Realize experts Rubi Das and and JeQuan Norris, SMB advertising sales managers, who between them bring years of experience to writing and testing advertising headlines. Here, we’ve gathered the three essential variables to test in your ad headlines, with details on what works best and why. ## The 3 Essential Variables to Test Your Ad Headlines ### 1. The Audience Filter: Specificity Over Generics When advertising on the open web, remember that the primary function of a headline is not to sell your product or service, but rather to pre-qualify the click. By naming your audience or the problem they’re facing, you can maximize the quality of the traffic that enters your funnel — and immediately sift out those who aren’t interested. Follow these steps to filter your audience up front: Specify the Persona or Problem  The headline should act as a litmus test to see if you’re reaching your target user: If the user does not relate to the headline, they won’t click. Remember to continuously A/B test your headlines to make sure you’re running the most effective version possible. “There is a common temptation to over-segment your audience by interest or niche right at the start, but that can actually restrict your data pool,” says Rubi Das, Realize Advertising Sales Manager. “I recommend starting without restrictive interest targeting. By letting the algorithm explore the ecosystem, you find the clicks that are truly viable. You might think your audience is only in one category, but there are massive overlaps. Use the testing phase to discover who they are, rather than assuming who they are up front.” Run Headlines That Anchor to a Specific Category Let the reader know immediately what category you’re advertising in, particularly if your business spans multiple sectors, such as clothing or electronics. You’ll have more success with specificity on the busy open web, especially with ad-fatigued users. Align Headline to Post-Click Content  Make sure there is a clear link between the headline's promise and the content's fulfillment. That includes the headline structure and ensuring that the headline's specificity matches the landing page content, and meets the expectation that has been set for the user. Be wary of using a broad headline that delivers a disappointing landing page. "When you’re targeting high-intent environments like the Yahoo or AOL Mail inbox, your audience is already logged in and in an 'action' mindset," says JeQuan Norris, Realize SMB Advertising Account Manager. "In these spaces, you can’t afford to be vague. You need very tight headlines that are directed to the point. The headline shouldn't just be a soft sell; it should be a mirror of the product experience they are about to have on the landing page." ### 2. Use the Curiosity Hook: Inform, Don't Sell Second, make sure you’re testing headlines that generate curiosity. This approach turns the ad into a recommendation rather than a sales pitch, which is key to overcoming ad fatigue and ensuring clicks onto the page where users can read further. Our experts recommend these tried-and-true methods for generating curiosity in headlines: Use Discovery Hooks  When you’re advertising on the open web, there are some particularly effective hooks, such as headlines that begin with or include terms that suggest exclusive, new, or insider information. They appeal directly to the reader's information-seeking mindset. Test Popular Phrasing Try headline phrasing that taps into reader curiosity — e.g., "See Why Is Soaring" or "Discover the 3 Key Signals Experts Are Watching" — for your particular category, to see what engages the audience. "Curiosity is a powerful tool, but it works best when it's paired with fresh data," notes Norris. "If a campaign starts to dip, it’s often a sign that your 'hook' has reached its limit with a specific audience. I always advise advertisers to treat their top publishers as living ecosystems—don't just block a site if it stops performing. Instead, use that as a signal to test a new creative angle or a trendier headline. Sometimes a simple tweak in phrasing is all it takes to unlock a new wave of qualified clicks from a premium publisher." Frame the Ad as a Time-Sensitive Alert  Headlines should convey an immediate need to know, without making claims that aren’t compliant with industry regulations, if applicable. For financial offers, for example, this means testing alert-based phrasing like "Investor Alert," "Time to Act," and "Don't Miss Out." Conduct Actionable Testing  You might use A/B testing for ad headlines, such as by comparing a purely informational headline ("A Report on Sector X") with an alert-based headline ("Investor Alert: Is Moving Now"). “When testing new curiosity-driven hooks, the number of variations is just as important as the copy itself,” Das advises. “For any single campaign, we recommend limiting your creatives to between six and eight variations. If you populate a campaign with too many headlines at once, you risk diluting your budget. You want enough variations to find a conclusion on what works, but not so many that the algorithm doesn't have the spend to properly test each one.” ### 3. Striking a Tone Balance: Compliance vs. Performance The final variable is testing the tone and word usage of an ad headline to maximize click-through rate (CTR), while operating safely within the platform's guidelines — a non-negotiable for stable scaling. Here’s what our experts recommend: Avoid Deceptive or Risky Wording  Since ad platforms prioritize compliance, the aggressive get-rich-quick tone common in other channels must be replaced with a professional, curiosity-driven tone. Make sure to double-check with your review team if there’s any possibility a headline might be categorized as misleading. "Scaling on premium tier publishers like MSN or Yahoo requires a higher level of 'branded' professionalism," Norris explains. "We find that headlines which lean too heavily into gimmicky or 'blind' curiosity often get flagged or limited by top-tier sites. To win in these auctions, your tone needs to be authoritative and heavily branded. By moving away from aggressive 'get-rich-quick' language and toward a more established, branded voice, you gain much more leverage to fight for placements on the web’s most valuable real estate." Consider Safe Wording Alternatives Along the same lines, think about how to use more platform-safe phrasing. Instead of, “Act now or lose!” for example, you could use the more thoughtful, “Why timing matters right now.” Or, instead of, "Guaranteed returns,” you could try, “What experts are watching.” “This is especially valuable for finance, health, and regulated categories,” says Das. Conduct Actionable Testing  When you’re considering the right tone balance for your ad campaign headlines, test language that focuses on education (e.g., "Learn How") versus guarantee (e.g., "You Will Get"). Maintain Consistency Across the Funnel  The tone used in the headline should match the tone of the body copy and the image. Even if the copy is specific to the image, make sure to take that into account, so the tone remains consistent to maintain user trust and funnel continuity. “Consistency isn't just about tone; it’s about language and localization,” says Das. “If you are promoting a product in a specific market—like Germany, for instance—your headlines must mirror the landing page exactly. If your page is in German, your headlines must be in German. A catchy headline in English that leads to a non-English landing page creates an immediate bounce. To maintain trust and performance, the language of the 'hook' must be the language of the fulfillment.” You may want to test image/headline pairs as a single creative unit to ensure tone match and consistency. ## Key Takeaways Ad campaigns won’t succeed without the right headline, whatever the end goal is. Effective ad headlines should both filter for the right audience and pique their curiosity to learn more. Advertising experts suggest creating and testing informative, relevant, consistent headlines to ensure the right targeting and reach. ## Frequently Asked Questions (FAQs) ### How can I efficiently test multiple headline variations on the open web without manually creating dozens of individual ads? The best practice for scaling creative testing like this is to use a platform's ad variations functionality, which is available through Realize. Instead of creating 10 separate ads, you can input multiple headlines, descriptions, and images into a single campaign element. The platform's algorithm will then automatically mix and match all possible combinations and use its machine learning to prioritize the highest performing pairs based on real-time data. This allows you to rapidly identify the winning headline/image pairing without manual rotation or micromanagement. ### We need to boost our click-through rate (CTR) to lower our cost per click. What platform tool can help generate fresh, high-performing headlines on demand? To accelerate creative ideation and combat ad fatigue, use Realize’s artificial-intelligence (AI)-driven creative assistance tools (often referred to as a "Help Me Write" or similar feature). These features use generative AI to analyze your existing copy and offer several immediate alternatives. You can input a base headline and the tool will instantly generate variations that use popular curiosity-based phrasing, emotional hooks, or urgency triggers, giving you a continuous stream of new ideas to test against your winners and drive a higher CTR. ### How can I ensure the specific demographics or audiences I mention in my headlines are actually performing well after the click? To move beyond simple ad performance (CTR) and analyze post-click lead quality, use Realize’s advanced reporting segmentation capabilities. By analyzing your campaign data and segmenting by the specific ad ID or headline text, you can isolate the conversion rate using the code ($\text{CVR}$) and cost per acquisition rate using the code ($\text{CPA}$), which are delivered by headlines targeting a specific persona (like "Seniors," "Investors," etc.). This allows you to verify that the headline that generated the cheapest click is also producing the most profitable lead, ensuring your creative testing is tied directly to bottom-line results. --- ### Attribution Bias Exposed: Are Google and Meta Favoring Their Ads Over Your ROAS? URL: https://www.taboola.com/marketing-hub/attribution-bias/ Last Modified: 2026-05-28 12:46:25 You open your marketing dashboard expecting strong results. Meta claims 50 conversions. Google takes credit for 45. But, your Shopify store only shows 60 actual sales. The math doesn’t add up. Is there a glitch or technical error? No — what you’re seeing is attribution bias. Self-attributing networks (SANs) like Google and Meta sell you ad inventory and measure their own performance, so they have every reason to make their numbers look as good as possible. What you get instead is inflated return on ad spend (ROAS), double-counted conversions, and a budget strategy built on data you can’t trust. ## What Is Attribution Bias in Marketing? Marketing attribution bias is a systemic error in how credit for conversions gets assigned — not a random mistake or a broken pixel, but a structural flaw baked into how performance is measured inside walled gardens. It doesn't occasionally skew results; it consistently shifts credit toward the platform doing the measuring, so your reported performance is reliably more flattering than reality. ## The Conflict of Interest: Why Ad Platforms Grade Their Own Homework Google and Meta are advertising businesses. Their revenue grows when you increase spend, so their reporting systems are built to reinforce the value of that spend and unlock more. When the same company controls both ad delivery and measurement, competing platforms and other touchpoints in your mix — like email, push notifications, or organic search — are more likely to be under-credited in favor of its own channel. In other words, the judge and the contestant are one and the same. ## Walled Gardens and Self-Attributing Networks (SANs) Explained A SAN, sometimes called a self-reporting network (SRN), is a platform that handles attribution internally and reports aggregated results without sharing the raw data behind the claim. Meta, Google, and TikTok are classic walled garden advertising environments: they control the inventory, the tracking, and the measurement, and you can’t fully audit the logic behind the numbers they report. ## The Illusion of Success: Why Your ROAS is High but Revenue is Flat Here’s how the illusion works: a user sees your Meta ad on Monday, clicks a Google search ad on Thursday, gets a discount email on Saturday, and buys on Sunday. Under Meta’s seven-day click and one-day view setting, Meta may still claim that conversion, and Google may claim it too. That’s how double-counting conversions leads to incorrect ROAS. With no cross-channel deduplication, a single sale can appear multiple times across your dashboards. The result is strong-looking ROAS, flat revenue, and budget decisions based on numbers that overstate what your campaigns are actually driving. ## 5 Common Types of Attribution Bias Distorting Your Data Not all attribution bias works the same way. Some forms come from how platforms assign credit, while others come from the ways that campaign systems optimize for the easiest conversions. ### 1. Platform Self-Attribution Bias Ad networks overclaim credit when they act as both ad server and measurement source. Google’s Performance Max (PMax) and Meta’s Advantage+ can be prone to this because they’re black-box, AI-driven campaign types that control targeting, bidding, and attribution in a closed loop, leaving advertisers with less transparency into how credit is assigned. Both may claim credit for conversions that were already in motion before the ad was served. This is also where cheap inventory bias shows up. Low-cost impressions can appear efficient simply because they occur near conversions that were already likely to happen, allowing the platform to report success without proving true lift. ### 2. In-Market (Retargeting) Bias In-market bias happens when algorithms target users who are already close to buying, then claim credit when they convert. The result is strong ROAS without proven incremental growth. Your incremental ROAS is often much lower than the platform-reported number. ### 3. Correlation-Based Bias Seeing an ad before converting doesn’t mean the ad caused the conversion. Correlation-based bias treats exposure as causation and gives full credit to a touchpoint that may have had little real influence. ### 4. Channel Proximity/Last-Click Bias When attribution defaults to last-click, the touchpoint closest to conversion gets all the credit, while the tactics that built awareness — display, native, video, email — get nothing. Advertisers cut upper-funnel spend, see short-term efficiency, and then watch the pipeline shrink. ### 5. Digital-Only Bias If it can’t be tracked with a pixel, it won’t show up in platform reporting. Events, sales calls, direct mail, and in-store visits all impact purchasing decisions, but none register in Google or Meta attribution, unless you take full advantage of CAPI, Measurement Protocol, or other forms of offline reporting in order to track your full funnel. This skews budget toward digital channels because they’re the only ones being measured. ## The Hidden Cost of Biased Attribution on Your Ad Budget When you trust platform-reported ROAS, retargeting looks efficient and awareness looks wasteful, so budget shifts accordingly. Over time, your new customer acquisition cost (nCAC) rises as you rely more on existing demand, rather than generating new demand. By the time revenue shows the damage, you’ve already cut the channels that keep the pipeline full. ## How to Expose and Fix Attribution Bias in Your Campaigns You can’t simply force SANs to become neutral measurement systems, but you can reduce how much control they have over your reporting and budget decisions. ### Implement Independent Third-Party Attribution Third-party attribution tools and mobile measurement partners (MMPs) ingest data from all channels, deduplicate conversions, and create a single source of truth outside the platforms. When Google and Meta both claim the same sale, it gets counted once and assigned by your model, not theirs. (To be clear, Google Analytics 4 does count as third-party, as it deduplicates conversions.) ### Incorporate Marketing Channel Data into Your CRM Make sure that every record in your CRM is infused with marketing-related data — the campaign channels the user clicked through to your website, and their unique identifiers (make sure you handle these in a manner that is compliant with privacy regulations such as GDPR). This way you’ll have the raw data necessary for painting a fuller picture of your user journey, and craft your attribution model accordingly. ### Take Advantage of Triangulated Measurement Simply put, no single attribution model can account for 100% of your conversions or ROAS. Only the combination of marketing mix modeling (MMM), incrementality testing, and a solid attribution model brings you closer to understanding exactly what drives your business. MMM is useful for strategic planning and accounting for external factors, while incrementality testing reveals whether an ad actually caused a sale, or if the customer would have purchased anyway. Meanwhile, attribution data remains best for the day-to-day management and optimization of campaign budgets. However, using these three methodologies together helps each methodology complement the others and serves as a powerful compass to guide your marketing efforts. ### Focus on New Customer Acquisition Cost (nCAC) Shift your primary metric from platform-reported ROAS to nCAC. It forces you to measure actual incremental growth: are you acquiring customers who haven’t bought before, or just recapturing existing demand? ### Use Server-Side Tracking and First-Party Data Signal loss from iOS 14+, ad blockers, and cross-device behavior creates gaps that platforms usually fill with modeled data. Server-side tracking helps close those gaps by sending high-quality conversion data directly from your server, rather than relying solely on browser-side pixels. Platforms like Realize support server-to-server (S2S) conversion tracking alongside pixel measurement. That helps advertisers connect campaign spend to verified customer relationship management system (CRM) events and order IDs, including conversions pixels may miss, such as phone orders, in-person appointments, and in-app events. Realize+ also supports MMP integrations for accurate attribution across web and app campaigns. More broadly, Realize+ is an open-web alternative to closed-loop systems like PMax and Advantage+, designed to deliver performance without platform bias. ## Should You Ditch Platform ROAS Completely? No, platform ROAS is still useful for in-platform optimization. Meta’s algorithm and Google’s Smart Bidding both need conversion signals to work, and platform reporting can still help with creative testing and within-channel comparisons. What it can’t tell you is how your overall budget is performing across channels. For business-level allocation decisions, you need third-party attribution tools that deduplicate conversions and apply the same logic everywhere. ## Key Takeaways Attribution bias is structural, not accidental, and without cross-channel deduplication, double-counting conversions is the norm. Biased attribution quietly raises nCAC by over-investing in retargeting and underfunding awareness, which is why you should use platform ROAS for in-platform optimization, but rely on third-party attribution tools for budget allocation. Use MMMs and incrementality tests to validate your attribution efforts. In addition, server-side tracking, MMPs, and nCAC will give you a stronger measurement foundation. ## Frequently Asked Questions (FAQs) ### What is biased attribution in marketing? Biased attribution occurs when a platform has a financial conflict of interest in the outcome it’s reporting. Because Google and Meta sell the ad space and measure the results, they may over-credit their own touchpoints. ### Why do Meta and Google overstate my ROAS? Both platforms use generous attribution windows. If a user sees a Meta ad and later clicks a Google ad before buying, both platforms can claim credit for the same sale. That’s how double-counting conversions leads to inflated ROAS. ### How do I fix ad platform attribution bias? Use third-party attribution tools or an MMP to deduplicate conversions across channels. Server-side tracking can also help by capturing verified conversion events that browser pixels miss, tying them back to real business outcomes. --- ### The ROI Case for Agentic AI: What the Performance Numbers Actually Say URL: https://www.taboola.com/marketing-hub/agentic-ai-roi/ Last Modified: 2026-05-24 07:23:13 As AI usage and tools become more sophisticated, business leaders are asking tougher questions about how it can drive growth and save on costs. To build a performance marketing return on investment (ROI) case and long-term strategy for using agentic AI, adopters need real evidence of how it can drive business growth. But, there hasn’t been enough data to really understand how marketing teams are getting value from AI, particularly agentic AI, in their daily work. That’s why the Realize team commissioned a survey specifically to understand the current landscape: We asked 200 senior performance marketers at companies spending $300,000 or more per month what they are seeing from their AI platforms. The results left no doubt that AI is already leaving its mark on performance marketing. ## The Performance Proof: What 76% of Marketers Are Actually Seeing The survey respondents are all people responsible for real performance budgets, and their response was overwhelming: 76% of them are seeing meaningful improvement from their use of AI-powered platforms such as Meta’s Advantage+ and Google’s Performance Max (PMax). To further break down that 76%: - 47% of respondents see a moderate performance lift. - 29% of respondents are seeing a significant lift from their AI platform. For 7% of respondents, they see a limited lift with their AI-powered solution. Just 1% report no impact on performance. For the vast majority of respondents, AI-powered campaign platforms are delivering visible, valuable impact. ### Why “Measuring But Too Early” is a Bullish Signal 17% of the decision-makers surveyed say they’re measuring the results of their AI-powered solution, but that it’s still too early to gauge the impact. Crucially, though, these survey respondents are conducting testing on how AI-powered platforms can work for their needs: they’ve committed budget and assigned resources to execute proof-of-concept projects and pilots. Platform adoption patterns suggest that when measurement is complete, it resolves positively for companies committed to performing testing. ## Why CPA/ROAS Optimization Is the Core Mechanism Driving Lift The survey data shows a clear picture of how agentic AI can help marketing and advertising teams. 41% of respondents say the top benefit of AI-powered solutions is the real-time optimization toward cost per acquisition (CPA) and return on ad spend (ROAS) — two of the most important metrics that advertisers use to gauge success and growth. Time savings was the second-ranked benefit for 14% of respondents, while improved budget allocation was top for 11% of respondents. These numbers show how important agentic AI results are to performance marketing teams, who are held accountable for specific, growth-focused metrics. Real-time ROAS or CPA optimization is far more valuable to these teams than the time savings, reporting, or creative automation, because it’s doing the right work at the right time faster than any human team could execute. All of the above is why it makes sense that Meta Advantage+ and Google’s PMax have achieved near-total adoption. Those platforms deliver for performance marketers against the metrics that matter most to them. This is also why agentic AI in performance marketing solutions can scale so well, and why it can help new channels grow very quickly. ### The Gap Between What Humans Optimize, and What Machines Can Real-time CPA and ROAS optimization for performance marketing is an ideal use case for agentic AI, because it offers a solution to a problem that humans alone can’t solve. Day-to-day, this type of operational optimization work happens continuously — adjusting bids, reallocating budget, rotating creative work according to its performance, and refining audience targeting. Human teams can only do this work in pieces, checking a dashboard once or twice a day, with a longer optimization cycle. Real-time, AI-driven performance marketing tools can respond to each signal as it arrives. Thousands of micro-optimizations compound quickly, as they’re completed faster and more consistently than any human team could execute. Over the course of a campaign, that cycle speed difference makes a big impact. The performance lift from agentic AI doesn’t rely on one concerted effort by human teams, but on continuous, back-end technical work. ## Making the Financial Case for Diversification Among Diminishing Returns This kind of performance lift in CPA/ROAS is a primary driver of budget decisions, too. Ad spend efficiency — ROAS — continuously drives performance marketing decisions, with search and social channels typically making up most of the budget. The survey found that 74% of respondents allocate 25% or more of their budget to paid search, and 67% do the same for paid social. These are the most heavily resourced channels in the performance mix and, tellingly, the biggest spenders are hitting the ceiling of these channels most acutely. ### When More Spend Stops Meaning More Results Search and social have served businesses very well, and agentic AI has proved in the survey data to be an effective tool to drive search and social performance. While search and social remain the dominant channels, though, they’re becoming saturated. That saturation shows up for performance marketers as diminishing returns based on the same or increased spending. Once marketing teams have optimized everything they can in search and social, incremental growth has to come from somewhere else, and that’s the open web, based on the data. A clear 70% of survey respondents spending more than $5 million per month say that it’s “extremely important” to find an incremental performance channel. Three quarters of survey respondents said it’s either very or extremely important to identify a performance channel to deliver incremental performance uplift beyond the walled gardens of search and social. Budget concentration in search and social reflects the typically strong returns of those channels, but the concern is scale. Returns that were strong with an investment of $500,000 a month does not automatically scale proportionally to $2 million or $5 million a month. Increased audience saturation and high-intent keyword competition mean that the cost of reaching an incremental customer just keeps rising. ## Budget Allocation Today vs. Where It’s Going Next The need for channel diversification is critical in the face of saturated performance advertising channels, and the survey data shows the decision point that many marketing leaders are facing. The current investment looks like this for performance marketing budgets, based on survey responses, with numbers reflecting the allocation of total budget: - Paid search: 22%. - Paid social: 21%. - Open web: 13%. - Connected TV (CTV): 12%. - Retail media networks (RMNs): 9%. - Affiliate: 8%. For the future of budget allocation, though, almost all of the survey respondents — 99% — say they would allocate budget to the open web if there were agentic AI solutions to use there. The average expected allocation would be about 24%, double what it is today, with the intent to invest even more strongly among the biggest spenders. That increase in investment in the open web would shift its importance in the channel mix and in the performance budget. With search and social remaining strategic pillars, new budget investment on the open web would be an ideal opening for performance marketers. ### What the Budget Intent Data Actually Says There’s a stark contrast between the 4% of performance marketers investing significantly in the open web, compared to those who would like to invest if agentic AI solutions existed for it — 39%. Those respondents say they would invest 26% or more if agentic AI solutions existed. At this decision point for performance marketers, the gap between those numbers reflects a strong demand signal for a new, confidence-inspiring channel. What’s missing in this moment is the infrastructure to make open web advertising investment easier. Marketers need the same automated, goal-based buying capabilities they can currently get from PMax and Advantage+, but for the open web. ## How Agentic AI Changes the ROI Calculus for the Open Web It’s clear that there are lots of potential opportunities for the performance marketers and advertisers working toward incremental reach and new audiences outside of search and social. With agentic AI proving its value in popular search and social platforms, the open web is next up to put performance budget to work in new ways. The proof is in the 81% of respondents who say they’d increase open web investment if they had automated, AI-powered campaign solutions to use. “Automated” is key here — the ROI calculation changes when AI-driven automation is in the picture for performance marketers. Survey respondents mentioned a few barriers to open web investment: too many vendors (74%); lack of unified attribution (71%); brand safety concerns (54%), and resource constraints (42%). Those concerns are the ones that a well-designed agentic system can reduce, mitigate, or remove entirely. A single interface, unified reporting, automated inventory controls, and autonomous execution, all within a single, AI-powered platform, can address the specific hesitations that lead to underinvestment in the open web. ## The Financial Argument Is Settled, So the Real Question Is Execution The takeaways for curious, data-driven performance marketers and leaders are clear from the survey: first, agentic AI has proved ROI value in two channels (Advantage+ and PMax), based on 76% of survey responses. It’s driven the specific performance results that marketers need — increased CPA and ROAS optimization. That’s a refreshing data point in the face of a lot of uncertainty around how AI can actually drive bottom-line business results. Second, based on survey responses, there is both strategic urgency and budget intent to expand this agentic AI model to the open web to grow performance marketing results. The missing variable is the infrastructure to convert intent into investment. Performance marketing leaders can rest assured the model exists, so the only question now is whether they have the operational capability to apply it. --- ### Black Box AI in Advertising: Risks, Walled Gardens, and Open Web Solutions URL: https://www.taboola.com/marketing-hub/black-box-ai/ Last Modified: 2026-06-10 09:01:08 You hand a platform your budget, creative assets, and conversion goals. A few days later, you get the results back. The dashboard shows everything is working, conversions are up, and cost per click (CPC) looks good. But, if you stop and ask how — which placements drove conversions, which audiences were targeted, and why the algorithm made one bid instead of another — you don’t get a clear answer. That’s black box AI in advertising, and it’s the defining tension of modern digital media buying. This article breaks down exactly what black box AI is, how platforms like Google Performance Max (PMax) and Meta Advantage+ use it, the real risks it poses to your campaigns and brand, and why the open web offers a more transparent path forward. ## What Is Black Box AI? Black box AI is any artificial intelligence system whose inputs and operations aren’t visible to the user or another interested party. The term “black box” comes from engineering: it describes a system you can observe from the outside but can’t inspect inside. In this case, you can see what goes in (your budget, creative, audience signals) and what comes out (clicks, conversions, return on ad spend ), but the decision-making in between stays completely hidden. In an advertising context, this means the AI is buying media, setting bids, choosing placements, and adjusting targeting, all without showing you its reasoning. Most black box AI systems are powered by deep learning, which is part of why they’re opaque. These systems are made up of layers of mathematical formulas and millions or even billions of connections that work together to answer queries or solve problems — in mysterious ways. The sheer scale of these neural networks makes them difficult to interpret, even for the engineers who build them. On the other side of the coin is explainable AI marketing, sometimes called “white box” AI. These are systems designed so that humans can understand, audit, and verify how decisions are made. In advertising, this looks like log-level data, placement-level reports, and transparent bidding logic — things you can actually act on. ## How Walled Gardens Weaponize Black Box AI (PMax & Advantage+) Google and Meta have built their automation products around the same core promise: You provide your goals and budget, and their AI takes care of everything else. It sounds great on paper, but you do give up some control in the process. ### Google Performance Max When Google launched PMax in 2021, it consolidated Search, YouTube, Gmail, Display, Discover, Maps, and Shopping into a single campaign type. For years, advertisers had no idea whether their budgets were being spent on high-quality search traffic or low-converting display placements. Advertisers also didn’t know what formats were used, or which specific videos or product pages in a shopping campaign were performing well or poorly. This lack of channel visibility created an impossible optimization scenario where advertisers couldn’t determine which elements of their campaigns were succeeding or failing. Google has made some transparency concessions under sustained industry pressure. In 2025, it rolled out channel-level reporting, search terms visibility, and campaign-level negative keywords. But, the fundamental architecture hasn’t changed, and while advertisers could finally see where their ads appeared, they quickly realized that visibility alone offers limited value without real control. The strategic question remains: Does adding transparency features make PMax as controllable as standard campaigns, or just somewhat less opaque? ### Meta Advantage+ Meta’s Advantage+ follows the same playbook. Marketers provide inputs such as goals, budgets, and product feeds, and Advantage+ takes it from there. While brands appreciated performance improvements, the lack of transparency mirrored the struggles with Google’s PMax. Advertisers have little visibility into how budgets are allocated or how audiences are targeted. This makes campaign optimization nearly impossible, as decisions are made inside that black box. Despite advertiser concerns over control and visibility, adoption of Meta Advantage+ continues to rise. The problem is that without transparency, you can’t tell whether results are coming from your best audiences or from brand-search arbitrage that would have converted anyway. There’s no way to know if the reduced costs are ultimately cost-effective without transparency. That’s the core logical problem with black box platforms: they’re asking you to trust outcomes you can’t verify, from a system that profits from your continued spend. ## The Hidden Risks of Algorithmic Opacity for Advertisers The risks that come with surrendering visibility are also real, and they compound until something breaks. ### Ad Spend Waste You Can’t Diagnose When a campaign underperforms in a transparent system, you can trace the problem: it could be the wrong audience segment, poor placement, or a low bid on a high-converting keyword. Black box systems strip out that diagnostic layer entirely. This challenge is costing businesses not only in a lack of ad spend efficiency and diminishing returns, but also vulnerability to ad fraud. When performance drops in a PMax or Advantage+ campaign, there’s no clear place to look. You can adjust inputs — change creative, revise your goal settings, tweak your audience signals — but you’re working in the dark. The algorithm learns, but you don’t. ### Brand Safety Exposure Black box AI doesn’t just decide who sees your ads, it decides where your ads appear. Without placement-level transparency, your brand can end up alongside content you’d never have chosen manually. Two thirds (65%) of marketing and advertising decision-makers worldwide worry about the suitability of ad placements on social platforms, per DoubleVerify’s 2025 Global Insights report. Over 70% of marketers have encountered an AI-related incident in their advertising efforts, including hallucinations, bias, or off-brand content. The consequences were significant: 40% had to pause or pull ads, over a third dealt with brand damage or PR issues, and nearly 30% had to conduct internal audits. On walled garden platforms, your ability to audit placements after the fact is limited. There’s no full log of every site, video, or inventory unit your ad appeared against. You’re trusting the platform’s brand safety tools to catch problems you can’t see yourself. Realize takes a different approach. When you run a campaign on the open web, you can see every site on which your content is shown, and bring in third-party verification tools like Integral Ad Science, MOAT, and DoubleVerify to run brand safety reports independently. ### Hidden AI Bias Organizations that deploy black box AI can face backlash or financial loss if the AI behaves unexpectedly. Without explainability, AI errors can escalate into major crises before they’re caught. In advertising specifically, hidden AI bias can show up in a few ways. One is systematic underdelivery to certain demographic groups. Another is over-indexing on audiences that are cheaper to reach, rather than those most likely to convert. You also see systems optimizing for vanity metrics that don’t translate into real business outcomes. Because the logic is opaque, these patterns can persist for months before anyone identifies them. ### The Cross-Channel Intelligence Gap Perhaps the most underappreciated cost of black box platforms is the data they withhold from you. Every conversion that runs through Google’s or Meta’s ecosystem feeds their algorithm, not yours. You don’t own the insight, and you can’t port the learning to other channels. Walled gardens centralize data, media buying, and measurement within a single platform. When you move budget out of a walled garden, even temporarily, you lose continuity. The algorithmic learning you paid for stays inside the platform, creating a subtle form of lock-in. The longer you stay, the more valuable the platform’s model becomes for your account, and the more you risk losing if you diversify. ## Escaping the Black Box: The Open Web Advertising Solution The open web operates on fundamentally different principles. Instead of one platform controlling inventory, measurement, and optimization behind closed doors, programmatic buying on the open web runs through interoperable systems where advertisers can see exactly where their money is going. In the open web programmatic advertising context, the open internet operates through interoperable technologies such as demand-side platforms (DSPs), supply-side platforms (SSPs), and ad exchanges. These systems work together to facilitate programmatic buying, allowing advertisers to access inventory from multiple sources more flexibly and transparently. In practice, this means: - Placement-level visibility. You know which publisher sites, apps, and contexts your ads ran against. You can exclude what doesn’t work, double down on what does, and build a real understanding of your best-performing inventory. - Log-level data ownership. Unlike walled gardens, open web campaigns can give you access to impression-level data. That data belongs to you, and you can use it to inform targeting decisions across every channel you run, not just one. - Bidding transparency. You can see how bids are set, where you’re competitive, and where you’re leaving opportunity on the table. That feedback loop is what makes optimization meaningful. - Explainable ROI. When performance goes up or down, you can trace why. That’s the foundation of a media strategy that improves over time. The Realize platform is built for exactly this kind of transparent performance advertising on the open web. Transparent, actionable data lets advertisers get a 360-degree look at what’s working best, and where they can improve to see their best campaign performance. Advertisers can see every publisher site where their content runs, control placement settings, and own the performance data their campaigns generate — none of which is available on a fully automated walled garden product. ## Key Takeaways Black box AI in advertising offers automation at the cost of understanding. For some campaigns with clear goals and enough conversion volume to feed the algorithm, fully automated products can perform well, but the risks of flying blind are real, and they’re cumulative. When you can’t see where your ads run, you can’t protect your brand. When you can’t trace performance drops, you can’t fix them. When your data stays inside the platform, you can’t build intelligence that compounds across channels. The balanced approach that most effective advertisers use is to treat walled-garden automation as one tool among several, not as a complete media strategy. Open web advertising through transparent programmatic channels gives you the visibility, data ownership, and bidding control that black-box platforms by design withhold. ## Frequently Asked Questions (FAQs) ### What does “black box” mean in AI? A black box AI is a system in which the inputs and outputs are visible, but the internal decision-making process remains entirely hidden. Black box AI models arrive at conclusions or make decisions without explaining how they were reached. You can observe results, but you can’t audit the logic that produced them. ### Is Google Performance Max considered black box AI? Yes. Google’s Performance Max (PMax) architecture treats channel distribution as an algorithmic optimization problem rather than a strategic business decision. The system prioritizes algorithmic learning over advertiser preferences, treating channel budget allocation as a variable to optimize, rather than a strategic parameter to control. While Google added some reporting features in 2025, the underlying optimization logic remains opaque and non-overridable. ### Why is a lack of algorithmic transparency a risk for marketers? Without transparency, marketers can’t diagnose performance problems, verify audience targeting decisions, or audit where their ads are appearing. ### How does open web advertising solve the black-box problem? Open web programmatic advertising gives advertisers access to placement-level data, transparent bidding logic, and log-level reporting that walled garden platforms don’t provide. The open web offers granular targeting, transparent metrics, and first-party data ownership — ideal for brands building long-term, privacy-compliant strategies. Platforms like Realize provide programmatic advertising transparency, showing advertisers exactly where their ads run and providing the data they need to make informed optimization decisions, without black-box automation stripping away that visibility. --- ### PMax: Google’s Performance Max Campaigns Explained URL: https://www.taboola.com/marketing-hub/pmax/ Last Modified: 2026-05-23 17:59:45 For most of digital advertising’s history, running campaigns meant making dozens of decisions every day. Decisions like which keywords to target, which placements to buy, how much to bid, and which creative to show to which audience. Effective? Sometimes. Scalable? Not really. Google Performance Max (PMax) is a complete departure from that model. Rather than requiring advertisers to manage individual levers across separate channels, PMax consolidates everything — inventory, bidding, creative, and targeting — into a single artificial intelligence (AI)-driven campaign that operates across Google’s entire network simultaneously. For performance marketers tasked with scaling revenue efficiently, understanding PMax isn’t just useful: it’s the price of admission. If you’ve been asking what PMax is in Google Ads, the short answer is that it’s Google’s most ambitious attempt yet to automate campaign management from end to end. ## What Is a Performance Max (PMax) Campaign? Performance Max is a goal-based campaign type within Google Ads that gives advertisers access to all of Google’s ad inventory from a single campaign. That means one campaign can serve ads on Search, YouTube, Display, Gmail, Discover, Shopping, and Maps, with placements determined not by the advertiser, but by Google’s machine-learning engine based on real-time signals about user intent. What sets PMax apart from other campaigns is its agentic nature. Unlike traditional campaigns that require continuous human optimization, PMax functions as an autonomous system. You set the objective and provide the inputs, and the algorithm handles execution. It decides where to show your ads, in what format, to whom, and at what bid, continuously optimizing across all surfaces to hit your conversion goals. That’s what makes it an agentic marketing solution in the truest sense. PMax doesn’t just automate tasks within a defined channel, it acts as an intelligent agent, making strategic decisions across the entire funnel, as well as the entire Google ecosystem. ## How Does the PMax AI Engine Work? At its core, PMax runs on Google’s smart bidding algorithms, which use machine learning to process large volumes of real-time data, such as search queries, browsing behavior, device type, location, and time of day. It uses this information to determine the optimal bid for every impression opportunity. Advertisers provide four primary inputs to get the system moving: - Budget and conversion goals: You tell the system what you want to achieve, whether that’s maximizing conversions or maximizing conversion value, and what you’re willing to spend to get there. - Audience signals: You provide seed data such as customer lists, website visitors, or interest-based segments to give the algorithm a starting point for identifying high-value users. - Asset groups: You upload the raw materials the system needs to construct ads tailored to each channel and format, including headlines, descriptions, images, logos, and videos. - Conversion tracking data: The more accurate and value-rich your conversion data, the smarter the algorithm becomes over time. From there, your PMax bid strategy is handled entirely by the algorithm. Google’s AI automatically creates ads and tests which combinations of creative and placement drive the most conversions. The system continuously refines this process, shifting budget toward what’s working and away from what isn’t. One notable development is the integration of generative AI models like Gemini, which allows advertisers to generate and scale new assets faster than ever, experimenting with fresh headlines, descriptions, and images without starting from scratch each time. ## Performance Max vs. Traditional Google Ads: What’s the Difference? The clearest way to understand PMax is to contrast it with the campaign types it was designed to work alongside and, in some cases, replace. Of all the Google Ads automated campaigns available, PMax is the most comprehensive, and understanding how it differs from traditional campaigns is key to using it well. Traditional Google Ads campaigns are channel-specific and require hands-on management. PMax is cross-channel and largely self-directed. In a traditional setup, an advertiser might run separate Search, Display, Shopping, and YouTube campaigns, each with its own targeting parameters, bids, creatives, and budgets. Managing all of these simultaneously is complex and resource-intensive, and performance data rarely paints a complete cross-channel picture. PMax collapses that complexity into a unified structure governed by a single conversion objective. The trade-off is control. Traditional campaigns give advertisers granular visibility and the ability to adjust individual keywords, placements, and bids. PMax shifts most of those decisions to the algorithm, prioritizing efficiency. ### PMax vs. Standard Search Campaigns Search campaigns are the workhorse of Google Ads, built for precision. They target users at the exact moment of intent, based on the terms they type into the search bar. That focus is their strength — and their limitation. Standard Search campaigns offer: - Placement exclusively within Google Search results. - Full advertiser control over keyword match types, ad copy, and bids. - Text-only ad formats targeted to specific queries. One difference between PMax vs. Search campaigns is that PMax casts a much wider net. It reaches users across all of Google’s properties and makes its own decisions about format and placement, based on real-time intent signals. That breadth is powerful for expanding reach, but it means giving up the keyword-level precision that Search provides. The practical recommendation for most advertisers is to run both simultaneously: Search locks in your highest-value keywords, while PMax expands reach beyond what Search alone can capture. ### PMax vs. Standard Shopping For e-commerce advertisers, the most relevant comparison is with Smart Shopping, the campaign type PMax directly replaced. Google retired Smart Shopping in 2022 and migrated all advertisers over, so if you were running it before, PMax is already your new normal. With PMax, the product feed foundation is the same, but PMax adds the ability to serve ads across: - YouTube. - Gmail. - Discover. - Maps. - Search and Display. Where Smart Shopping was limited to Search and Display, PMax dramatically expands how many touchpoints your product catalog can reach, all from a single campaign. ## The Pros and Cons of Google PMax PMax offers genuine advantages for advertisers who are set up to use it well, but it also introduces friction points that can be frustrating if you’re used to having control over your campaigns. The advantages: - It consolidates cross-channel management significantly, freeing up time for higher-level strategy rather than day-to-day bid adjustments. - It opens up audience segments and placements that manual campaigns often overlook, giving you more coverage across the full funnel. - Google reports that advertisers who’ve adopted PMax see an average 18% increase in conversions at a similar cost-per-action compared to standard Shopping campaigns alone. - Because it learns continuously, AI-driven advertising performance typically improves over time as the algorithm accumulates more conversion data. The drawbacks: - Keyword-level transparency is limited. You can access search term reports, but not the detailed data available in standard Search campaigns. - Ad cannibalization is the real risk. Without careful campaign architecture, PMax can compete with your existing Search and Shopping campaigns for the same auctions. - The “black box” nature of its decision-making makes it harder to diagnose performance issues, or understand why the algorithm is making specific choices, which can complicate attribution analysis. For advertisers with the right setup, the efficiency gains are hard to ignore. The key is understanding upfront that you’re trading detailed control for scale, and building your campaigns accordingly. For advertisers who do aim for more control and visibility, there are alternatives. ## Winning PMax Strategies for Performance Advertisers Because you can’t manually adjust bids in a PMax campaign, the optimization lever shifts from bid management to input quality. The better your inputs, the better the algorithm performs. Here are some tips to help get the most from your efforts. ### Invest in Conversion Tracking Quality First PMax is only as smart as the data it learns from. Before launching, make sure your conversion tracking is accurate, comprehensive, and wherever possible, value-based. Assigning different values to different conversion types — a purchase vs. a newsletter signup, for example — allows the algorithm to optimize for outcomes that actually matter to your business. ### Build Rich, Diverse Asset Groups The system needs creative variety to test effectively across formats. Provide the full range of image sizes, multiple headline and description variations, and video. Campaigns without video assets default to auto-generated videos, which are rarely optimized for performance. ### Use Audience Signals Strategically When it comes to audience signals, PMax treats them as starting points rather than hard constraints. Upload your customer list, retargeting audiences, and high-intent in-market segments to give the algorithm a strong foundation. ### Segment Google Ads Asset Groups by Product Category or Audience Intent Rather than lumping all your products or services into a single asset group, create separate groups aligned with distinct categories. This helps the algorithm match creatives more precisely to relevant inventory and gives you cleaner performance data for each segment. ### Monitor for Cannibalization Run a brand-excluded campaign setup and regularly review search term reports to make sure PMax isn’t encroaching on your high-performing search keywords. Think of PMax optimization less like tuning a campaign and more like briefing a very capable colleague. The clearer and richer the information you provide on the front end, the better the results you can expect on the back end. ## Is Performance Max Right for Your Advertising Goals? PMax performs best when certain conditions are in place. It thrives on data, so the more conversion history your account has, the more effectively the algorithm can optimize. For this reason, it’s especially well-suited to e-commerce brands with established purchase funnels, SaaS companies scaling user acquisition at volume, and advertisers with clean customer relationship management system (CRM) data and value-based conversion tracking configured. The campaign type is less suited to advertisers operating on tight daily budgets. Smart bidding is the only option here, since manual bidding isn’t available, and the algorithm needs sufficient daily spend to generate useful data. Most advertisers set that threshold at $50 per day or higher. Without it, the learning phase drags on, performance stays volatile, and the campaign never fully matures. Similarly, advertisers who need strict control over every placement — whether for brand safety reasons, competitive considerations, or regulatory constraints — may find PMax’s opacity difficult to manage. The algorithm’s autonomy is a benefit for some and a liability for others. Knowing which camp you’re in before you launch will save a lot of frustration. It’s worth noting, too, that Google’s network, as broad as it is, still represents a walled garden. Advertisers looking to extend their reach beyond Google’s ecosystem can explore performance campaigns on the open web, which offer access to premium publisher inventory outside of search, YouTube, and Shopping. For brands already running PMax, adding an open-web strategy can fill in the gaps that even a cross-channel Google campaign can’t cover. ## Key Takeaways Google Performance Max represents a genuine shift in how omnichannel ad campaigns are managed. The advertiser’s job hasn’t disappeared, it’s just moved upstream — less time on bid adjustments and placement management, and more time on the conversion data, creative assets, and audience signals that drive AI behavior. Advertisers who make that mental shift and invest in high-quality inputs consistently see PMax outperform siloed, manually managed campaigns. Those who treat it like a set-it-and-forget-it tool are likely to be disappointed. ## Frequently Asked Questions (FAQs) ### What is the main difference between PMax and Search campaigns? Search campaigns run exclusively within Google’s search results and are governed by keyword targeting. Your ads appear when users search for terms you’ve explicitly bid on, giving you full control over match types, bids, and ad copy. PMax operates across all of Google’s properties simultaneously, using machine learning to automatically determine placements, formats, and bids. The two campaign types serve different purposes and work best when run in tandem. ### Does PMax replace my other Google Ads campaigns? No. PMax is built to complement existing campaigns, particularly standard keyword-based Search campaigns. Running both allows you to maintain precise control over your highest-value, specific-intent keywords while giving PMax room to expand reach across the broader network. Most advertisers should treat PMax as an addition to their campaign mix, not a replacement. ### How do you optimize a PMax campaign if it’s automated? Optimization with PMax happens at the input level rather than the execution level. Focus on the quality and accuracy of your conversion tracking data, the richness and variety of your creative assets, the precision of your audience signals, and the logical segmentation of your asset groups. Improvements in any of these areas give the algorithm better information to work with, which typically translates to better performance. ### Are Performance Max campaigns good for lead generation? PMax can be effective for lead generation at scale, but it requires the right infrastructure. Because it serves ads across the Display network, it has a higher potential for attracting low-quality or spam leads than Search-only campaigns. To mitigate this, strong CRM integration and value-based conversion tracking are essential. You need to be able to tell the algorithm not just that a lead came in, but what that lead is actually worth. Without that signal, the campaign may optimize for volume rather than lead quality. --- ### Understanding Native Advertising: How It Works, Types, Best Practices URL: https://www.taboola.com/marketing-hub/native-advertising/ Last Modified: 2026-05-23 18:43:47 Today, over 90% of users ignore traditional banner ads and 52% use active ad-blocking software. In other words, the traditional interruptive marketing model has lost its effectiveness, leaving marketers facing a conundrum: how to expand consumer awareness without triggering their defense mechanisms. The answer? Native advertising. With this advertising format, marketing teams design paid media to match the look, feel, and function of the platform on which it appears. It’s the open web equivalent of the seamless experience we’re accustomed to in social feeds, and now you can supercharge it with agentic intelligence. ## What Is Native Advertising, Really? If you immediately think native advertising equals sponsored content, you should expand your definition. This integrated ad format functions as a natural extension of the user’s content journey. Its core characteristics include: - Visual consistency. The ad uses the same fonts, layout, and image styles as the surrounding editorial content. - Functional alignment. The ad behaves like the rest of the site. If the site is a news feed, the ad becomes a story; if the site is a search engine, the ad is a result. - A value-first approach. Unlike display ads that demand an immediate Buy Now! action, native ads offer information, entertainment, or utility first. ## What Business Is Native Advertising Best For? The native advertising industry’s value was projected to hit $400 billion by 2025. The takeaway? It’s here to stay — and now it’s essential for: - B2B and high-consideration brands. It allows brands to explain complex return on investment (ROI) or technical specs like medical, legal, and financial, in a format users already trust. - Direct-to-consumer (D2C) and e-commerce. Brands selling aesthetic or lifestyle products (gardening tools, watercolor supplies, boutique fashion) can use visual storytelling that feels like a more relatable peer recommendation. - High-growth performance marketers. Agencies needing absolute return on ad spend (ROAS) use native to access premium demographics on sites like The Washington Post or Der Spiegel, where banner blindness is the highest. ## How Does Native Advertising Work? The mechanics of native advertising have evolved from static advertorials to agentic auctions. As third-party cookies continue their phaseout, native advertising has returned to its roots: contextual relevance. Placing an ad for hiking boots in an article about Best Trails in the American Midwest ensures that audiences see it as a helpful suggestion, not a creepy tracking attempt or irrelevant interruption. ### The Supply and Demand Chain - The publisher (supply): High-quality news and lifestyle sites provide slots within their article feeds or recommendation sections. - The advertiser (demand): Brands provide a creative portfolio (multiple headlines, images, and URLs). - The agentic engine: While standard demand-side platforms (DSPs) bid on users, agentic engines act as an autonomous agent, analyzing thousands of combinations of devices, time of day, and contextual signals to determine which specific creative yields the best outcome. ## Types of Native Ads There are six primary types of native advertising, each designed to align with specific user behaviors, like passive scrolling or high-intent searching. ### In-feed Content In-feed ads are the most common form of native advertising. They appear directly in the natural flow of content on a publisher’s site or social media platform. Think: a promoted post on LinkedIn or a sponsored article on The New York Times. - They match the typography, image aspect ratio, and layout of organic posts. - Because users are typically in discovery mode (scrolling for news or updates), they evaluate the ad using the same mental framework as editorial content. - Best use case: High-level brand storytelling and thought leadership. For example, a medical technology firm might publish an article about the future of artificial intelligence (AI) in surgery on a health news site. ### Recommendation Widgets These are the “you might like…” or “recommended for you” grids you find at the bottom or side of an article, powered by content discovery networks. - They typically look like a grid of thumbnails with catchy headlines. While they resemble internal site links, they include labels like “Around the web” or “Sponsored.” - Users engage with these widgets after finishing an article and looking for what’s next. Widgets capitalize on curiosity. - Best use case: Driving high-volume traffic to landing pages or listicles that nurture leads. ### Sponsored Search and Promoted Listings These ads appear at the top of search engine results (Google/Bing) or within e-commerce marketplaces (Amazon/Etsy). Unlike other native ads, a specific user query usually triggers them. - They’re identical to organic search results or product listings and distinguished only by a small “Ad” or “Sponsored” tag. - Users typically have high intent because they’re looking for a solution or product. The ad addresses an immediate problem. - Best use case: Direct response marketing and e-commerce sales. If someone searches for “best gardening tools for sore backs,” a promoted listing for an ergonomic shovel is the perfect native fit. ### In-map Content Integrated into navigation apps like Google Maps or Waze, these ads appear as branded pins or promoted search results based on a user’s geographic location. - A branded pin (like a logo) appears on the map as a user searches for “coffee near me.” It includes one-tap directions and store hours. - This strategy works for local, immediate intent. The user is physically near the business and needs a service now. - Best use case: Drive-to-store campaigns for retail, restaurants, and service providers (like a repair shop or car dealership). ### Branded Video Native video ads integrate into the content stream rather than appearing as forced interruptions like annoying mid-roll YouTube ads. - These videos appear within an article or feed and may start muted or require a user to click to begin, to avoid disrupting the reading process. - These users are willing to watch if the content provides entertainment or value. Its opt-in approach builds brand trust (instead of generating annoyance). - Best use case: Emotional brand building and product demos. ### In-game Narrative and Rewarded Video These ads, which are common in mobile gaming, are integrated into the game’s aesthetic or economy. The most popular format is rewarded video, where users choose to watch an ad in exchange for a life or currency. - You might see an in-game character wearing a branded shirt, or an ad break might pop up, offering you an optional way to unlock a new level or earn credits. - This strategy creates a positive association with the brand because users receive a tangible benefit (a reward) in exchange for their attention. - Best use case: App installs, high-frequency brand exposure, and reaching younger, tech-savvy demographics. Format Description Best for… In-feed content Ads that look exactly like the news or social posts surrounding them. Awareness and brand storytelling. Recommendation widgets “Recommended for You” blocks at the bottom of articles. Driving traffic and performance. Sponsored listings Promoted products on search results or e-commerce pages (Amazon, Google Search). Direct conversions. In-map content Ads within navigation apps like Google Maps that appear when searching for nearby locations. Local business and foot traffic. Branded video Click-to-watch or in-stream video that provides entertainment value. Engagement and high intent. In-game narrative Rewarded video or aesthetic-matched assets within mobile games. App installs and retention. ## Native Advertising: Pros/Cons ### Combatting Ad fatigue and Banner Blindness After years of digital exposure, is it any wonder that users have developed banner blindness — a cognitive filter that unconsciously allows them to ignore the top and side rails of a website where ads typically live? Native ads sit in the center of the page where users are already focused. Because they don’t look like traditional clutter, these ads bypass the mental filter. Research also shows that consumers look at native ads 53% more frequently than display ads (and are more likely to process them as content). ### Ad Blocker Resistance Third-party scripts serve up traditional display ads that browser extensions can easily identify and block. Native ads are designed to match the publisher’s CSS (styling) and are thus more often integrated more deeply into the site’s structure. While ethical disclosure is still required, the format is less likely to be stripped from the page, increasing the chance that intended recipients will see your message. ### Higher Click-through Rate (CTR) and Retention Banner ads often attract accidental or impulsive clicks. The click on a native ad is intentional. Native ads deliver up to 8.8x higher CTR than banners. User retention is three times higher. Users who clicked to learn or engage (rather than react to a flashing box) arrive on your landing page with a higher degree of pre-qualification. ### The Brand Halo Effect Consumers associate the quality of an ad with the quality of the environment in which it appears. If it appears as a partner content piece on The Wall Street Journal or National Geographic, your brand inherits those institutions’ authority. This halo effect builds trust faster than a standalone pop-up on a generic site. Pros Cons Better UX: Nondisruptive; fits the flow of a user’s journey. Content intensity: Requires high-quality writing and imagery. You can’t just slap a logo on it and hope for the best. Bypasses ad fatigue: Engagement remains high, even for tech-savvy audiences. The risk of deception: If not labeled clearly, users can feel tricked, which will damage brand trust. Higher intent: Clicks represent a genuine interest in the topic, leading to better leads. The complexity challenge: Manually optimizing headlines and images across 1000+ sites is impossible. Contextual resilience: Works well in a world without third-party cookies by focusing on page content. Harder to measure (at first): Focuses on post-click engagement, which requires more advanced tracking tools. ## Benefits of Native Advertising ### Authentic Engagement in a Natural Environment Native ads combine image and title units styled to match publisher content. They drive authentic engagement and consistent performance by blending ads seamlessly into the page experience. ### Full-funnel Versatility - Top (awareness): Use long-form sponsored articles to introduce a brand story. - Middle (consideration): Use interactive native video to build interest. - Bottom (conversion): Use native ads with embedded CTAs or in-search results to drive immediate action. ## Considerations of Native Advertising ### The Transparency Mandate Because native ads blend so well with organic content, there’s a fine line between integration and deception. If you don’t clearly label an ad (e.g., sponsored content), users may feel misled once they realize they’re engaging with paid material. Transparency is key. Label your native ads appropriately, otherwise, you risk losing trust that’s hard to recover. ### Mind Your Budgets Native marketing ads up, especially when you compete against prominent brands with huge budgets, as auctioned placements go to the highest bidder. Incorporate native marketing as part of a larger, diversified strategy instead of relying on it as the singular business driver. ### Content-heavy Strategy Native ads work because we build them from themes that drive good content: cohesive storytelling, audience awareness, and quality. You need more than a quick collection of banner ads or paid search spots. Commit to content creation — whether video, articles, or social placements — and give your team the resources and time to deliver quality work. ## How to Create a Native Ad Campaign in 6 Steps ### Step 1: Define your hitchable outcomes Don’t chase clicks. Define your objective: Is it a newsletter signup? A whitepaper download? A tire purchase? Platforms need a definitive goal to optimize effectively. ### Step 2: Consolidate for AI Modern agencies are moving away from hyper-granular ad sets. Consolidate your campaign structure. Give the algorithm enough budget and room to A/B test and find the best option across thousands of publishers. ### Step 3: Master the brand script You wouldn’t ask someone to marry you on your first date, right? Position your brand as the helpful partner. Tell a story that addresses the customer’s specific pain points (e.g., why 2026 is the year for ergonomic gardening) rather than attempting a hard sell. ### Step 4: Use the creative portfolio approach Feed the system a large volume of high-quality inputs. Use images of people, not logos. Keep your headline and image congruent because together, they can tell a logical, compelling story. ### Step 5: Champion a mobile-first user experience (UX) With over 50% of ad spend now on mobile, you need a flawless landing page. A slow-loading, desktop-style article will result in a 90% bounce rate. ### Step 6: Embrace continuous agentic optimization Native isn’t set-it-and-forget-it. Use agentic AI to automatically block underperforming supply paths and refresh visuals to avoid creative fatigue. ## How to Measure the Effectiveness of Your Native Ad Campaign Measuring native advertising’s effectiveness requires a shift from surface-level metrics to intent-based analytics. A click is just the story’s beginning: to really understand ROI requires determining how deeply the user engaged with content and whether that engagement led to a measurable business outcome. ### Click-through Rate (CTR) Definition: The ratio of users who click on your native ad to the number of total users who view it. - Why it matters: In native advertising, CTR is a resonance test. Because the ad competes with editorial headlines, a high CTR indicates that your headline and image combo provides perceived value to the audience. ### Engagement and Dwell Time Definition: The amount of time a user spends consuming your content after the initial click. - Why it matters: Native advertising is a content-first strategy. If a user clicks your ad but bounces within five seconds, they probably perceived it as clickbait. ### Scroll Depth Definition: A metric that tracks how far down a page a user scrolls (e.g., 25%, 50%, or 100% of the article). - Why it matters: This key performance indicator (KPI) is the best sign of content stickiness. If users consistently drop at the 50% mark, your branded script may be too sales-focused too soon. ### Conversion Rate (Cost Per Acquisition / Return on Ad Spend ) Definition: The percentage of users who take a specific hard action, like signing up for a newspaper, downloading a whitepaper, or making a purchase, after engaging with the content. - Why it matters: This is what to look for in the outcome-led model. Engagement is great, but performance advertisers need to see the bottom-line impact, too. ### Brand Lift and Sentiment Definition: Surveys or data analyses that measure changes in audience perception or brand awareness after exposure to a campaign. - Why it matters: Native advertising can benefit from a publisher’s halo effect. Users might not buy today, but they’re more likely to consider the brand later because they saw it in a trusted place, like The Wall Street Journal. ### Earned Media and Social Shares Definition: The frequency with which users share your native content on social platforms or link to it from other sites. - Why it matters: High-quality native content often goes viral within professional niches. If users share your sponsored LinkedIn article with their own networks, your paid reach effectively becomes earned reach, lowering your overall cost per impression. ## Alternatives to Native Ads ### Native vs. Standard Display Standard display (banners) relies on interruption. Native relies on integration. The brain often processes display ads as peripheral noise, but processes native ads as core content. While display often costs less on a cost per mille (CPM) basis, it can suffer from a quality gap. Native clicks represent higher intent and a lower cost per lead (CPL) in the long run. ### Native vs. Paid Search Search is reactive; it waits for a user to type a specific query. Native is proactive; it finds users who are consuming related content but haven’t yet searched for a solution. Use search to capture existing demand. Use native to create that demand by educating users before they register that they need you. ### Native on the Open Web vs. Walled Gardens Google PMax and Meta Advantage+ are powerful, but they prioritize their own inventory (think: YouTube and Instagram) and often lack transparency in reporting. The open web brings the same agentic, set-it-and-forget-it performance to premium, brand-safe publishers (like The Guardian or Forbes). It removes platform bias, directing your budget to where performance is strong (not where the platform wants to dump excess inventory). Feature Native advertising (open web) Standard display Walled gardens Paid search UX High: Non-disruptive; fits site aesthetic. Low: Often seen as clutter or spam. Medium: High-quality but repetitive. High: Provides direct answers to queries. Placement Control High: Direct code integration on premium sites. Variable: Can end up on long-tail, low-quality sites. Low: Black box automation; limited placement visibility. High: Specific to search engine results pages (SERPs). Targeting Logic Contextual and agentic: Moves with user intent. Audience-based: Relies heavily on (often being phased out) cookies. Platform-driven: Proprietary social/search data. Keyword-driven: High intent but limited scale. Ad Blocker Risk Low: Integrated into site creative specs (CSs). Very high: Frequently stripped by browsers. Medium: Mostly in-app (safe from blockers). Low: Usually bypasses blockers. Cost Model Performance-led: Focus on ROI/CPA. CMP-led: Focus on inexpensive impressions. Outcome-led: High performance, but high platform tax. Cost-per-click (CPC)-led: Capped by search volume. Scale Vast: Reaches the entire open web. Vast: But with low engagement quality. Limited: Restricted to the platform’s own apps. Limited: Capped by search volume. ## Key Takeaways Native advertising works because it treats the customer as a reader first and a buyer second. It brings the intelligence of social buying to the world’s best publishers. Its adaptive learning, with continuous optimization, identifies new supply paths and blocks underperformers. ## Frequently Asked Questions (FAQs) ### What are some of the best examples of native advertising ads? Klarna (Ad format: tailored carousel placements) This leader in the global fintech space monetizes its user interface by integrating sponsored offers into its home screen carousel. These placements use first-party data to personalize promotions to individual users. Visually, these ads are nearly indistinguishable from organic deals, except for a subtle ad identifier. Spotify (Ad format: immersive multimedia takeovers) Spotify leverages its core strength — curated discovery — to build in-depth native experiences. A great example was its collaboration with Netflix for Stranger Things, which allowed fans to activate a themed interface and access character-specific playlists. Bed Bath & Beyond (Ad format: integrated newsletter promotions) To monetize its large email subscriber database, Bed Bath & Beyond incorporates vendor promotions into its weekly newsletters. It uses direct-sold placements that mirror the aesthetic of its product highlights, rather than relying on generic third-party ad networks. ### What are some of the best practices for native ads? To maximize the benefits of native advertising, prioritize strategic partnerships with experienced publishers and technology providers to ensure that your content meets high editorial standards. Success starts with clear KPIs like brand awareness or conversions. Your brand “script” should focus on storytelling and solving customer pain points rather than a hard sales pitch. Lean into contextual targeting by placing ads in relevant editorial environments and making sure they align with the reader’s current intent. Your content should be mobile-optimized for fast loading, and leverage A/B testing and regular content refreshes to mitigate potential ad fatigue. ### What are the best platforms or networks for native ads? Realize+ is a premier choice for agentic (currently in Beta), outcome-based buying on the open web. Other reputable networks include LinkedIn (for professional B2B reach), Outbrain, and Nativo. ### Is there a difference between native and banner ads? Banner ads live in a page’s gutters and are visually separated from content, like a billboard on a highway. Native ads integrate into the content itself, like a product placement in a movie. By matching a site’s fonts and layout, native ads bypass banner blindness and attract higher engagement. --- ### Best Performance Platforms for Native Advertising in 2026 URL: https://www.taboola.com/marketing-hub/native-advertising-platform/ Last Modified: 2026-05-23 19:06:28 Running ads on social media is no longer enough. Rising costs per mille (CPMs), tightening audience-targeting controls, and algorithmic volatility have pushed performance marketers to look beyond the walled gardens of Meta and Google. Native advertising platforms on the open web offer something different. They exist in premium publisher environments where audiences are actively researching, more receptive, and less saturated with ads. The challenge is choosing the right platform. Some excel at scale, some at optimization depth, and some at specific verticals. Getting it wrong means budget burned on traffic that won’t convert. This guide breaks down the leading native advertising platforms for performance campaigns in 2026 — what each does well, what it doesn’t, and who it’s best suited for. ## Best Native Advertising Platforms Compared Platform Why It’s Essential Core Use Cases and Features Best for (Performance Advertisers) Pricing Model (Indicative) 1. Realize Performance-first platform built around outcome-based optimization. Artificial intelligence (AI)-driven bidding, conversion tracking, audience modeling, and automation for performance campaigns. Cost per acquisition (CPA)-focused advertisers, e-commerce, lead generation , return on ad spend (ROAS) optimization. Performance-based; cost per click (CPC) for native, CPM for programmatic. 2. Teads Premium open internet platform combining Outbrain’s performance DNA with Teads’ video and branding expertise. In-feed native ads, outstream video, contextual targeting, audience segmentation, conversion optimization. Lead generation, finance, high-lifetime value (LTV) verticals, full-funnel campaigns. CPC/CPM (CPC typically 20 cents-50 cents). 3. MGID Global scale with a performance-first approach. Native widgets, contextual and behavioral targeting, advanced optimization tools. International campaigns, affiliates, and aggressive scaling. CPC/CPM (CPC about 5 cents-25 cents.) 4. Revcontent High engagement inventory with strict publisher quality control. Native placements, granular targeting, fast-loading widgets, A/B testing. Arbitrage, lead generation, high click-through rate (CTR) campaigns. CPC (auction-based, varies by vertical). 5. Life360 Ads (formerly Nativo) Strong focus on true native storytelling within premium publisher environments. In-feed native ads, branded content distribution, and contextual targeting. Brands balancing high-quality native with performance goals. CPM (often premium pricing). 6. Meta Audience Network Extends Meta demand beyond its owned apps into third-party inventory. Native, banner, interstitial formats, strong audience targeting via Meta data. Retargeting, app installs, scalable audience extension. CPC/CPM/CPA (auction-based). 7. Google AdSense One of the largest contextual ad networks with native ad formats. Native ads, contextual targeting, automated placements across publisher sites. Broad reach, contextual acquisition, long-tail scale. CPC (revenue share model, auction-based). 8. Yahoo Native Access to Yahoo’s owned-and-operated ecosystem and strong first-party data. Native plus display plus search integration, premium placements, audience insights. Brand-safe performance and hybrid campaigns. CPC/CPM (auction-based). 9. AdPushup Revenue optimization platform for publishers with native ad integrations. A/B testing, layout optimization, header bidding, native ad optimization. Publishers optimizing yield (less direct advertiser use). Revenue share/SaaS-based. ## The 9 Best Native Advertising Platforms For Performance Campaigns ### 1. Realize Why it’s essential: Realize is an AI-powered native advertising and performance platform designed to help marketers scale their customer acquisition beyond the confines of social media walled gardens. By leveraging a proprietary predictive engine, the platform identifies high-intent users across a massive network of premium news, lifestyle, and medical publisher sites, matching them with dynamically optimized ads in real time. Performance marketers use Realize as an automated command center to manage the entire campaign lifecycle, from generating AI-driven creatives to executing complex bidding strategies. The platform is built to deliver high-quality traffic and sustained performance stability, using machine learning algorithms that continuously ingest conversion data to refine targeting and minimize wasted ad spend. Showcased features: - GenAI AdMaker: This tool uses generative AI to instantly create and optimize high-quality static and motion ad variants, tailored to specific performance objectives. - Maximize Conversions: An automated bidding strategy that uses real-time signals to adjust cost-per-click, ensuring optimal scale and performance within a defined budget. - Predictive Audiences: This AI-driven feature identifies untapped, high-converting customer segments by mirroring the behaviors of your existing pixel or conversion data. - Social Importer: Marketers can quickly repurpose top-performing Facebook and Instagram posts into native display ads, maintaining brand consistency across the open web. - SpendGuard: A suite of automated safeguards powered by predictive algorithms that minimizes wasted spend by automatically de-prioritizing underperforming sites and creatives. - Abby (AI Assistant): A generative AI performance expert that automates campaign onboarding, troubleshoots creative rejections, and provides real-time optimization advice. Best for: Realize is particularly effective for high-growth D2C and high-consideration brands that need to reach educated, high-intent audiences during their informational research phase. It’s an ideal fit for teams that may lack a massive internal data science department, but want to leverage enterprise-grade AI to automate the heavy lifting of creative production, technical troubleshooting, and 24/7 bidding optimization. Pricing model: Performance-based model; campaigns billed on CPC basis, or CPM for programmatic. Pros: - Data-Driven Creative Speed: The GenAI AdMaker allows for rapid production of performance-optimized assets, effectively eliminating creative fatigue bottlenecks. - Automated Budget Protection: Features like SpendGuard and Custom Rules act as a 24/7 safety net, automatically pausing underperforming segments to protect ROI. - Predictive Performance: The platform’s ability to build predictive audiences from simple conversion signals allows for efficient scaling without the need for complex manual segmenting. Cons: - Baseline Data Requirements: The platform’s most powerful AI features, such as the Performance Simulator, require a consistent threshold of historical conversion data to reach peak efficiency. - Technical Implementation Depth: Achieving full-funnel visibility and maximum algorithm training often requires a more technical Server-to-Server (S2S) tracking setup, compared to basic pixels. - Regional Feature Availability: Some advanced targeting tools, such as Search Keyword and Mail Domain Targeting, are currently limited to specific global markets. ### 2. Teads Why it’s essential: Teads is one of the largest advertising platforms on the premium open web, with direct partnerships across more than 10,000 publishers and a reach of 2.2 billion consumers globally. For performance advertisers, the merger combines Outbrain’s content discovery and conversion-optimization heritage with Teads’ premium video inventory and brand-safety positioning. Where Teads differentiates itself is in the quality of its publisher relationships: CNN, The Washington Post, and similar editorial environments where contextual targeting reaches high-income, engaged audiences. Its AI-driven Smartlogic optimization layer continuously adjusts bids and placements based on engagement signals, making it well-suited for lead generation campaigns in finance, insurance, healthcare, and other high-LTV verticals. Showcased features: - In-feed native and outstream video formats: Seamless placements within premium editorial content, across article feeds, and in-article video environments. - Smartlogic AI bidding: Real-time optimization across placements and audiences based on predicted conversion probability. - Contextual targeting: Content-matched placements without third-party cookie reliance — particularly relevant following the deprecation of Privacy Sandbox technologies in late 2025. - Smart Bid and Maximize Conversions modes: Automated bidding for CPA optimization once campaigns accumulate sufficient conversion data. Best for:  Lead-generation advertisers in regulated or high-consideration verticals, such as finance, healthcare, and insurance, that need premium publisher brand safety alongside performance optimization. Also suited for full-funnel campaigns that blend awareness-stage video with lower-funnel conversion objectives. Pricing model: Teads’ pricing is auction-based, using CPC and CPM. CPC typically ranges from 20 cents to 50 cents for campaigns in the United States, with variation by vertical and targeting depth. Pros: - Premium publisher inventory with strong brand safety controls. - Combined branding and performance capabilities across the full funnel. - Strong reach into the 55+ demographic, which is relevant for Medicare, financial products, and insurance. Cons: - CPCs run higher than open-web alternatives like MGID or Revcontent. - The platform’s premium positioning makes it less suited for aggressive direct-response creative strategies that push against editorial standards. - The Outbrain/Teads integration is relatively recent; some advertisers may encounter ongoing product consolidation. ### 3. MGID Why it’s essential: MGID is a global native advertising platform with meaningful reach across the United States, Europe, Latin America, and Southeast Asia. Its “brandformance” positioning gives advertisers flexibility to run awareness-oriented native creatives while still optimizing for conversions and ROAS. The platform provides advertisers access to a broad publisher network with competitive auction pricing. Its contextual and behavioral targeting draws on first-party publisher signals, rather than third-party cookie dependencies. Best for: Performance marketers running international campaigns or extending reach beyond premium publishers in the United States. Particularly suited to affiliate marketers, lead generation in non-regulated verticals, and advertisers targeting emerging markets with competitive CPCs. Pricing model: CPC and CPM auction-based. CPCs typically range from $0.05 to $0.25. Pros: - Relatively low CPCs among major native platforms, allowing more volume for budget testing. - Strong international reach for global campaigns. - Contextual targeting is well-developed and cookie-independent. Cons: - Publisher quality is more mixed than premium networks; site-level brand safety controls require close monitoring. - Optimization depth is less sophisticated than AI-first platforms, requiring more manual campaign management. ### 4. Revcontent Why it’s essential: Revcontent is a performance-oriented native network focused on high engagement rates and publisher quality control. It occupies a distinct position in the native landscape: more selective about publisher partners than open-web networks, but more accessible and performance-friendly than premium-only platforms. Revcontent’s granular site-level targeting and A/B testing capabilities support the rapid creative iteration that performance campaigns depend on. Its fast-loading widget formats reduce page-speed penalties, which matter for downstream conversion rate. Best for: Direct-response advertising, content arbitrage, and lead-generation campaigns that require high volume at competitive CPCs. Works well for advertisers with strong creative testing capabilities who want to identify winning placements and scale quickly. Pricing model: CPC auction-based; rates vary by vertical and placement quality. Pros: - Strong publisher quality control relative to its pricing tier. - Fast-loading ad formats support better post-click performance. - Granular site-level controls give meaningful optimization levers. Cons: - Smaller reach than Realize, Teads, or MGID limits scale for broad campaigns. - Less sophisticated AI bidding than larger platforms. - Requires active management; automation capabilities are more limited. ### 5. Life360 Ads Why it’s essential: Life360 Ads (formerly Nativo) takes a different approach to native advertising than pure performance networks. Rather than sponsored widgets sitting alongside editorial content, Life360 Ads places ads directly within publisher article feeds in a format that reads like editorial content. This drives higher attention and engagement metrics than traditional native units. Life360 Ads reaches more than 228 million monthly unique users across 7,000+ publishers, including Time and MotorTrend, with ad formats spanning native article, native display, native video, and stories. Best for: Brands in high-consideration categories like automotive, financial services, and healthcare that need performance results without sacrificing creative integrity or brand safety. Also relevant for content marketers running sponsored article strategies. Pricing model: CPM-based, typically at premium rates. Pros: - True native integration drives higher attention and engagement metrics. - Premium publisher relationships support strong brand safety. - Effective for brands with strong content assets. Cons: - CPM pricing runs higher than most native alternatives. - Smaller reach limits top-of-funnel scale. - Less suited for aggressive direct-response campaigns. ### 6. Meta Audience Network Why it’s essential: Meta Audience Network extends Meta’s advertiser demand outside its owned-and-operated properties into third-party apps and websites. For advertisers already running Meta campaigns, the platform provides incremental reach using the same audience data and targeting infrastructure — custom audiences, lookalike audiences, interest-based targeting — at lower costs than core Facebook and Instagram inventory. Best for: Advertisers already running Meta campaigns who want incremental reach beyond owned-and-operated inventory. Particularly effective for retargeting, app installs, and campaigns where Meta’s audience data is the primary targeting mechanism. Pricing model: CPC, CPM, and CPA auction-based pricing within Meta Ads Manager. Pros: - Seamless integration with existing Meta campaigns — there’s no new platform to learn. - Access to Meta’s first-party audience data. - Natural starting point for extending high-performing Meta creatives into third-party inventory. Cons: - Limited transparency into where ads appear within the network. - Less publisher quality control than dedicated native platforms. - Functions as an extension of Meta, not a standalone native strategy. ### 7. Google AdSense Why it’s essential: Google AdSense is the largest contextual advertising network, with native ad formats that reach publishers across the web at scale. Advertisers access this inventory through Google Display Network campaigns, where native formats are available alongside standard display. The strength is reach and contextual precision, as Google’s crawling of publisher content enables placement matching at a scale no other network can match. Best for: Advertisers seeking broad reach across long-tail publisher inventory with strong contextual relevance. Works best as part of a broader Google Ads strategy, rather than as a standalone native campaign. Pricing model: CPC auction-based pricing through Google Ads; costs vary significantly by vertical and targeting parameters. Pros: - Unmatched scale and publisher reach. - Strong contextual targeting precision. - Easy management within Google Ads for advertisers already on the platform. Cons: - Less transparent publisher-level control than dedicated native platforms. - Not purpose-built for direct-response performance optimization. - Brand safety controls are less granular than premium native networks. ### 8. Yahoo Native Why it’s essential: Yahoo Native provides access to Yahoo’s owned-and-operated properties like Yahoo Finance, Yahoo News, Yahoo Sports, and AOL, plus a broader publisher network. Its differentiation is first-party audience data at scale. Yahoo’s registered user base enables demographic and interest targeting that doesn’t depend on third-party cookies. Best for: Advertisers targeting finance, news, and sports audiences in a premium, brand-safe environment. Also useful for hybrid campaigns that combine native display with Yahoo search-intent signals. Pricing model: CPC and CPM auction-based pricing. Pros: - First-party data targeting reduces cookie dependency. - Premium owned-and-operated inventory with strong brand safety. - Finance-focused inventory reaches high-income demographics. Cons: - More limited reach than Realize, Teads, or Google for broad campaign scale. - Optimization tools lag behind purpose-built performance platforms. ### 9. AdPushup Why it’s essential: AdPushup is primarily a publisher-side revenue optimization platform, not a direct advertiser tool. It’s included here because it sits at the intersection of native advertising and publisher monetization. Publishers using AdPushup optimize native ad placements alongside header bidding and layout testing, which affects ad quality and viewability for advertisers purchasing that inventory programmatically. Best for: Publishers optimizing ad revenue. Less directly applicable to performance advertisers, but worth understanding as context for the publisher-side quality optimization that shapes inventory across native networks. Pricing model: Revenue share and SaaS-based pricing for publishers. Pros: - Meaningful lift in publisher revenue (AdPushup reports an average 33% uplift), attracting higher-quality publisher partners to platforms carrying their inventory. - Sophisticated testing capabilities improve ad quality over time. Cons: - Not a direct advertiser tool. There’s no direct campaign management or media buying capability. - Only accessible to publishers generating $5,000+ in monthly ad revenue ## More About Native Advertising Platforms For Performance Campaigns ### What Is a Native Advertising Platform? A native advertising platform is a technology system that connects advertisers with publisher inventory and delivers ads that match the form and function of the surrounding editorial content. Unlike display banners that look like ads, native ads blend into the page and carry the visual and structural characteristics of the publisher’s own content. The defining characteristic is non-disruptiveness. A native ad in a news feed looks like an article. A native ad in a product grid looks like a product listing. This design reduces ad avoidance behavior and increases the likelihood that users engage with the content. For performance advertisers, native ad platforms also allow users to target users, optimize bids, and track conversions, turning publisher reach into measurable outcomes. ### How to Choose a Native Advertising Platform Choosing the right native advertising platform is a question of which platform’s infrastructure, inventory, and optimization logic aligns with your specific campaign goals. Here are some features to look out for: #### Publisher Network Quality and Direct Relationships Understand whether the platform has direct relationships with its publishers, or aggregates inventory through intermediaries. Direct publisher relationships give you greater visibility into where your ads appear, more control over placement exclusions, and stronger protection against fraudulent traffic. Platforms that operate on direct, first-party code across their publisher network give you the transparency that matters when you’re optimizing for conversion quality, not just volume. #### Targeting Capabilities and Data Sourcing The method a platform uses to target audiences determines the precision and stability of your campaigns. Contextual targeting, for example, which matches ads to page content rather than user identity, is increasingly important as third-party cookie signals erode. Understand what data each platform relies on and how it holds up in a cookieless environment. #### Comprehensive Format Portfolio Advertisers and agencies today expect a platform that covers the full funnel, not one that handles a single format. Look for platforms that support native, display, video, and programmatic formats from a single interface. A broader format portfolio means you can run unified, full-funnel campaigns without fragmenting your measurement or creative workflows across multiple platforms. #### Bidding and Optimization Sophistication Evaluate which automated bidding options are available. Examples include smart CPA bidding, maximizing conversions, target ROAS, and similar modes. These require sufficient conversion data to function well, so ask about minimum volume thresholds. Platforms with more advanced AI bidding reduce the manual optimization burden and sustain performance at scale. #### Creative Support and Tools Native creative fatigue is real. Platforms that offer AI-powered creative generation, social import tools, or built-in A/B testing frameworks reduce the burden on in-house creative teams and accelerate iteration cycles. #### Tracking and Measurement Depth Evaluate whether the platform supports pixel-based tracking, server-to-server (S2S) conversion events, and integration with your attribution tools. Platforms with stronger S2S infrastructure tend to produce more accurate optimization signals, especially for campaigns where post-click conversion paths are complex. See Realize’s campaign reporting capabilities for a sense of what best-in-class transparency looks like. ### What Is the Average Cost of Native Advertising Platforms? Native advertising costs vary by platform, vertical, and targeting parameters. Understanding both the pricing model and the realistic budget required is essential before committing. Common pricing models: - CPC (cost per click): You pay each time a user clicks your ad. Cost per click is the most common model for native discovery platforms, and the most practical for performance advertisers, because you pay for engagement, not just exposure. - CPM (cost per mille): You pay per 1,000 impressions. This is more common on premium platforms like Life360 Ads or programmatic-oriented buys. CPM is better suited for awareness objectives where reach matters more than direct clicks. - CPA (cost per action): You pay when a specific conversion event occurs. Less common as a direct pricing model but available on some platforms through automated bidding. Most performance-oriented native platforms recommend a budget of $100–$500 per day minimum for self-serve campaigns, as lower budgets limit the volume of conversion data the platform’s algorithm needs to optimize effectively. Start with a test budget, validate the conversion economics, and scale once the unit economics make sense. ### Best Native Ad Platforms for Small Businesses Small businesses running native advertising face a specific constraint in that they have limited creative resources and budget, which makes it harder to run the volume of tests that native platforms reward. Realize is worth considering even for smaller advertisers, particularly those with a clear conversion goal and at least some initial conversion data to help the algorithm learn. Its AI tools reduce the creative and optimization workload, which helps teams without dedicated media buyers. MGID is often the most accessible entry point for businesses new to native advertising. Its lower CPCs allow more test volume on limited budgets, and its self-serve interface doesn’t require significant technical expertise to operate. Meta Audience Network is a natural starting point for small businesses already running Meta campaigns. It extends existing creatives and audience targeting without requiring new platform setup or additional creative production. Regardless of platform, two principles apply: Invest in clear conversion tracking before spending anything on traffic, and start with a test budget large enough to generate statistically meaningful data, typically 50–100 conversion events per ad variant, before drawing optimization conclusions. ### Realize vs. Outbrain Outbrain no longer exists as a standalone platform. In February 2025, Outbrain completed its $900 million acquisition of Teads and subsequently rebranded the combined company as Teads in June 2025. Realize and Teads are both premium open-web native platforms with AI-driven optimization, but they serve different primary use cases. Realize is purpose-built for performance. The product architecture, AI tools (Maximize Conversions, Predictive Audiences, SpendGuard), and campaign management features are designed for advertisers optimizing toward CPA, ROAS, and conversion volume. It’s strongest for direct-response advertisers in e-commerce, lead generation, and D2C. Teads brings together Outbrain’s performance DNA with Teads’ premium video and branding heritage. It’s better positioned for full-funnel campaigns that blend awareness-stage video with lower-funnel conversion objectives, and for advertisers for whom brand safety in premium editorial environments is a core requirement. For pure performance campaigns, Realize’s conversion-first infrastructure and AI tools make it the stronger choice. For campaigns that need to balance branding and performance — particularly in finance, healthcare, or high-consideration consumer categories — Teads’ full-funnel positioning and premium inventory offer a compelling alternative. ## Key Takeaways Native advertising on the open web offers a meaningful alternative to social platforms, particularly as audience targeting precision within walled gardens continues to be constrained. Platform selection should start with conversion goals, not platform features — remember, the best platform is the one whose optimization logic aligns with your specific CPA, ROAS, or lead-generation objective. Publisher network transparency and direct relationships also matter, as knowing where your ads appear gives you better optimization leverage and stronger protection against low-quality inventory. Finally, your test budgets should be large enough to generate meaningful conversion data before drawing optimization conclusions — underfunding the test phase leads to premature decisions. ## Frequently Asked Questions ### How to start with native advertising (step guide) - Define your conversion event. Before spending anything, decide exactly what action you’re optimizing for, whether that’s a lead form submission, a purchase, or a subscription signup. Set up conversion tracking (pixel or S2S) before your campaign goes live. - Choose a platform that fits your vertical and budget. For most performance advertisers, Realize is a practical first choice. - Create three to five ad variations. Native ads depend on thumbnail images and headlines. Test different approaches — problem-focused vs. curiosity-driven, lifestyle imagery vs. product-focused — to find what resonates. - Launch with a test budget. Aim for $500–$1,000 over your initial test window. You need enough conversion volume (typically 50+ events) for the platform’s algorithm to learn effectively. - Analyze by placement and creative. Review site-level and creative-level performance. Pause underperforming placements and creatives, then reallocate budget to what’s working. - Scale what converts. Once you’ve identified winning creative and placement combinations, increase budget gradually — typically in 20–30% increments, rather than sudden jumps that destabilize algorithm optimization. ### What is the difference between native ad networks and programmatic platforms? Native ad networks are closed or semi-closed ecosystems where the platform owns or directly controls publisher relationships. Advertisers buy inventory through the platform’s own interface, using its targeting and bidding infrastructure. Realize, MGID, and Revcontent are examples. Programmatic platforms are open-auction environments where buyers access inventory from multiple publishers and ad networks through a demand-side platform (DSP), using standard IAB protocols. Native ad formats are available programmatically, but the buying and optimization layer sits in the DSP rather than with any single network. ### Which platforms are best for scaling native campaigns globally? - Realize operates a network reaching 1.4 billion unique users per month through premium publishers across multiple regions, with AI optimization tools that work across markets. - MGID is the most accessible platform for international campaigns, with competitive CPCs across Europe, Latin America, Southeast Asia, and the United States. - Teads has a global publisher footprint across 36 countries, with particular strength in Europe, the Middle East, and Africa (EMEA) markets inherited from the original Teads platform. - Meta Audience Network provides global reach for advertisers leveraging Meta’s audience infrastructure, though its publisher quality controls are less transparent than those of dedicated native platforms. For truly global campaigns, the practical approach is to test platforms by region rather than assuming one platform delivers uniform performance everywhere. CPCs, audience quality, and optimization dynamics vary meaningfully across markets. What works at scale in the United States may require a different platform or strategy in Southeast Asia or Latin America. --- ### Google PMax Alternatives: 5 Best Options to Scale on the Open Web in 2026 URL: https://www.taboola.com/marketing-hub/google-pmax-alternatives/ Last Modified: 2026-05-28 12:47:51 Google Performance Max (or Google PMax) promised to handle everything: budget allocation, placements, bidding, and creative testing. Hand it the keys and let the algorithm drive! For a lot of advertisers, that trade-off worked well enough to stick with. But, PMax is a Google product and remains within Google’s ecosystem. Every dollar you spend runs through Google’s inventory at Google’s prices, while your audience spends more than half their digital time somewhere else entirely, such as on news sites, streaming platforms, financial publishers, and apps that have nothing to do with Google. In this guide, we break down the five best Google PMax alternatives. ## The 5 Best Google PMax Alternatives For Open Web Scaling These five platforms represent different approaches to open web performance advertising, from agentic AI solutions to full-service DSPs. Each one solves different parts of the PMax problem: the right choice depends on where your biggest constraint is, whether that’s control, reach, transparency, or access to first-party data-driven targeting. Platform Best for Special features Pricing 1. Realize+ (In Beta) Agentic open web performance at scale. Autonomous AI agents, Campaign Generator, Predictive Audiences, placement-level transparency. Performance-based model; campaigns billed on CPC basis, or CPM for programmatic. 2. Criteo Commerce Growth E-commerce customer acquisition and retention. Purchase-intent data from more than 22,000 commerce partners, dynamic creative optimization. CPC/CPM, custom. 3. The Trade Desk Enterprise programmatic with full control. Kokai AI, more than 500 data integrations, no inventory conflicts. About 15–20% of media spend. 4. StackAdapt Multi-channel programmatic for mid-market teams. Contextual AI targeting, accessible UI, no minimum spend. Custom. 5. Yahoo DSP Identity-resilient reach, CTV, premium streaming. ConnectID, Yahoo Backstage, no ad serving fees. Custom CPM/CPC. ### 1. Realize+ (In Beta) — Agentic Solution For the Open Web Realize+ is an agentic performance advertising platform, built to automate campaign buying and optimization at scale, but across the open web instead of within Google’s closed ecosystem (currently in Beta). Where PMax locks you into Google’s properties, Realize+ operates across a network of premium publishers, including TIME, Yahoo News, the Weather Channel, USA TODAY Network, and thousands of others. It’s designed for advertisers who want autonomous, AI-driven campaign management without ecosystem dependency. The platform’s agentic layer is what distinguishes it from conventional programmatic tools. Realize+ doesn’t require you to monitor performance and make manual adjustments: its AI advertising agents make buying decisions continuously, testing placements, adjusting bids, and reallocating budget. They do this with full visibility into reporting, so you know exactly what decisions were made and why. Features: - The Campaign Generator function automatically generates and optimizes campaign elements (campaigns, creatives, audiences) to fit your evolving strategy. - Realize+’s AI agents monitor campaign performance in real time and adjust bids automatically to hit your targets, automating optimization without surrendering visibility into where your budget goes. - Predictive Audiences uses first-party data signals to identify your best converters (some advertisers have grown conversions by up to 270% while holding CPAs stable). Realize+ continuously refines these audience models as campaigns run. - Realize+ operates across a curated network of vetted publisher sites, giving you blocklist controls at the placement level, a self-serve capability PMax doesn’t offer. The agentic system uses these constraints as hard guardrails while continuing to optimize within them. - Realize+ integrates with your CRM, customer lists, and conversion data to build and activate custom audiences, without relying on third-party cookies. Pricing model: Performance-based model; campaigns billed on CPC basis, or CPM for programmatic. Pros: - Agentic AI makes real-time optimization decisions without requiring manual intervention. - Full placement-level transparency so you see exactly where your ads ran. - Access to premium publisher inventory outside Google’s ecosystem. - Self-serve brand safety and placement exclusion controls. - Predictive audience discovery drives incremental reach beyond your known audience. - Automatically generates and optimizes your campaigns, creatives, and audiences. - Integrates with major analytics platforms, Google Analytics, and conversion APIs. Cons: - Network is strongest in native and content formats; not a fit if you need pure search or social inventory. - Minimum spend thresholds apply for managed campaigns. - Best results require clean first-party conversion data to fuel the agentic optimization. ### 2. Criteo Commerce Growth — Best For E-commerce Performance Marketing Criteo Commerce Growth is an AI-powered performance platform built on one of the largest commerce datasets outside of Amazon. The platform’s core advantage is its access to purchase-signal data from more than 22,000 brand and retailer partners, which it uses to predict who’s most likely to buy, when, and at what price point. For e-commerce advertisers, that kind of purchase-intent data is a meaningful differentiator over generic behavioral targeting. The platform covers customer acquisition and retention across the open web, social, video, and connected TV (CTV). Its dynamic creative optimization automatically generates product-specific ads personalized to each user’s browsing and purchase history. Features: - Criteo’s AI-powered predictive bidding is based on user value and purchase intent signals. - Criteo offers dynamic product catalog ads that update in real time, based on the live inventory available. - The platform includes lookalike modeling for new customer acquisition beyond simply retargeting audiences. - Cross-channel activation is available across open web display, social, video, and CTV. - The Commerce Insights dashboard offers users transparent attribution and measurement. Pricing model: Cost-per-click (CPC) and CPM-based, with custom pricing based on spend levels and campaign scope. Pricing is available on request through Criteo’s website. Pros: - Unmatched commerce dataset for purchase-intent targeting. - Strong retargeting capabilities that go well beyond basic pixel-based remarketing. - Cross-channel campaign management in a single platform. - Works without a heavy technical setup, so campaigns can go live quickly. Cons: - Less suited for lead generation or non-commerce verticals. - Some users report a learning curve on the reporting interface. - Customer support quality can vary, per reviews, depending on account size. ### 3. The Trade Desk — Best For Programmatic Transparency and Scale The Trade Desk is the largest independent among DSP open web platforms in the programmatic advertising market, and its independence is the point. Unlike Google’s DV360, which routes spend through its own inventory whenever possible, The Trade Desk has no owned inventory to protect. It gives advertisers access to open web supply across display, video, audio, CTV, digital out-of-home (DOOH), and in-game advertising, with no algorithmic bias toward any particular publisher or format. The platform’s Kokai AI engine brings a layer of predictive optimization to campaigns. Still, The Trade Desk is primarily designed for teams that want expert-level control over their media strategy. It processes more than 13 million impressions per second and provides advertisers with granular visibility into bid decisions, cost structures, and performance by placement. Features: - The Trade Desk utilizes Kokai AI for predictive pacing, audience modeling, and in-flight optimization. - The platform offers omnichannel programmatic advertising across display, video, CTV, audio, DOOH, and more. - The Trade Desk has a robust data marketplace with more than 500 audience data provider integrations. - You get full transparency on CPMs, bid outcomes, and the entire programmatic supply chain. - You also get goals-based buying with multi-key performance indicator (KPI) optimization. Pricing model: Platform fee of approximately 15–20% of media spend. They have custom contracts for agencies and enterprise advertisers. Pros: - No inventory conflicts as optimization is neutral across all supply sources. - Industry-leading transparency into where ads run and what you pay per impression. - Strong CTV and streaming inventory access through direct publisher deals. - Extensive third-party data integrations for precise audience targeting. Cons: - Significant minimum spend requirements make it inaccessible for smaller advertisers. - Steep learning curve; it’s designed for experienced programmatic buyers and agencies. - Contracting and customer service processes have received critical feedback from smaller accounts, per reviews. ### 4. StackAdapt — Best for Multi-Channel Programmatic Without a Steep Learning Curve StackAdapt occupies an interesting position in the DSP market. It has the multi-channel capabilities of an enterprise platform, but employs an interface that experienced programmatic buyers can pick up without a week-long onboarding process. The platform consistently earns high marks for usability in G2 and Gartner reviews, particularly from teams expanding into programmatic for the first time, or moving away from managed service relationships. It covers native, display, video, CTV, audio, in-game, and DOOH advertising, with machine learning optimization across all channels. Its contextual AI targeting is a practical solution for cookieless environments, targeting based on content signals rather than user-level identity data. Features: - StackAdapt offers multi-channel programmatic across native, display, video, CTV, audio, and DOOH. - The platform’s contextual AI targeting works without third-party cookie dependency. - The planner tool includes campaign budget and performance forecasting by campaign parameters. - StackAdapt gives you real-time reporting and customizable dashboards. - Account management support is included at most spend levels. Pricing model: Custom pricing based on campaign budget and spend. No publicly listed minimums, but pricing is available on request. Pros: - It has one of the most user-friendly DSPs in the market. - Strong customer support and account management across account sizes. - Contextual targeting handles cookieless environments without workarounds. - No minimum spend threshold, which makes it accessible for mid-market advertisers. Cons: - Reporting customization has some limitations compared to more mature enterprise platforms. - Less depth in data marketplace integrations than The Trade Desk. - Premium inventory access is more limited for smaller accounts. ### 5. Yahoo DSP — Best For Identity-Resilient Reach With No Ad-Serving Fees Yahoo DSP is one of the more underestimated platforms in the open web advertising space. Its core differentiator is ConnectID, Yahoo’s first-party identity solution, which achieves among the highest match rates on the open web as third-party cookies continue to lose relevance. For advertisers building privacy-first targeting strategies, that identity backbone is a significant operational advantage. The platform provides access to Yahoo’s owned-and-operated inventory across Yahoo News, Yahoo Finance, Yahoo Sports, and AOL, plus a broader programmatic marketplace spanning display, video, native, and CTV. Yahoo also recently expanded its premium publisher access through Yahoo Backstage, which offers direct deals with major streaming and content brands, including Paramount, Disney, Hulu, Spotify, and Roku. Features: - Yahoo DSP offers ConnectID first-party identity matching for cookieless targeting at scale. - The platform includes Yahoo Backstage for direct-deal access to premium streaming and publisher inventory. - Full programmatic supply-chain transparency with the Association of National Advertisers and the Trustworthy Accountability Group (ANA/TAG) TrustNet certification. - No ad serving fees, with transparent audience pricing visible in-platform before launch. - CTV and streaming capabilities across major platforms. Pricing model: Custom CPM/CPC-based pricing. No ad serving fees. Audience data costs are visible inside the platform before campaign launch. Contact Yahoo Advertising for account setup. Pros: - Industry-leading identity resolution via ConnectID, particularly strong in a post-cookie environment. - Transparent ad buying with no hidden ad serving fees. - Direct access to premium streaming inventory through Yahoo Backstage. - Full supply-chain transparency certified through ANA and TAG. Cons: - Platform’s owned-and-operated inventory skews toward news, finance, and sports audiences. - Reporting can lag for CTV and video campaigns. - Creative review and approval timelines can slow launch schedules. ## Key Features to Look for in an Open Web Advertising Platform Not every alternative to PMax serves the same use case. When you’re evaluating platforms, these are the capabilities that distinguish a capable open web solution from one that’s just another walled garden: ### Placement-Level Transparency Channel reporting tells you that Display spent 35% of your budget. Placement-level reporting tells you which sites the budget went to and the cost per conversion. The first is interesting. The second is actionable. Any platform worth moving budget to should give you the second. ### First-Party Data Integration As third-party cookies decline, your customer relationship management system (CRM) data, customer lists, and server-side conversion signals become your most durable targeting infrastructure. Look for platforms that can ingest this data directly and use it to build lookalike audiences or optimize toward your best-converting segments. ### Agentic Optimization Capabilities A system that only executes your manual rules isn’t much better than a spreadsheet. The platforms in this guide all offer some form of AI-driven optimization. The question to ask is how transparent that optimization is. Can you see what decisions the system made, why the platform made them, and where it spent your money as a result? ### Brand Safety and Exclusion Controls Self-serve blocklists, content category exclusions, and placement-level controls matter more than most advertisers realize until they see a screengrab of their ad somewhere unexpected. Confirm that these controls exist, that they’re self-serve, and that they apply at the campaign level, not just as account-level settings you have to ask a rep to configure. ### Cookieless Targeting Pathways Third-party cookies aren’t fully dead yet, but they’re no longer a reliable targeting infrastructure. Platforms with first-party identity solutions, contextual AI, or direct publisher data relationships are better positioned for the direction targeting is heading than those that still primarily rely on cookie-based segments. ### Cross-Channel Measurement One of the benefits of moving your budget outside a single platform is cleaner performance comparisons. Make sure the platform you choose supports third-party measurement partners, so you can apply consistent attribution across channels without being locked into a proprietary measurement framework. ## Preparing Your Data For a Post-PMax Strategy Moving budget to open web platforms involves a change in your data strategy. The platforms that perform best on the open web are those with the richest signals to work with. Here’s what to sort out before you shift spend. ### Clean Your First-Party Data Conversion APIs, CRM integrations, and server-side event tracking give agentic platforms the signal quality they need to optimize effectively. If your only conversion data is browser-based pixels, you’re giving the algorithm a partial picture. Prioritize implementing server-side conversion feeds before scaling spend on any of the platforms in this guide. ### Define Your Optimization Goal Clearly Agentic platforms optimize toward a specific objective. Typically, that objective is a target CPA, ROAS, or to maximize conversions within a budget. The clearer the goal is, and the more conversion data you have feeding it, the faster the system reaches stable performance. Starting a campaign with vague goals and a limited conversion history will lead to slow, costly learning periods. ### Build Audience Segments From Your CRM Customer lists, high-lifetime value (LTV) segment exports, and lookalike seed audiences built from your best customers gives a platform like Realize+ the raw material for predictive audience modeling. Upload these before launch, not after. ### Set Up Cross-Channel Measurement Decide how you’ll attribute conversions across platforms before you start spending. Whether that’s a third-party attribution tool or a consistent Urchin Tracking Module (UTM)-based framework in your analytics stack, having a measurement approach that doesn’t favor any single platform will let you make honest spend allocation decisions as campaigns scale. ## More About Google Performance Max ### The Problem With Google Performance Max When PMax launched, advertisers handed Google a level of control it had never formally asked for before. Prior to PMax, advertisers had full authority over where their ads ran and how their budgets were allocated across every Google-owned surface. The pitch was simple: better AI means better outcomes. However, the results were mixed. PMax drove conversions for many advertisers. It also developed a reputation for some specific behaviors that frustrated campaign managers, media buyers, and performance marketing leads alike, leading them to look for Google Ads PMax competitors. Budget Cannibalization of Branded Search PMax and branded Search campaigns often compete for the same users. When PMax intercepts a branded query that a Search campaign would have converted at a lower cost per acquisition (CPA), it inflates reported return on ad spend (ROAS) without adding incremental revenue. The user was already coming to you, but Google’s system counted it as a win anyway. Channel Reporting Without Channel Control Google’s May 2025 update brought channel-level performance reporting to PMax campaigns, a significant improvement: advertisers could finally see how spend was distributed across Search, YouTube, Display, Discover, Gmail, and Maps. But, transparency isn’t the same as control. Now, you can see, e.g., that your Display allocation is underperforming, but you can’t pause Display while keeping the rest of the campaign running. The algorithm rebalances on its own timeline. Ecosystem Lock-in PMax is, by design, a Google product. Expanding your reach beyond Google’s properties requires separate platforms, separate budgets, and separate measurement frameworks. For advertisers trying to diversify their performance marketing mix, PMax doesn’t scale outward — it scales deeper into the same ecosystem. A real-world example from Grow My Ads makes this concrete: After cutting PMax spend by 80% and redistributing budget to Standard Shopping and Search campaigns, three e-commerce brands saw revenue increases ranging from 35% to 45% with only marginal increases in total spend. The automation wasn’t the problem, but rather the lack of control over how that automation worked. ## What Are Agentic Solutions in Advertising? Basic programmatic automation follows rules. You set bid caps, frequency limits, targeting criteria, and budget allocations. The system executes within those constraints. When something underperforms, you adjust the rules. When you’re not watching, nothing changes. Agentic advertising works differently. An agentic system doesn’t wait for marketers to notice a problem and update a parameter. Instead, it: - Identifies performance patterns, draws conclusions, and makes buying decisions autonomously. - Sets and adjusts bids in real time. - Reallocates spend toward placements that are converting and away from ones that aren’t. - Tests creative variants and shifts budget toward what’s working, without a campaign manager manually reviewing a report first. The distinction is important because it changes what “automation” delivers. Traditional automation is a set of instructions that runs without your involvement. Agentic AI is a system that learns, adapts, and operates with a degree of independent judgment. The goal is the same: efficient ad spending at scale. The mechanism, however, is substantially more capable. What separates a good agentic solution from a basic one is transparency. The best agentic solutions for advertising in this category let you see what their agents are doing and why. They surface the decisions the AI made, show you where your spend went, and give you meaningful controls when you want to override. ## The Benefits of Scaling Campaigns on The Open Web Google’s advertising network is large, but it’s not the whole internet. Consumers move constantly between search engines, news publishers, streaming apps, financial sites, sports platforms, and retail properties. The Reuters Institute’s Digital News Report consistently finds that a substantial share of digital news consumption occurs directly on publisher sites, rather than through Google’s platforms. Streaming audiences on platforms like Peacock, Paramount+, and Spotify aren’t Google inventory, either. Nor are the thousands of premium publisher sites that participate in programmatic marketplaces outside of Google Display Network (GDN). Open web advertising lets you reach these audiences. Here are some specific advantages worth considering: ### Lower Costs Per Mille (CPMs) in Many Categories Google’s inventory is competitive. Premium publisher inventory outside Google’s network is often priced more efficiently, particularly for display and native formats. Lower CPMs don’t automatically mean better return on investment (ROI), but they create more room for testing and optimization without burning through budget quickly. ### Access to Curated, Premium Publisher Inventory Demand-server platforms (DSPs) and performance platforms with strong publisher relationships give you access to brand-safe, premium environments. Think major news outlets, financial publishers, lifestyle media, and CTV networks. ### Escape From Self-Referential Measurement When you run PMax, Google measures the results in Google’s attribution system. When you diversify across open web platforms, you can apply consistent cross-channel measurement and make comparisons that aren’t built on Google’s own math. ### Reduced Dependency Risk Advertisers who concentrated their spend entirely in walled gardens discovered how exposed they were when platforms changed policies, raised prices, or altered algorithms. Open web diversification distributes that risk. ## Key Takeaways PMax remains a useful tool for advertisers with straightforward conversion goals and no strong preference for where their ads appear within Google’s properties. The 2025 transparency updates were real improvements. But, transparency without control is still a constraint, and it’s a constraint that doesn’t exist on the open web. The five Performance Max alternatives in this guide represent different entry points to open web performance advertising, although your choice will ultimately come down to your company’s specific needs. ## Frequently Asked Questions (FAQs) ### What is an agentic solution in advertising? An agentic advertising solution uses AI to make independent campaign decisions without waiting for human input. Instead of executing a fixed set of rules you’ve defined, an agentic system monitors performance data in real time, draws its own conclusions about what’s working, and adjusts bids, budget allocations, and placements accordingly. The key difference from standard automation is adaptability: Agentic platforms respond dynamically to changing conditions, rather than remain within a static parameter set. ### Is Performance Max going away? Performance Max isn’t being discontinued: Google continues to develop it and has been pushing advertisers toward it as the default campaign type for full-funnel conversion goals. The 2025 updates, particularly channel-level reporting, addressed some of the transparency criticisms that had built up over several years. What hasn’t changed is the fundamental constraint: PMax operates exclusively within Google’s ecosystem, and advertisers who want to reach audiences on premium publisher sites, streaming platforms, and the broader open web need additional platforms to do so. ### Why scale on the open web instead of just Google and Meta? Consumers spend more than half of their digital time on properties unrelated to Google or Meta. Publisher sites, news platforms, streaming apps, financial media, and content networks are all walled garden alternatives that capture meaningful audience attention that neither walled garden can reach. Scaling on the open web through DSPs and performance platforms like Realize+ gives you access to that audience, often at lower CPMs than you’d see inside Google’s competitive inventory auction, with the added benefit of transparent placement-level reporting that tells you exactly where your ads ran. ### How do PMax alternatives provide better brand safety? PMax’s brand safety controls are largely at the account level and require working with a Google rep to implement domain exclusions. Most open web DSPs and agentic platforms, such as Realize+, offer self-serve placement exclusions, content category blocklists, and domain-level controls that you can configure directly in the campaign interface. This gives you faster, more granular control over where your ads appear, without needing to submit a request and wait for it to take effect. --- ### What's Holding Back Agentic AI Adoption? Risks of Delay Explained URL: https://www.taboola.com/marketing-hub/risks-agentic-ai-adoption-performance-advertising/ Last Modified: 2026-05-23 20:18:19 An overwhelming 82% percent of marketers say agentic AI on the open web represents a real growth opportunity, but most still aren’t using it. That gap isn’t a fluke, it’s where the industry stands right now: strong belief, slow action, and a growing divide between the companies experimenting early and the ones waiting on the sidelines. The hesitation is understandable. Agentic AI is moving fast, and many teams are still trying to figure out where the technology fits into existing workflows, budgets, and measurement models. But, while others debate the timing, early adopters are already learning what works, refining their strategies, and building advantages that won’t be easy to catch up to later. ## The Paradox: High Belief, Low Action — What the Data Actually Shows In performance marketing, 82% agreement on anything is rare. This is an industry built on skepticism, where marketers learn to question vendor claims, inflated projections, and “next big thing” narratives. Getting four out of five professionals to agree on a growth opportunity almost never happens unless the results already feel undeniable. Interestingly, in this case, the results aren’t fully proven yet. Not at scale, anyway. That’s what makes the data so unusual: most of the marketers who believe in agentic AI haven’t actually acted on that belief. They aren’t testing the channel and walking away disappointed, they aren’t running pilots that failed to perform. Instead, many are stuck in a middle ground: convinced of the opportunity, but slow to operationalize it. That distinction matters. From the outside, hesitation and disbelief can look identical. In practice, they’re completely different market signals. A skeptical market has already decided something won’t work. A hesitant market believes it probably will, but hasn’t acted yet. The survey data points clearly toward hesitation, not rejection. Marketers aren’t dismissing the channel — they’re waiting, evaluating, delaying rollout, or struggling to determine how it fits into existing systems and priorities. All of which raises a more interesting question: Why has a near-consensus belief, in an industry that rarely agrees on anything, produced so little action so far? ### Where the Market Actually Stands The survey breaks that 82% agreement into three distinct groups, each representing a different stage of open web advertising adoption: - The Untapped Potential (46%): Nearly half of marketers see agentic AI as a meaningful growth opportunity, but have yet to act on it at scale. - The Hesitant Believers (19%): This group recognizes the value, but is waiting for the right solutions or internal conditions to align. - The Early Scalers (17%): These marketers already treat open web campaigns as a proven growth driver and run them at scale. What’s striking is how small the skeptical segment actually is. Only about 15% question whether the incremental impact is real, while just 3% are openly dismissive or haven’t seriously evaluated the category at all. Again, that means 82% of the market falls on the believing side of the equation. Just 18% remain skeptical or unevaluated, and even within that group, most aren’t rejecting the technology outright. Many are simply unconvinced about one or two aspects of its effectiveness. In other words, belief isn’t the barrier. The real disconnect is between conviction and execution. That, in turn, raises a bigger question that the survey can’t fully answer on its own: If so many organizations already believe the opportunity is real, why are scaled adoption rates so low? ## The Real Barrier Isn't Skepticism, It’s Integration — Unpacking the 54% Finding The biggest obstacle to adoption isn’t doubt. It’s operations. When marketers were asked to identify the single largest internal barrier to adopting agentic AI, more than half pointed to one issue: integrating it into existing workflows. No other challenge came close. Among all performance marketing barriers surfaced in the survey, marketing technology integration stood out as the dominant friction point. Here’s how the responses broke down: - Difficulty integrating into existing workflows: 54%. - Lack of team knowledge or expertise: 12%. - Uncertainty about which technology or vendor to choose: 9%. - Budget constraints: 6%. - Insufficient investment in training and upskilling: 5%. - No major internal barriers: 5%. - Resistance or skepticism from certain teams: 5%. - Leadership misalignment on priorities: 4%. The data is hard to ignore. The industry isn’t being held back by budget concerns. It isn’t even stalled by skepticism about whether the technology works. The dominant challenge is far more practical: figuring out how an autonomous system fits into a marketing infrastructure that was never designed for it. Agentic AI doesn’t operate in isolation. It has to connect to existing tech stacks, attribution models, approval processes, reporting systems, and cross-functional workflows that are already deeply embedded inside organizations. In many cases, the companies that stand to benefit the most are also the ones dealing with the highest operational complexity. That tension shows up clearly in the data. The hesitation isn’t necessarily a sign that organizations are falling behind. In many cases, it’s the natural result of trying to integrate a functionally new operating model into systems built for a very different era of marketing. ### What Integration Difficulty Actually Looks Like on the Ground Imagine a senior media buyer managing $2 million a month across paid channels. That budget already runs through a carefully built system — vendor relationships, attribution logic, reporting cadences, approval workflows, and budget authorization processes that have evolved together over the years. In that environment, adopting an autonomous platform isn’t as simple as adding another tool to the stack. It changes how the system itself operates. Marketing technology integration at this level forces teams to rethink questions like: - Where does an autonomous platform fit alongside existing vendors and channel partners? - How do you measure attribution when the system is making its own optimization decisions in real time? - What happens to reporting workflows that were designed around human-led campaign management? - How do creative approval processes adapt when campaigns move faster than traditional review cycles? - What does budget authorization look like when a machine is continuously allocating spend on its own? None of these problems are impossible to solve, but none of them are minor, either. That’s what makes the 54% figure so important: it’s not a vague objection or a sign of organizational laziness, it’s a reflection of how modern marketing operations are actually built. The gap between belief and adoption doesn’t exist because marketers fail to see the opportunity. It exists because integrating a fundamentally different operating model into an already complex system is difficult, time-consuming, and organizationally disruptive — even for teams that believe the payoff is worth it. ### Why the Other Barriers Are Smaller Than You’d Expect The smaller performance marketing barriers in the survey data are revealing in their own right: - Budget constraints (6%) suggest that most organizations have already concluded that the investment is financially justifiable. The issue isn’t whether they can spend the money, it’s whether they can operationalize the technology effectively once they do. - Vendor uncertainty (9%) points to a market that’s already maturing. Buyers may not agree on the best partner, but most no longer seem confused about whether credible options exist. - Team knowledge gaps (12%) are real, but they’re also manageable. Training teams and building expertise takes time, yet even that challenge ranks far below the complexity of integration itself. The data tells a clear story, that the industry has long moved past the upstream questions. Should we be doing this? Most marketers think yes. Can we afford it? Again, mostly yes. Are there legitimate vendors in the space? For the most part, yes. What remains is the harder question, and it’s one that slows adoption even after belief is established. How do you integrate an autonomous system into the way the organization already works today? ### The Same Pattern Appears in External Barriers The operational bottlenecks don’t stop at internal workflows. When marketers were asked what has limited additional investment in open web advertising channels specifically, the responses followed a similar pattern: - 74% cited the complexity of managing too many vendors and partners. - 71% pointed to the lack of unified attribution and measurement. - 54% raised brand safety concerns. - 42% said they lacked the resources to manage additional channels effectively. Meanwhile, outright skepticism barely registered. Only 5% said they don’t believe the channel can reach incremental users. Just 2% said they don’t believe incremental performance is achievable at all. Once again, the market isn’t signaling a lack of confidence in the opportunity itself, it’s signaling friction in the systems required to support it. Marketers largely believe the performance upside exists. What they’re struggling with is the complexity that comes with adding another layer of vendors, measurement frameworks, governance, and channel management into already crowded marketing ecosystems. ## Why Larger Organizations Feel This Most Acutely The integration challenge isn’t evenly distributed across the market. In fact, it scales almost directly alongside monthly budget size. When asked to identify their primary obstacle, the percentage of organizations pointing to integration grows significantly as spend increases: - $300K–$499K budget: 9%. - $500K–$999K budget: 38%. - $1M–$4.9M budget: 74%. - $5M+ budget: 68%. At first glance, that seems backward. Larger organizations typically have more resources, more robust teams, stronger technical infrastructure, and more specialized expertise. In theory, they should be better positioned to absorb a new system than smaller, leaner companies, but the data points in the other direction. The reason is fairly straightforward: Complex systems are harder to change than simpler ones. As organizations scale, so does the complexity surrounding every operational decision: - More infrastructure creates more integration points. - More legacy systems create more dependencies and workarounds. - More stakeholders introduce more competing priorities and approval layers. - More established performance creates greater risk if implementation goes poorly. In other words, the very sophistication that made these organizations successful is also what makes transformations more difficult. These companies aren’t starting from scratch: they’ve spent years building performance engines, attribution frameworks, reporting structure, vendor ecosystems, and cross-functional operating rhythms that already work at scale. Integrating a new autonomous system into that environment is inherently more complicated than adding it to a smaller, less mature operation. That complexity shouldn’t be mistaken for resistance or failure. It’s a predictable consequence of scale. But, it also means the pressure to solve the integration problem is greatest where the opportunity is largest. The organizations managing the biggest budgets have the most to gain from successful adoption — and potentially the most to lose if they fall behind when competitors figure it out first. ## The Cost of Waiting: Why Senior Leaders and Big Spenders Feel the Urgency More The integration barrier helps explain why many organizations haven’t moved yet, but the survey reveals something equally important: The cost of waiting isn’t distributed evenly. It falls hardest on the leaders managing the largest budgets. Overall, 75% of marketing leaders say finding an incremental performance channel beyond search and social is very or extremely important, but the urgency becomes much clearer when the data is segmented by seniority. Of senior management weighing in, the following percentage rated it as “extremely important”: - 53% of VPs. - 20% of Directors. - 15% of Senior Managers. The same pattern appears when segmented by monthly spend, showing a dramatic escalation in urgency: - 3% of $300K–$499K spenders. - 5% of $500K–$999K spenders. - 24% of $1M–$4.9M spenders. - 70% of $5M+ spenders. That sharp jump at the highest spending tier is revealing. The people managing the biggest budgets are also the people seeing diminishing marginal returns most clearly. When an organization is spending millions each month across search and social, the effects of saturation become impossible to ignore. Every additional dollar works a little less efficiently than the one before it. At that scale, finding a new source of incremental growth becomes a financial necessity. The organizations feeling the greatest urgency are also the organizations with the most capital available to reallocate once operational barriers are solved. The budgets already exist. The intent already exists. What’s missing is the ability to integrate and scale confidently. That’s where the competitive advantage begins to form. The first-mover advantage in agentic AI on the open web isn’t theoretical. It shows up in budget share, learning curves, operational maturity, and performance efficiency, and it will likely flow to the organizations that solve the integration challenge before everyone else does. ### The Diminishing Returns Problem Nobody Talks About When a company spends tens of millions of dollars a year on paid search and paid social, diminishing returns stop being theoretical. They become a reality. At a certain scale, every additional dollar produces less incremental impact than the one before it. That’s the underlying force driving much of the urgency in the survey data. The mechanics are familiar to anyone managing large performance budgets: - High-intent audiences have already been reached: The customers most likely to convert are already seeing your ads — often alongside competitors targeting the same users. - Ad fatigue compounds over time: Repeated exposure to the same creative naturally reduces responsiveness. - CPAs begin to rise: As the most efficient inventory becomes more competitive, incremental conversions get more expensive. - Auction pressure intensifies: More advertisers compete for the same keywords, audiences, and placements, pushing costs higher across the board. The data reveals a heavy reliance on these saturated channels: - 74% of respondents allocate at least 25% of their performance budget to paid search. - 67% do the same with paid social. These channels still drive enormous value, but they’re also where the limits of scale become most visible, especially to the executives closest to the numbers. For a senior leader overseeing a $5-million-per-month performance program, saturation isn’t an abstract economic concept. It’s the day-to-day experience of watching marginal returns flatten across the two channels consuming the majority of the budget. Once that happens, finding a new source of incremental performance stops being optional. ## What Crossing the Adoption Threshold Looks Like in Practice The 17% of organizations already running open web performance campaigns at scale didn’t get there by waiting for the integration problem to solve itself. They got there by treating agentic AI adoption as an operational transformation, rather than a technology purchase. If the barrier is operational, the solution has to be operational, too. In practice, that usually comes down to three disciplines the rest of the market is still working through. ### Mapping Workflows Before Adoption, Not After The fastest-moving teams identify the friction points early. Before contracts are signed or pilots begin, they already understand which approval chains, reporting cadences, attribution models, and decision-making processes will need to evolve. That work happens up front, not reactively in the middle of implementation. ### Building Measurement Infrastructure in Advance One of the fastest ways for a pilot to stall is trying to measure incremental impact using a framework that was never designed for autonomous optimization in the first place. Organizations that scale successfully tend to establish their measurement logic early, so the system’s contribution can be evaluated from the start instead of debated later. ### Starting With Contained Pilots The organizations succeeding in this space rarely try to automate their entire performance operation at once. Instead, they begin with a tightly defined audience, objective, and budget range. That creates room for the operating model to mature in a controlled environment before broader rollout begins. None of this is especially flashy. In fact, most of it looks like disciplined operational planning — the kind of work that determines whether transformational efforts actually scale. The 17% who have crossed the threshold weren’t necessarily early because they had better technology. They moved first because they treated workflow design, measurement, and organizational alignment as core parts of the adoption process from the beginning. ## Closing Thought: The Barrier Is Real, but It’s Operational, Not Philosophical The 17% already running open web performance campaigns at scale aren’t simply ahead on technology. They’re ahead on the operational learning curve — the part competitors can’t shortcut later by writing a bigger check. That’s the underappreciated dynamic in the survey data. Belief is no longer a differentiator. Budget isn’t either. What separates the Early Scalers from the Untapped Potential is the willingness to treat workflow design, measurement infrastructure, and organizational alignment as the actual work of adoption, not the prerequisites to it. At this point, the philosophical debate about whether agentic AI belongs in performance marketing is largely over. Most of the industry has already answered that question for itself. What matters now is execution. Who can integrate the technology effectively? Who can adapt workflows, measurement systems, and reporting structures quickly enough to scale? Who can move from belief to operational maturity before competitors do? Because conviction alone no longer creates an advantage. Execution does. --- ### 8 Best AI Performance Platforms: The Top Options in 2026 URL: https://www.taboola.com/marketing-hub/ai-performance-advertising-platforms/ Last Modified: 2026-07-06 08:43:21 It’s no secret that artificial intelligence (AI) is quickly changing how performance advertising works. What used to take hours of manual optimization can now happen in real time, across targeting, bidding, and creative. Today’s ad platforms aren’t just automating tasks, but are actively making decisions based on live data. For digital advertisers, that means better efficiency, faster learning, and more scalable growth. With so many AI-powered performance platforms to choose from, though, how do you know which one will work best for your campaigns? This guide breaks down the best AI performance platforms to help you choose the right one. ## Best AI Performance Platforms for Advertising in 2026 Platform Why It Matters Core Features and Use Cases Best for Pricing 1. Realize Uses predictive AI to optimize spend and placements across the open web. Predictive optimization, automated bidding, creative generation, placement insights. Advertisers looking to scale beyond the walled gardens of search and social. Performance-based model; campaigns billed on CPC basis, or CPM for programmatic. 2. Albert.ai Fully autonomous campaign management across channels. AI handles bidding, budgeting, segmentation, and optimization. Enterprise teams wanting hands-off execution. Custom (typically enterprise). 3. Madgicx Focused AI optimization for Meta campaigns. Budget automation, creative insights, performance tracking. Advertisers heavily invested in Meta. Tiered/custom pricing. 4. Smartly.io Combines creative production with AI-driven media buying. Dynamic creative, predictive budgets, cross-platform automation. Large brands running multi-channel campaigns. Enterprise pricing. 5. Ryze AI Simplifies bid and budget optimization using AI. Automated bidding, anomaly detection, campaign adjustments. Teams wanting easy-to-use AI tools. Varies, see provider. 6. Optmyzr Blends automation with human control for pay-per-click (PPC). Rule engine, optimization workflows, reporting tools. Agencies and in-house PPC teams. Subscription tiers. 7. Birch (formerly Revealbot) Rule-based automation for campaign control. Automated triggers, scaling rules, and performance monitoring. Advertisers in need of constant automation. Subscription/usage. 8. OmniReach AI Centralizes cross-channel campaign management. Unified dashboard, automated budget allocation. Teams managing complex, multi-touch journeys. Enterprise pricing. ### 1. Realize Why it’s essential: Realize is a performance-driven AI advertising platform designed to help growth teams scale beyond the walled gardens of search and social by accessing the open web. It's primarily used to bridge the gap between automated creative generation and high-intent media buying, allowing advertisers to reach audiences on premium news, lifestyle, and tech sites. By utilizing a proprietary predictive engine, the platform identifies users most likely to convert and matches them with ads in real time. Marketers use Realize as an automated command center to manage the entire campaign lifecycle, from producing high-volume AI creatives and landing pages to executing complex bidding strategies. The platform is particularly effective for those seeking sustained performance stability, as its machine learning algorithms continuously ingest conversion data to refine targeting and minimize ad waste. This enables brands to maintain a consistent, scalable presence across a massive publisher network, without the need for a large internal operations or data science team. Showcased features: - Maximize Conversions: An automated bidding strategy that leverages real-time signals to prioritize spend on the specific user opportunities most likely to complete a conversion event. - SpendGuard: A 24/7 budget protection algorithm that automatically identifies and blocks underperforming publisher sites or creatives. - AI Creative Suite: Gen-AI Ad Maker and Landing Page builder for hyper-localized, high-CTR assets. - Advanced Targeting: Predictive Audiences and Contextual Topic targeting using semantic AI analysis. Best for: Performance advertisers requiring hands-off optimization and creative-led scaling across premium publisher inventory (Open Web), rather than social walled gardens Pricing model: Performance-based model; campaigns billed on CPC basis, or CPM for programmatic. Pros: - The platform uses “Improved Matchmaking” algorithms that predict top-performing sites, resulting in a reported 20% increase in CVR during the early stages of the advertiser lifecycle. - Unlike platforms that only offer AI for bidding, Realize includes a Gen-AI Ad Maker and Gen-AI Landing Page builder. This ensures that the ad creative and the destination page are contextually aligned for maximum conversion. - Tools like SpendGuard and Custom Rules act as a 24/7 safety net, automatically pausing underperforming ads and capping spend on low-quality supply without requiring manual intervention. Cons: - Some advanced AI features, such as the Performance Simulator, are currently in Beta, which may mean less historical stability for risk-averse advertisers. - While AI accelerates results, the predictive models (like Predictive Audiences) still require seed data to fully optimize, which may challenge advertisers with very short-term or low-volume campaigns. - Although using the GenAI toolkit integrated with Realize boosts your ad approval rate, even if you decide to proceed with manual uploads, you may still benefit from AI assistance. If it’s required, though, the Real-time Ad Compliance feature (powered by Abby) provides auto-fixes for your convenience and the quickest possible resolution. ### 2. Albert.ai Why it’s essential: Albert.ai is one of the closest things to fully autonomous campaign management. It removes much of the manual work from performance marketing by taking over execution across channels. Instead of optimizing campaigns piece by piece, Albert analyzes performance data holistically and automatically adjusts budgets, bids, and targeting. This makes it especially powerful for large teams managing complex campaigns. It can process far more data than a human team and react faster to changes in performance. However, the trade-off is control. Albert makes decisions for you, not with you. Showcased features: - Autonomous optimization: AI continuously adjusts bids and budgets without manual input. - Cross-channel execution: Manages campaigns across search, social, and other channels. - Audience segmentation: Identifies and targets high-value user groups automatically. Best for: Enterprise advertisers looking for hands-off campaign management. Pricing model: Custom (enterprise). Pros: - Reduces workload by automating most campaign decisions. - Very good at identifying patterns across large datasets. Cons: - Limited transparency into how decisions are made. - Less flexibility for marketers who want granular control. ### 3. Madgicx Why it’s essential: Madgicx focuses on improving performance within the Meta ecosystem. It combines automation with strong creative and audience insights, making it a practical tool for teams already investing heavily in these platforms. Rather than replacing the advertiser, it enhances decision-making. It surfaces what’s working, suggests optimizations, and automates repetitive tasks like budget shifts. Showcased features: - Budget automation: Dynamically reallocates spend to top-performing campaigns. - Creative insights: Analyzes which visuals and messages drive results. - Audience targeting tools: Helps identify and scale high-performing segments. Best for: Advertisers focused on Meta Ads. Pricing model: Tiered plans starting at $32 per month, paid annually. Pros: - Strong creative analytics help improve ad performance quickly. - Balances automation with human control. Cons: - Primarily limited to the Meta ecosystem. - Can feel overwhelming with so many optimization suggestions. ### 4. Smartly.io Why it’s essential: Smartly.io is built for scale. It connects creative production with media buying, allowing teams to launch and optimize large campaigns efficiently. This is especially important in today’s environment, where creative volume often drives performance. The platform excels at helping teams produce variations of ads and test them quickly, while AI handles budget allocation and optimization. Showcased features: - Dynamic creative optimization: Automatically tests and scales high-performing ad variations. - Cross-platform automation: Manages campaigns across multiple channels from one interface. - Predictive budgeting: Allocates spend based on expected performance. Best for: Large brands running high-volume, multi-channel campaigns. Pricing model: Enterprise. Pros: - Excellent for scaling creative production and testing. - Strong integration between creative and media buying. Cons: - High cost makes it less accessible for smaller teams. - Requires significant creative input to maximize value. ### 5. Ryze AI Why it’s essential: Ryze AI focuses on making AI optimization accessible. It strips away complexity and delivers practical improvements in bidding and budgeting without requiring deep technical knowledge. This makes it a good entry point for teams that want to start using AI without overhauling their entire stack. Showcased features: - Automated bid optimization: Adjusts bids in real time based on performance. - Budget management: Shifts spend toward campaigns delivering results. - Anomaly detection: Flags unusual performance changes quickly. Best for: Small to mid-sized teams looking for simple AI tools. Pricing model: Varies, visit the provider for details. Pros: - Easy to implement and use. - Provides quick performance improvements. Cons: - Lacks the advanced features found in enterprise platforms. - Limited customization for complex campaigns. ### 6. Optmyzr Why it’s essential: Optmyzr is designed for advertisers who want automation without losing control. It provides tools to streamline PPC management while allowing teams to define their own rules and strategies. This makes it especially popular with agencies and experienced marketers who want efficiency, but still want to steer the ship. Showcased features: - Rule engine: Automates tasks based on custom conditions. - Workflow tools: Simplifies campaign management across accounts. - Optimization suggestions: Provides actionable recommendations. Best for: Agencies and in-house PPC teams. Pricing model: Subscriptions start at $209 per month, paid annually; enterprise. Pros: - Strong balance between automation and control. - Saves time on repetitive tasks. Cons: - Requires setup and ongoing management. - Not fully autonomous compared to newer AI tools. ### 7. Birch (formerly Revealbot) Why it’s essential: Birch gives advertisers control through automation. Instead of relying on fully autonomous AI, it allows users to define rules that automatically trigger actions. This makes it ideal for teams that want consistent optimization without giving up control. Showcased features: - Automated rules: Executes actions based on performance thresholds. - Budget scaling: Increases or decreases spending automatically. - Real-time monitoring: Tracks campaigns continuously. Best for: Advertisers who need reliable, rule-based automation. Pricing model: Subscriptions start at $45 per month, paid annually; enterprise. Pros: - Highly customizable automation. - Great for maintaining consistent performance. Cons: - Requires time to set up effective rules. - Not as advanced in predictive AI. ### 8. OmniReach AI Why it’s essential: OmniReach AI is designed for advertisers managing campaigns across multiple channels. It centralizes data and optimization, making it easier to coordinate efforts across platforms. This is especially useful for brands running complex customer journeys. Showcased features: - Unified dashboard: Brings all campaigns into one view. - Automated budget allocation: Distributes spend across channels. - Cross-channel insights: Identifies performance trends. Best for: Teams managing multi-touch campaigns. Pricing model: Enterprise. Pros: - Strong visibility across channels. - Helps align strategy across platforms. Cons: - Requires integration with multiple systems. - Can be complex to implement. ## More About AI for Performance Advertising AI has quickly become a critical component of modern performance marketing. Tasks that were once done manually, like adjusting bids, testing audiences, and shifting budgets, are now handled in real time by machine learning models. The real shift isn’t just automation, though, it’s decision-making. AI platforms can analyze thousands of signals at once and act on them instantly, which is something no human team can replicate at scale. For advertisers, that means faster optimization cycles, more efficient spend, and the ability to scale campaigns without adding headcount. ### What Is AI-driven Performance Advertising? AI-driven performance advertising uses machine learning to continuously optimize campaigns based on real-time data. Instead of relying on manual rules or static targeting, the system learns from user behavior, conversion patterns, and engagement signals. It then adjusts bids, placements, and creative delivery automatically to improve outcomes. In practice, this means campaigns are always evolving, rather than being set up once and left to run. The goal is simple: Get better results, faster, with less manual intervention. ### Benefits of AI in Digital Advertising The biggest benefit of AI is speed. It can process and act on data far faster than any human team, which leads to quicker optimization and better performance over time. AI also improves efficiency by reducing wasted spend, since it continuously shifts budget toward what’s working. Another key advantage is scale, because it allows advertisers to manage larger, more complex campaigns without increasing their workload. Finally, AI enables deeper insights, helping marketers understand not just what worked, but also why. ### How to Measure the ROI of AI Advertising Campaigns Measuring ROI with AI campaigns starts with the same core metrics: conversions, cost per acquisition (CPA), and return on ad spend (ROAS). But it’s also important to look at how quickly campaigns improve over time, since AI systems typically get stronger as they learn. Metrics like time-to-optimization and performance stability can reveal how effective the AI really is. You should also track incremental lift to understand whether AI is driving new results or just optimizing existing demand. In short, return on investment (ROI) isn’t just about the end result, but how efficiently you get there. ### Examples of Successful AI Performance Advertising Campaigns Case Study #1: Olight Doubles ROAS with Realize Retargeting Olight, a global portable lighting brand, faced a common plateau: search and social performance had begun to show diminishing returns, and they needed a way to reach high-intent buyers at scale. Their solution was to use Realize to run retargeting campaigns across the open web, focusing delivery on users who had already demonstrated purchase intent by visiting product pages, adding items to cart, or beginning checkout — but not yet completing a purchase. The results were significant. During Realize campaign periods, Olight achieved nearly 2x higher ROAS. Through continued optimization with their account manager — testing creative variations and refining targeting — they ultimately doubled their overall ROAS, with Realize performing on par with Olight's highest-performing established channels. This case illustrates a core advantage of AI-driven platforms: the ability to identify warm audiences and re-engage them with precision at exactly the right moment, outside the limitations of traditional walled-garden channels. Case Study #2: NYDJ & iQuanti Drive 3X ROAS with Dynamic Creative Optimization Premium women's apparel brand NYDJ, working with digital marketing agency iQuanti, set out to drive incremental sales and reach a 2.5X ROAS target in the U.S. market — specifically among fashion-forward women over 40 who couldn't be found through other channels. Their strategy centered on Realize Product Ads with Dynamic Creative Optimization (DCO), which automatically customized ad creatives in real time for users who had already engaged with NYDJ's content, retargeting them with ads featuring the exact products they had shown interest in. The campaign exceeded its goal, generating 3X ROAS — above the 2.5X target — and ultimately yielded one of the highest ROAS figures across all of NYDJ's advertising channels. Across both campaigns, the pattern is consistent: AI performance platforms work best when they can learn from behavioral signals, personalize creative delivery in real time, and automatically shift spend toward what's driving results. That combination of speed, scale, and continuous optimization is what turns good campaigns into high-performing ones. ## Key Takeaways AI is no longer a future trend in performance advertising; it’s already the foundation. The platforms featured here are moving beyond simple automation and into full campaign orchestration, where targeting, creative, and bidding all work together. As an advertiser, your opportunity is clear: to achieve better performance with less manual effort. That said, the real advantage comes from choosing the right level of control. Some teams will benefit from fully autonomous systems, while others will prefer tools that enhance decision-making. Either way, the advertisers who lean into AI now will have a clear edge as competition continues to grow. ## Frequently Asked Questions (FAQs) ### What are the best AI advertising strategies for small businesses? Start with AI tools that improve efficiency right away, like automated bidding and budget optimization. These help you get better results without increasing spend. Next, use AI-driven audience targeting to find high-intent customers and reduce wasted impressions. As you gather more data, layer in creative testing tools to quickly identify which messaging and visuals convert best. ### What are the key AI technologies used in advertising? Most AI advertising platforms rely on machine learning to analyze performance data and improve results over time. Predictive analytics is used to forecast outcomes and guide budget allocation before campaigns scale. Natural language processing (NLP) powers features like ad copy generation and contextual targeting. Together, these technologies help advertisers make faster, smarter decisions across targeting, creative, and bidding. ### How can I start using AI for e-commerce advertising? Begin by integrating AI tools into the channels you already use, like Meta or Google Shopping. Focus first on automated bidding and dynamic product ads to improve efficiency and scale. Then, use AI-powered creative tools to test variations of product images, headlines, and offers. As your data grows, the AI will become more effective at identifying high-value customers and maximizing return on ad spend. --- ### The State of Agency: How Autonomous AI Is Rewriting the Rules of Performance Marketing URL: https://www.taboola.com/marketing-hub/how-autonomous-ai-is-changing-performance-marketing/ Last Modified: 2026-05-23 20:42:14 Performance marketing has crossed a pivotal line. Marketers are no longer just using artificial intelligence (AI) to inform decisions, but deploying AI systems that make decisions in real time, autonomously, without waiting for a human to sign off. The shift from manual campaign management to always-on autonomous operation has already happened. According to a new survey of 200 senior performance marketers conducted by Realize in March 2026, Google PMax has reached 91% adoption at scale, while Meta Advantage+ sits at 88%. Those are not early-adopter numbers, that’s almost the whole industry. Meanwhile, everything else, like TikTok Smart+, the open web, and every channel outside those two, is still in testing or beta mode. This isn’t a story about where AI is going, but one about a transformation that has already taken place in search and social, and is now working its way across the rest of the marketing map. ## From Automation to Autonomy: What “Agentic” Actually Means in Practice The word “agentic” is often used loosely, so let’s be clear about what it means in reality. Traditional performance marketing automation executes instructions a human already set. You define the audience, set the bid cap, decide when to run and when to pause. The system follows the rules you wrote. The decisions are still yours, just made in advance. Agentic AI works differently. You set a goal, like a target cost per acquisition (CPA) or a return on ad spend (ROAS) threshold, and the system figures out how to get there. It reads performance signals in real time and adjusts continuously without waiting for a human to intervene. As the report describes the process, these are systems that continuously execute strategies and optimize performance in real time. For performance marketers, this changes the nature of the job. The role shifts from managing campaigns day-to-day to setting the conditions for the system to manage them well as an autonomous strategist. That’s still skilled work, but it’s a different kind of skilled work, and it’s already the reality for most teams running search and social. When the system is making real-time decisions about audiences, placements, bids, and creative combinations, the marketer is no longer the one deciding upon those variables directly. That’s a change in how accountability works, and it requires a different relationship with data, with campaign goals, and with the platforms themselves. Getting clear on what you’re optimizing toward matters more than it ever did, because the machine will optimize toward exactly what you tell it to. ### The Shift from “Set It” to “Set the Goal” Not long ago, a performance marketer’s week included pulling reports, identifying audiences that weren’t converting, adjusting bids by segment, rotating creatives, monitoring pacing, and checking placements. Every one of those tasks required time and hands-on manual work or, at best, rules-based updates. In channels where AI campaign management has scaled, most of that work no longer falls to the marketer. You give Google PMax a target CPA and a budget, and upload your creative assets and product feed. The system handles audience selection, bid levels, placement decisions, and creative combinations. The marketer’s focus becomes defining the goal, providing quality inputs, and knowing when to step in if something looks off. ## The Google/Meta Blueprint: Why Mass Adoption Happened Fast The adoption numbers for autonomous advertising on Google and Meta are difficult to overstate. Again, 91% of respondents running Google PMax and 88% running Meta Advantage+ at scale doesn’t describe a trend, but instead captures the new default operating model of the industry. Most new advertising technologies take years to reach anything close to majority adoption, but that wasn’t the case with agentic AI. Why? Performance. When enough marketers ran these tools alongside their manually-managed campaigns and saw better results with less operational overhead, increased budget followed. The survey found that 76% of respondents report moderate to significant performance lift from AI-powered solutions such as PMax and Advantage+, with 29% describing that lift as significant. Adoption at this scale only happens when the proof is in the data. It’s also worth noting that Google and Meta had structural advantages that made this easier to demonstrate. First-party data at scale, closed-loop attribution, and platform-controlled inventory meant these platforms could measure what was working and optimize toward it with confidence. That helps explain why bringing the same approach to other channels is harder, and why the rest of the industry is still working to catch up. ### What Mass Adoption Looks Like in the Data The survey data makes this two-tier reality very clear. Google and Meta are in a category of their own. TikTok Smart+, the next most widely engaged platform, is being tested by 73% of respondents but runs at scale by only 9%. Open web campaign solutions show 36% at scale, with 44% in active pilots. Wide testing and scaled adoption are not the same thing. When 91% of an industry is running something at scale, that’s the default. When 73% are testing something with 9% at scale, the breakthrough hasn’t happened yet, even if the direction is clear. Budget tends to follow proof, and proof takes time to accumulate in new channels. The concentration of scaled adoption in just two platforms means that for most performance marketing programs, the autonomous optimization capabilities that now define best practice are only being applied to a portion of total spend. The rest of the budget — in TikTok, on the open web, in connected TV (CTV), in retail media — is still largely managed the traditional way. That’s the opportunity that the next phase of agentic adoption is moving toward. ### The Proof That Unlocked the Budgets The budget-shift cycle with PMax and Advantage+ is simple and repeated: performance improved, budgets were pulled from manually-managed campaigns to agentic campaigns, further improvements were seen, and adoption continued to scale. Right now, that’s what’s missing from other advertising channels. Of the 76% of respondents seeing meaningful improvement from AI-powered solutions, 29% describe it as significant and 47% as moderate. Only 7% report limited lift. One percent report no impact. Zero percent say they are not measuring at all. The 17% who say it’s too early to determine impact is worth noting, too: that’s not a sign of failure, but it does reflect how long real performance proof cycles take. It also suggests the current 76% figure has room to grow as measurement matures. The top perceived benefit of these solutions, cited by 41% of respondents, is real-time optimization toward CPA and ROAS goals. These performance gains are not random; they come directly from systems that can process more signals and make faster adjustments than any human campaign manager can. ## Where the Market Is Heading: TikTok, the Open Web, and Beyond The history of agentic adoption in search and social gives a useful frame of reference for reading where other channels are right now. PMax and Advantage+ didn’t go from launch to near-universal adoption overnight. There was a period of wide testing, limited scale, and accumulating proof, then the results justified the budget, and scale followed quickly. That pattern is visible in the current data for the channels that come next. TikTok Smart+ is the clearest near-term signal. Seventy-three percent of respondents are currently testing it, with only 9% running it at scale. That specific combination — broad organizational commitment to testing and limited scaled deployment — is what the PMax and Advantage+ adoption curves looked like initially. The open web tells a more complex version of the same story. Forty-four percent of respondents are in active pilots, 36% are already at scale, and 82% see AI-powered goal-based buying there as a meaningful growth opportunity. The demand signal is strong and consistent across the data, but the gap between belief and scaled action is wider than it was for TikTok, and the reason is structural. Google and Meta had closed-loop attribution, first-party data at scale, and platform-controlled inventory. Those advantages made it straightforward to demonstrate that autonomous optimization was working. The open web doesn’t have those same conditions built in, which means proving performance is harder and managing campaigns at scale is more complex. ### TikTok Smart+: The Next Wave Taking Shape The 73% testing figure for TikTok Smart+ carries real weight. Testing at that level requires budget allocation, operational bandwidth, and organizational buy-in. When almost three-quarters of senior performance marketers are actively piloting a platform, the question of whether agentic adoption will happen there is largely answered. If performance proof for TikTok Smart+ starts replicating what PMax and Advantage+ have shown, the conversion from widespread testing to scaled adoption could happen quickly. That’s the pattern the industry has already shown it follows. What makes TikTok an interesting near-term signal is the gap between the testing number and the scale number. A 73% to 9% split looks like a platform that has the attention of almost the entire market, but hasn’t yet produced the consistent performance results that would justify moving significant budget. ### The Open Web: Biggest Opportunity, Biggest Gap The open web is where the most significant tension in this data sits. The demand signal is strong: 82% of respondents view AI-powered goal-based buying on the open web as a meaningful growth opportunity. Seventy-five percent of marketers surveyed rate finding a performance channel delivering incremental outcomes beyond search and social as very or extremely important. Among the highest spenders, those at $5 million or more per month, 70% call it extremely important. The urgency also increases with seniority. Among vice presidents (VPs), 53% rate finding an incremental performance channel beyond search and social as extremely important, compared to 20% of directors and 15% of senior managers. The push to diversify beyond walled gardens is seen most at the level where budget decisions get made. That’s a meaningful signal about where organizational priority sits, even when the investment data doesn’t yet reflect it. Only 4% of companies currently put significant budget (25% or more of performance spend) into the open web. Most maintain a moderate presence. The channel accounts for an average of 13% of total performance marketing budgets today. That gap between recognition and investment reflects something specific about where the open web sits in the current ecosystem. It’s not that marketers don’t believe the channel can work. It’s that the tools required to make it work at scale don’t yet exist at the same level of maturity for the open web. Marketers are being asked to manage something complex manually that, in other channels, the machine handles automatically. The barriers cited in the survey are almost entirely operational: 74% point to too many vendors and the complexity of managing multiple partners, 71% cite lack of unified attribution and measurement, and 54% flag brand safety concerns. These are infrastructure problems, not belief problems. Very few respondents question whether the open web can deliver incremental value. They are held back by the difficulty of proving it and managing it at scale. This is exactly the situation that existed in search and social before agentic solutions arrived. The platforms that solved the measurement problem unlocked the budgets. The open web uses the same automated, goal-based buying tools that make performance proof achievable and operational complexity manageable. The survey data suggests the market is ready for that solution. Ninety-nine percent of respondents said they would allocate open web budget if agentic AI-powered solutions were available, with an average expected allocation of 24% of total performance spend. Seventy-four percent of $5 million or more spenders strongly agree they would increase open web investment if it offered the same automated campaign solutions available in search and social. The market is waiting for the infrastructure to catch up with the intent. ## So What’s Next? Agentic AI has already redefined what performance marketing means in search and social. It has proved, at scale and in account data, that autonomous optimization outperforms human campaign management when the infrastructure is right. The remaining question is not whether this model extends further. Marketers want it to and they will direct their budget toward it when it arrives. The question now is about timing and readiness. The marketers who are building toward agentic capability on the open web by running pilots, establishing measurement foundations, and learning how autonomous optimization behaves outside walled gardens will be ahead of the shift when it lands, not racing to catch up with it. --- ### Dynamic Keyword Insertion: Serving Search Terms To Drive Conversions URL: https://www.taboola.com/marketing-hub/dynamic-keyword-insertion/ Last Modified: 2026-05-17 11:34:31 You know those signs you see plastered across bus stop benches, or up on billboards, that read something like, “Your Ad Here”? We’ve all seen signage like this, and rather than compelling most marketers to jump at the chance to place ads in these locations, these signs seem to confirm that, in fact, said advertising opportunities are less than desirable. Now, a digital ad that’s served directly to an online party based on their browser history, search terms, social media algorithms, and more? That can be a more compelling opportunity. And when advertisers can serve users ads containing the very search terms they just used, the chance for conversion goes up dramatically. The use of search terms in ad copy is the digital equivalent of someone walking up to a bus stop while wondering aloud where to find good Italian food in the area, only to see an advertisement for an Italian restaurant posted on the bench before them. In the “real” world, that level of serendipity is improbable. Online, however, dynamic keyword insertion allows for highly customized ads delivered at just the right time. Here’s how marketers can make the most of this strategy. ## What Is Dynamic Keyword Insertion (DKI)? Dynamic keyword insertion (or DKI, for short) is an online ad feature that automatically replaces a specific part of a digital ad's copy with a keyword or phrase matching a user's search query. This makes ads more relevant by showing the exact term the user searched for, which can improve click-through rates (CTR) and quality scores, and can save advertisers time, as you won’t have to create separate ads for every single keyword variation. ## How Dynamic Keyword Insertion Works ### Dynamic Keyword Insertion in Google Ads In Google Ads, DKI works by inserting a keyword from your ad group (that matches a user’s search query) into a predefined part of your ad copy. Advertisers place a placeholder in the ad text using a format like: {KeyWord:Default Text} If the keyword fits character limits and formatting rules, it’s dynamically inserted. If it doesn’t, the default text appears instead. Advertisers can also control capitalization (keyword, Keyword, KeyWord) to ensure the inserted term looks natural. This approach is especially useful for search campaigns, where matching the user’s exact query can significantly boost perceived relevance and CTR. ### Dynamic Keyword Insertion in Realize Realize approaches DKI differently. Instead of inserting search terms, Realize uses dynamic macros to insert contextual information into titles. These values are generated based on where or how the user is viewing the campaign item. Examples of dynamic values supported by Realize include: - Location: Country, region, DMA, or city. - Device: Platform. - Timing: Day of week. A Realize DKI title might look like this: People in ${city:capitalized}$ Can’t Get Enough of This Razor Subscription Service When the campaign runs, ${city:capitalized}$ is replaced with the viewer’s city (for example, “Phoenix”), creating a highly specific, personalized title. Where to set it up in Realize DKI can be applied when adding items to a new campaign or when editing items in an existing campaign. Navigate to the Ads Report, locate the campaign item you want to update, and add the DKI macro directly into the title field. To use DKI in Realize titles, advertisers must follow a strict macro format: ${keyword:capitalizationtype}$ When you enter your text, both the keyword and capitalization type must be written in all lowercase, or the campaign item will be blocked from going live (you can choose the form of capitalization you wish to be displayed later). Supported Keywords - Country - Region - DMA - City - Platform - Dayofweek Supported Capitalization Types - Uppercase – ALL CAPS - Capitalized – First Letter Capitalized Only one dynamic keyword is allowed per title, and dynamic keywords can only be used in titles — not descriptions or body copy. Common setup errors to avoid - If the macro is not formatted exactly as specified, the campaign item will not go live on the network. - All characters in the macro must be lowercase — any uppercase letter will block the item. - Only one DKI macro is permitted per title. A second macro in the same title will cause the item to be blocked. - Always double-check spelling and casing before saving. If a value can’t be determined (due to GDPR restrictions, privacy limitations, or unavailable location data), the campaign item may be blocked, so careful testing and validation are essential. ## Setting Up and Implementing DKI Implementing DKI effectively requires careful keyword selection and ad group structuring so that the inserted terms remain relevant and natural. In Google Ads, this often means matching a user’s search query—for example, using {KeyWord:Shoes} to dynamically show "Running Shoes" or "Leather Shoes." In certain performance advertising platforms, implementation focuses on contextual macros rather than search terms. To set this up, you must use a specific dollar-sign syntax: ${keyword:capitalizationtype}$. For example, a title like “People in ${city:capitalized}$ Love This Service” will automatically update to “People in Phoenix...” or “People in Chicago...” based on the viewer’s location. It might be critical to remember that for some platforms, the keyword and capitalization type must be entered in all lowercase during setup (e.g., ${city:capitalized}$), or the campaign item will be blocked from going live. ### Placing the Code in Ad Headlines or Descriptions When placing DKI code, the best approach is to prioritize clarity and natural flow. In Google Ads, headlines are the most visible spot for DKI, though it can also be used in descriptions to reinforce relevance. However, you must ensure the inserted term doesn’t exceed character limits, or the default text will be triggered. You may find that for some performance advertising software, the rules are more strict: dynamic keywords can only be used in titles—they are not supported in descriptions or body copy. You would be allowed for only one dynamic keyword per title. Because some platforms recommend keeping titles under 60 characters to avoid truncation, advertisers should ensure that the combination of their static text and the dynamic value (like a long City or DMA name) stays within those bounds. If a dynamic value cannot be determined due to privacy restrictions or missing data, the campaign item may be blocked, making careful validation essential during the setup phase. ### Formatting for Correct Capitalization Managing capitalization with dynamic keyword insertion is all about choosing the right capitalization setting so your ads look polished and professional. Google Ads offers three capitalization formats when you insert {KeyWord:Default Text}: Keyword: Capitalizes only the first word, as in: “Running shoes.” KeyWord: Capitalizes each word, as in: “Running Shoes.” KEYWORD: Capitalizes all letters, as in: “RUNNING SHOES.” This choice depends on where the DKI appears. In headlines, using “KeyWord” is often best because it makes each word stand out and aligns with typical Title Case formatting. In descriptions, “Keyword” tends to feel more natural and conversational. Full uppercase (“KEYWORD”) should be used sparingly, usually for emphasis, since too much can look spammy or aggressive. The key is consistency: Match the capitalization style to the rest of your ad copy so the dynamically inserted terms blend seamlessly. ### Using Query Parameters for Landing Page DKI Using query parameters for landing page dynamic keyword insertion is a smart way to align the ad experience with the user’s search intent. The basic idea is that when someone clicks your ad, the keyword they searched for can be passed through the URL as a query parameter (e.g., ?keyword=running+shoes). Your landing page then reads that parameter and dynamically updates certain elements — like headlines, product categories, or call-to-action text — to match the visitor’s query. This creates a seamless experience where the messaging in the ad and the landing page are consistent, which can improve relevance, engagement, and conversion rates. It’s best to keep the implementation simple by only swapping text in places where it feels natural, ensuring that default values are set so the page doesn’t break if no keyword is passed, and avoiding overusing DKI so the page doesn’t look repetitive or robotic. It’s also important to test formatting and capitalization to ensure dynamically inserted terms look professional. ### Guided vs. Manual Setup Methods When thinking about guided vs. manual DKI setup methods, the difference comes down to convenience versus customization. Google Ads provides guided setup through its interface, where you insert {KeyWord:Default Text} directly into headlines or descriptions with prompts and built-in formatting options. This method is expedient and beginner-friendly because the platform handles capitalization rules, default text fallbacks, and ensures compliance with character limits. It’s ideal if you want to quickly add DKI without worrying about syntax errors. Manual setup, on the other hand, involves writing the DKI code into the ad copy yourself. This gives you greater control over how keywords appear, how defaults are structured, and where insertion happens. For example, you can strategically place {KeyWord:Default Text} in multiple parts of the ad, experiment with capitalization styles, and tailor default text to match brand tone. Manual setup is more flexible but requires careful testing to avoid clunky phrasing or mismatched keywords, and it brings a greater chance of error. In practice, guided setup is best for advertisers new to DKI or managing large campaigns where speed matters. Manual setup suits advanced advertisers who want to fine-tune messaging and ensure every keyword insertion feels natural. Many marketers combine both, using guided setup for efficiency, then manually refining ads for polish and performance. ## Expert Tips for Using DKI Effectively in Realize A few words to the wise about DKI and Realize in particular: Keep titles simple and flexible: Write titles that still read naturally once the dynamic value is inserted. Use tightly scoped intent: Pair DKI with strong targeting so personalization feels intentional, not gimmicky. Follow macro rules exactly: One typo, uppercase letter, or extra macro will block the campaign item. Test before scaling: Always preview and QA dynamic titles to ensure values render correctly. Don’t over-automate: Combine DKI-driven titles with strong static messaging to maintain brand voice and clarity. ## Key Benefits and Advantages of DKI DKI is a powerful feature that automatically inserts the keyword a user searched for into your ad copy. This creates a sense of personalization and relevance that generic ads often lack. The first major advantage is increased relevance: When a searcher sees their exact query reflected in your headline or description, the ad feels directly connected to their intent, which builds trust and encourages clicks. Another benefit is higher click-through rates. Because ads appear more tailored, users are more likely to engage. This often leads to improved Quality Scores in Google Ads, which can reduce your cost-per-click and improve ad placement. Advertisers also save time with DKI, since you don’t need to manually create dozens of ad variations for every possible keyword. Instead, one ad template can dynamically adapt to multiple queries. Finally, DKI can improve conversion rates by ensuring consistency between the ad and the landing page. When users see their exact keyword echoed in both places, they feel reassured they’ve found what they’re looking for, reducing bounce rates and boosting conversions. ### Maximizes Ad Relevance and User Experience DKI maximizes ad relevance and user experience by making ads feel directly connected to what a person is searching for. When a user types in a query, DKI automatically inserts that exact keyword into the ad headline or description. This means the ad mirrors the searcher’s intent, which immediately signals that the advertiser offers exactly what they’re looking for. That heightened relevance increases the likelihood of clicks because the ad doesn’t feel generic — it feels personalized. From a user experience perspective, DKI reduces friction. People see their own words reflected back in the ad, which reassures them they’re on the right path. When the landing page also reflects those keywords (often through query parameter insertion), the journey feels seamless: The search term leads to an ad that matches, which leads to a page that delivers. This consistency builds trust, lowers bounce rates, and improves satisfaction. ### Drives Higher Click-Through Rates (CTR) DKI drives higher click-through rates by making ads feel more relevant and personalized to each user’s search query. When someone types in a keyword, DKI automatically inserts that exact term into the ad headline or description. This creates a strong psychological effect: Users see their own words reflected back, which signals that the ad is directly addressing their intent. That immediate relevance increases the likelihood they’ll click because the ad doesn’t look generic — it looks tailored to them. Also, DKI helps scale personalization without requiring dozens of separate ad variations. Instead of writing individual ads for every keyword, one DKI-enabled ad can adapt to multiple queries, ensuring that each impression feels customized. This combination of relevance, visibility, and efficiency is why advertisers often see CTR improvements when using DKI strategically. ### Improves Quality Score Another way DKI boosts CTR is by improving visibility. Ads with dynamically inserted keywords often match the searcher’s query more closely, which can make them more eye-catching in search results. As noted above, this alignment also contributes to better Quality Scores in Google Ads, meaning your ads can show in higher positions at lower costs, further increasing the chances of clicks. ### Streamlines Ad Group Management DKI streamlines ad group management by reducing the need to create dozens of separate ads for every possible keyword variation. Normally, advertisers would have to manually write multiple ads to cover different search terms, but with DKI, one ad template can dynamically adapt to each keyword in the ad group. This means you can maintain fewer ads while still achieving highly personalized messaging for a wide range of queries. This method also simplifies organization because you don’t have to split ad groups as finely to ensure keyword relevance — DKI automatically handles that by inserting the right term into the ad copy. As a result, campaign structures can be leaner, easier to manage, and less time-consuming to update. For example, instead of building separate ad groups for “running shoes,” “dress shoes,” and “leather shoes,” you can keep them together and let DKI adjust the headline to match each query. ## Expert Strategies and Optimization Tips Here are some additional ways to make the most of DKI for increased conversions: Use tightly themed ad groups: Keep ad groups focused on closely related keywords. This ensures that dynamically inserted terms always make sense in the ad copy and avoids stilted or irrelevant phrasing. Set smart default text: Always set a clear, compelling default word or phrase in your {KeyWord:Default Text} code. This protects your ad from breaking when a keyword is too long or doesn’t fit character limits. Control capitalization: Use the right capitalization format (Keyword, KeyWord, or KEYWORD) depending on placement. Headlines often benefit from title case (KeyWord) for professionalism, while descriptions may look more natural with sentence case (Keyword). Avoid branded or competitor terms: Experts caution against using DKI with competitor names or trademarks. This can lead to misleading ads, policy violations, or even legal issues. Align your landing page: Ensure your landing page reflects the dynamically inserted keyword. Passing query parameters to the page can help maintain consistency, which improves user trust and conversion rates. Test and monitor performance: DKI isn’t a “set it and forget it” tool. Regularly test different placements (headline vs. description), monitor CTR and Quality Score, and refine ad groups to maximize results. Balance automation with control: While DKI saves time, don’t rely on it exclusively. Combine dynamic ads with carefully crafted static ads to maintain brand voice and avoid over-automation. ## Common Mistakes to Avoid (When Not to Use DKI) Using broad match keywords: Broad match keywords can result in non-relevant or confusing terms being inserted into your ad. It's best to stick to an exact match to ensure the most relevance. Instituting poor ad group structure: Lumping unrelated keywords into the same ad group leads to a higher risk of awkward or nonsensical ad copy, which will turn wary customers off to your brand. Using irrelevant or branded keywords: Avoid using DKI for branded keywords, as they should be in their own ad group. Additionally, avoid using keywords that are irrelevant to your ad's message, as this can confuse users and lead to a lower Quality Score. Ignoring character limits: Long keywords can break your ad if they exceed the character limits for the headline or description. Violating trademarks: Be cautious about using brand names as keywords, as this can lead to trademark violations and ad rejections, if not more serious legal issues. ## Key Takeaways Dynamic keyword insertion, known as DKI for short, can greatly enhance your online ads’ click-through rates (CTR) and can lead to better conversions. DKI can lead to improved ad relevance, with ads matched to user intent more closely. Beyond leading to higher CTRs, it can also mean better quality scores with better performance and lower costs. DKI can also save time, as fewer manual ad variations will be needed. Finally, DKI can lead to much better conversion rates, thanks largely to the customized but also the consistent messaging users see as they travel from ad to landing page. ## Frequently Asked Questions (FAQs) ### What happens if the keyword is too long to fit in my ad text? If a user’s keyword is too long for the space designated in your ad, the system uses your fallback (default) text instead. This ensures your ad still makes sense and doesn’t get cut off or display awkwardly. In Realize, when using DKI in your title, you must follow their macro format (${Keyword:…}$). If the dynamic value would make the title too long (or violates formatting rules), the ad can be blocked in Real-Time Ad Compliance. Realize recommends keeping titles under 60 characters to prevent truncation. ### Does dynamic keyword insertion (DKI) only work for Google Ads? No, DKI is a broader online marketing technique used across many different advertising systems, not just Google Ads. It’s a way to personalize ad text by dynamically replacing placeholders with contextually relevant content. Realize supports its own version of DKI. In Realize, you can insert dynamic values (like location, device, day of week, etc.) into titles using a macro format like ${city:capitalized}$ or ${platform:uppercase}$ to make ads more personalized. ### Will using DKI automatically improve my Quality Score? Not necessarily. While DKI can improve relevance (which can help CTR), Google Ads’ Quality Score also depends on other factors, like landing page experience and expected click-through rate. So, DKI helps, but it's not a magic bullet. In Realize, there’s no “Quality Score” equivalent like in Google Ads, but using DKI can boost CTR and engagement by making your titles more relevant. Realize’s help docs recommend DKI to increase relevancy and engagement, but performance still depends on your title, thumbnail, targeting, and bid strategy. --- ### AI Ad Optimization Beyond Rules: Why Traditional Automation is Your Performance Bottleneck URL: https://www.taboola.com/marketing-hub/ai-ad-optimization-why-automation-became-bottleneck/ Last Modified: 2026-05-11 19:07:36 Rules-based automation in performance marketing was a step forward, but managing a matrix of hundreds of “if/then” triggers is a complexity trap. The more rules you add, the more time you spend auditing conflicting automations instead of actually improving performance. The advertisers pulling ahead are moving to agentic artificial intelligence (AI): autonomous systems that continuously read live signals, make decisions, and execute in real time. If you're still relying on static automation to manage your campaigns, your legacy systems have become your performance bottleneck. Here's what the shift looks like, and how to get ahead of it. ## What is AI Ad Optimization (and Why Rules Aren't Enough)? AI ad optimization uses machine learning to continuously improve campaign performance. To hit goals like lower cost per acquisition (CPA) and higher return on ad spend (ROAS), in real time, it adjusts: - Bids. - Budgets. - Audiences. - Creative. Most advertisers have already experienced a version of this through automated rules, smart bidding, and scheduled budget adjustments. These tools are useful, but they have a ceiling: Traditional ad automation runs on simple “if this, then that” logic. Bid cap hit? Pause the campaign. Click-through rate (CTR) drops? Rotate the creative. It’s reactive by design, which means it can only respond to conditions you’ve already anticipated. It can’t learn or adapt on the fly. Agentic AI advertising works differently. Instead of waiting for a trigger, it continuously analyzes signals across your full campaign, makes decisions, and executes in real time, essentially employing return on ad spend (ROAS) AI. At scale, that distinction matters. When your strategy spans hundreds of audience, creative, and environment combinations, static rules can’t keep up. Budgets drift, winning strategies get starved of spend, and opportunities close before anyone can react. That’s the ad performance bottleneck most advertisers don’t see coming until it’s already hurting their results. ## The Evolution: From Static Automation to Agentic AI For years, automated ad management followed the same basic pattern: A campaign goes live, data builds up, a human reviews performance, tweaks the settings, and the cycle repeats. Automation helped speed up parts of that loop, but the underlying logic stayed the same: react to yesterday’s data, within parameters someone set in advance. That model made sense when campaigns were simpler, but it doesn’t scale anymore. Agentic AI takes a different approach. Instead of responding to historical data within fixed rules, it’s goal-oriented. You define the outcome you want, whether that’s a target CPA or a ROAS threshold, and the system works backwards from there. It continuously tests strategies, reads live signals, reallocates budget, and adjusts creative delivery without needing human input at every step. Traditional automation asks, “Did this condition occur?” Agentic AI asks, “What’s the best decision right now?” One is a checklist. The other is a campaign strategist that never sleeps. ## Escaping the “Complexity Trap” of Manual Matrix Management A modern campaign isn’t a single thing, it’s a strategy built from overlapping elements: audience targeting, creative format, device, placement environment, and bidding approach. On the open web, those variables combine to create hundreds of distinct strategy combinations for a single objective. Add one new creative format, and that number jumps. Add a new bidding dimension, and you’re looking at thousands. More combinations should mean more opportunity, but in practice, it creates a management problem that rules-based systems make worse. Every new rule you add interacts with the ones already in place. Budget caps clash with bid rules. Audience exclusions overlap. A rule that made sense three weeks ago is now quietly throttling your best performer. Soon, you’re spending more time auditing conflicting automations than actually improving campaigns. That’s the complexity trap. Real-time campaign optimization cuts through it. Instead of stacking rules on top of rules, machine learning continuously evaluates the full matrix, identifies what’s working, and automatically shifts resources there. The system handles the complexity so your team can focus on strategy. ## The Speed Advantage: Real-Time vs. Reactive Data Analysis Traditional campaign optimization runs on a delay. Data is collected, reported, reviewed, and then acted on, often hours or days after the fact. Even well-configured automation operates this way, executing rules against historical snapshots rather than what’s happening right now. The trouble is, the ad marketplace doesn’t work on that timeline: audience behavior shifts throughout the day; publisher inventory fluctuates; competitor bids move; a placement that drove strong ROAS this morning may be underperforming by afternoon, and a new opportunity may have opened up somewhere else entirely. Agentic AI operates on live data streams, not yesterday’s reports. It continuously processes signals across audiences, creatives, placements, and bids, identifying patterns and making adjustments in real time. The kind of micro-optimizations that would take a human analyst hours to spot and act on happen automatically, at a speed and frequency no manual process can match. For advertisers, that speed compounds over time. Every real-time adjustment is a conversion that wouldn’t have happened under a slower system. At scale, those gains add up fast. ## 3 Core Capabilities of an Agentic AI Optimization Engine Rules-based automation handles individual tasks. An agentic system connects them. Here’s what that looks like across the three areas that drive the most performance impact. ### 1. Autonomous Budget Reallocation and Bid Management In a manual setup, budget allocation is a periodic decision. A human reviews performance, identifies the stronger strategies, and shifts spend. By the time that happens, the window has often already moved. AI bidding optimization works continuously. The system monitors performance signals across every campaign strategy in real time, moving budget toward what’s working and pulling back from what isn’t, without waiting for a weekly review. Bids adjust dynamically based on live auction conditions, audience quality, and conversion probability, keeping spend efficient even as the marketplace shifts around it. ### 2. Rapid Creative Testing and Iteration Creative fatigue is one of the most common causes of performance decay, and one of the hardest to catch manually. By the time declining CTR shows up in a report, an audience has already been overexposed. Agentic AI identifies fatigue signals early and rotates creative automatically, serving different ad variations to different audience segments based on what’s most likely to convert. Rather than running a handful of creatives and waiting to see what sticks, the system continuously tests and iterates without requiring constant manual input. Some performance platforms take this a step further. Rather than waiting for creative to underperform, they automatically build and optimize campaign elements on an ongoing basis, keeping the creative portfolio fresh and aligned with whatever strategy the decision engine is prioritizing at any given moment. ### 3. Predictive Audience Targeting Demographic targeting is a starting point, not a strategy. Age, location, and interests tell you something about who someone is, but they tell you very little about whether that person is ready to convert right now. Machine learning ad targeting shifts the focus from profile-based assumptions to behavioral intent signals. By analyzing patterns across browsing behavior, content consumption, and past interactions, the system identifies users who are in-market, not just on-demographic. The result is spend that reaches people with genuine purchase intent, not just people who fit a broad description. ## Feeding the Engine: Why Complete Customer Journey Data is Non-Negotiable An agentic AI system is only as good as the data it runs on, and this is where a lot of advertisers undermine their own results. If the only signal you’re feeding back to the platform is the initial click, that’s what the system will optimize for: more clicks. Not more customers or higher order values. The AI will do exactly what you’ve asked, even if you haven’t asked the right question. Full-funnel data changes that. When the system can see what happens after the click, it can optimize for outcomes that actually matter to your business, whether that’s form completions, purchases, repeat visits, or revenue value. That means connecting server-side tracking, syncing conversion events accurately, and passing back revenue data so the AI understands not just who converted, but what that conversion was worth. This is especially important for ROAS-focused campaigns. An agentic system that can distinguish between a high-value customer and a low-value one will allocate budget and adjust bids very differently from one working from click data alone. The setup investment is worth it. The more complete the signal you provide, the more precisely the engine can target, bid, and optimize on your behalf. Garbage in, garbage out has never been more consequential than when an AI is making thousands of decisions a day based on it. This is also where the quality of your platform’s data infrastructure matters. Realize’s direct code-on-page integration across thousands of premium publishers generates proprietary first-party signals that go well beyond standard third-party data. ## The New Role of the Marketer: From Executor to Strategic Coach One of the most common concerns around agentic AI advertising is straightforward: If the system handles optimization automatically, what does that leave for the marketer? The answer is, the parts that require human judgment. In an agentic setup, the media buyer’s role shifts from managing the day-to-day mechanics of campaign execution to setting the strategic conditions the AI operates within, defining goals, establishing guardrails, briefing creative direction, and identifying the audiences that matter most. Think of it less as handing over control and more as moving up a level. The advertisers who get the most from agentic AI aren’t the ones who set it and forget it, they’re the ones who treat it like a high-performance team member: give it clear objectives, feed it good inputs, and hold it accountable to outcomes. The AI handles the execution, while the marketer coaches the strategy. ## Top AI Ad Optimization Platforms Paving the Way Agentic AI advertising isn’t a future concept. Several platforms are already operating with these principles today. ### Meta Advantage+ Meta Advantage+ is the most widely adopted example. It handles audience selection, placement, and creative delivery automatically across Meta’s inventory, using behavioral data to find users most likely to convert. Meta Advantage+ optimization has driven real CPA improvements for many advertisers, though it’s limited to Meta’s ecosystem and offers little transparency into how decisions are made. ### Madgicx Madgicx acts as a 24/7 AI account auditor, continuously scanning performance and surfacing recommendations. It’s particularly strong on creative intelligence, flagging fatiguing ads before performance visibly drops. ### AdRoll AdRoll takes a full-funnel approach, using machine learning ad targeting to coordinate retargeting and prospecting across display, social, and email. Its strength is cross-channel consistency, keeping optimization logic aligned across multiple touchpoints. ### Zocket Zocket focuses on speed, using AI ad creation tools to generate and launch campaigns quickly, with built-in optimization that adjusts delivery based on early performance signals. ### Realize+ Realize+ is an agentic system built specifically for the open web. It continuously decides, executes, and adapts campaign strategies in real time, turning advertiser goals into outcomes without requiring constant human intervention. Think of it as the performance power of PMax and Advantage+, applied to the world’s best premium publishers, with zero platform bias. For advertisers who have hit the ceiling on search and social, Realize+ offers a direct path to scaled, outcome-based performance on the open web, without the complexity trap or the ad tech tax. ## How to Transition Your Campaigns to Agentic AI Today Shifting from rules-based management to an agentic approach doesn’t have to happen overnight. For most teams, it’s a gradual process of letting go of manual controls in the right order. ### Start with campaign consolidation Hyper-granular campaign structures are the enemy of machine learning. When budget is fragmented across dozens of tightly segmented campaigns, no single strategy gets enough data to learn from. Consolidating into fewer, broader campaign groups gives the algorithm the signal volume it needs to make smarter decisions faster. ### Switch to goal-based bidding Move away from manual bid adjustments and toward objective-led strategies like Maximize Conversions or target CPA. This is the most direct way to hand optimization logic over to the system and start seeing what it can do with a clear performance target. ### Feed it better data Before scaling any agentic setup, make sure your conversion tracking is accurate and complete. The system needs full-funnel signals, not just clicks, to optimize for outcomes that actually matter. ### Run experiments before going all-in Test agentic optimization against your existing approach on a portion of budget. Let the data make the case rather than making a wholesale switch based on assumptions. ### Resist the urge to over-manage This is the hardest part for experienced media buyers. Agentic AI needs room to learn, and frequent manual interventions reset that learning. Set clear goals, define your guardrails, and give the system time to perform. ## Key Takeaways The shift from rules-based automation to agentic AI is already underway. The advertisers who adapt now will scale more efficiently and outpace competitors still managing campaigns manually. Here’s what to keep in mind: - Static rules can’t manage the complexity of modern campaign strategy at scale. - Agentic AI redirects the marketer’s role toward strategy, not execution. - Full-funnel data is the foundation — better inputs mean better outcomes. - The transition starts with small steps: consolidate campaigns, set goal-based bidding, and trust the system to learn. ## Frequently Asked Questions (FAQs) ### What is agentic AI in advertising optimization? Agentic AI refers to autonomous systems that work toward a specific advertising goal, like maximizing ROAS or hitting a target CPA, by continuously analyzing live data and making decisions in real time. Unlike traditional automation, it doesn’t need a human to define every rule in advance. It reasons, adapts, and executes on its own, based on what the data is showing right now. ### How is AI ad optimization different from traditional automation? Traditional automation follows static rules. If CTR drops below a threshold, pause the ad. If CPA exceeds a limit, reduce the bid. It only responds to conditions you’ve already anticipated. AI ad optimization uses machine learning to continuously analyze performance data, predict outcomes, and adapt campaigns to shifting market conditions. It doesn’t wait for a trigger. It’s always working, and it gets smarter over time. ### Will AI replace media buyers? No. The role evolves rather than disappears. As agentic AI takes over the day-to-day mechanics of campaign execution, media buyers shift their focus to the things that actually require human judgment: setting business goals, defining guardrails, shaping creative strategy, and making sure the system has the right data to work with. The AI handles the micro-decisions. The marketer sets the direction. ### How does AI improve return on ad spend (ROAS)? AI improves ROAS by doing things manual optimization simply can’t do at speed or scale. It identifies behavioral patterns that aren’t visible in a standard dashboard, shifts budget toward high-intent audiences in real time, and adjusts bids continuously based on conversion probability. On the creative side, it tests and personalizes ad variations automatically, so spend is always weighted toward what’s most likely to convert. Less waste, more return. --- ### How to Optimize Sports Betting Ad Spend: Realize Expert Recommendations URL: https://www.taboola.com/marketing-hub/optimize-sports-betting-ads/ Last Modified: 2026-07-05 11:34:30 The only metric that matters in the sports betting marketing world is performance. But, as every user acquisition lead knows, effecting an efficient transition from high-energy social media campaigns to the open web can be challenging at first. In the real world case being discussed here, a leading sports trading and prediction platform saw its cost per acquisition (CPA) climb above $1,000. The team knew that more budget wouldn’t solve the problem — rather, they required a creative overhaul. That’s why they partnered with Realize, reengineering their stagnant campaign by shifting from static social assets to performance-driven motion and lifestyle creatives. By doing this, they slashed their CPA to a much more sustainable $200 during the heart of football season. Keep reading to see how a Realize-driven framework can help you optimize your spend and surpass your competition on the open web. ## Moving Beyond Static: Why Motion Ads Are Critical for Sports Betting When set loose on the open web, your creative competes not only with other ads, but the latest trade rumors, injury reports, game highlights, weather conditions, matchup analyses, and more. If your ad looks like a digital billboard, users will scroll right past it. To drive performance like an offense focused on a two-minute drill, your creative must interrupt the scroll with the same kinetic energy as a 50-yard slot fade/rub (not a desperate Hail Mary). ### Breaking Through the Learning Phase If an ad doesn’t generate enough early engagement signals, the algorithm starves it of spend before it ever finds its audience. The static images prevalent in the betting space often lack the scroll-stopping power needed to exit this phase quickly. That’s why Realize allows you to make adjustments to underperforming ads quickly. “If we don’t see the spend pick up on new ads, we don’t just wait — we leverage more variance,” says Stephen Hollinshead, senior account manager at Taboola. “The goal is to move beyond the static images audiences ignore. Leaning into motion ads moves the algorithm through the learning process faster to start hitting those efficiency targets.” For a sports analogy, imagine if a new offensive coordinator waited until a week before the season opener to hand out playbooks. Or, if the DC didn’t bother to review the previous season’s tapes to see where the defense needed to shore up its positioning. The team would never get the traction necessary to become a playoff contender. ### The Psychology of Subtle Movement in Betting Performance advertising hits its stride when ads feel like breaking news. By imitating a live ticker or play-by-play update like those found in the NFL app, motion adds a layer of urgency that a static ad can’t replicate. “We’ve seen that even subtle motion can radically change a user’s perception of a betting ad,” says Taboola sales manager Caroline Berardi. “We take those powerful breaking news themes and reengineer them into lifestyle layouts that perform best in a news-reading environment.” ## Precision Placement: Doubling Down on High-Affinity Sports Environments Ad spend optimization is a two-way street (or offense and defense): the creative gets the click, but context prepares the user to convert. Realize experts focus on matching high-impact creative to publisher environments where fans are already primed to place a trade or bet. At midseason, when teams are competing for spots and wildcard berths, a generic news placement might reach a sports fan, but will it catch them in a betting mindset? Potentially not. ### Winning the Home Field Advantage For this client, the Realize team noted that generic news placements had a 3x higher CPA than sports-specific apps like ESPN. “We found our segment,” says Hollinshead. “In the sports betting vertical, doubling down on premium environments like the ESPN app during peak season provides the exact inventory we need. We match the ad creative to the high-energy mindset of a passionate fan already deeply engaged in the game.” ## From Social UGC to Sports Betting Performance on the Open Web Many betting brands lean heavily on their Instagram presence, filling their feed with high-energy user-generated content (UGC). But, a raw video of a tailgating Bills Mafia fan jumping onto a table doesn’t exactly translate into a native ad unit, so those videos fall flat on the open web like a 60-yard kick that doinks off the uprights. Ads that do well on Instagram may fail elsewhere because they rely on a high-energy, scroll-stopping social context that contrasts sharply with the editorial, reading-focused environment of publisher websites. Nativizing these hooks requires adapting content from intrusive, loud, overly polished (or too raw) formats to quieter, more context-driven, and user-initiated narratives. The Realize team has its own trick plays to use, nativizing social hooks for a performance marketing hook. ### Reengineering the Betting Hook An Instagram video or other successful social media post might just need a little tweaking to resonate with your target audience. You just have to extract its DNA: If your top social performer is a video about “Legal Betting in 50 States,” for example, use the headline as your anchor on the open web, but adjust the visual delivery to fit the editorial flow. “I’ve been building off the best creatives with the strongest hooks on social,” explains Berardi. “We take that core breaking news energy and adapt it for the specific image and text flows we know convert successfully on the open web. In other words, modify what works on Instagram and make it feel like part of the sports fan’s news-reading experience.” ## Key Takeaways If you want to scale a sports betting app on the open web, you need a specialized creative pipeline. By partnering with the Realize experts, our client transformed a $1,000 CPA into a scaling success story, by replacing its one-size-fits-all social assets with performance-driven motion and lifestyle imagery. Score a touchdown, win the game (and engineer every creative variant for return on investment) by: - Using motion to bypass the learning phase and signal urgency. - Targeting premium sports publishers like ESPN to attract fans during peak mid-season. - Repurposing social hooks into lifestyle-driven, editorial-style layouts. ## Frequently Asked Questions (FAQs) ### Why is my sports betting CPA higher on the open web than on traditional social platforms? Users scrolling the open web are operating in consumption mode, reading analyses, checking scores, or bemoaning the injury of a key player. Standard social banners often look like ads (and are easy to ignore). High CPAs on the open web often happen because loud social media ads clash with the analytical mindset of users reading the sports news (and hunting for stats). To lower costs, shift ads from disruptive flashes to contextual, lifestyle-driven creative that blends with premium editorial content. By matching the aesthetic of premium publishers, these ads naturally increase click-through rate (CTR) and lower your CPA. The result? You build deeper trust and capture high-intent users already in a research-and-bet mindset and committed to a high-intent action. ### How do I know which sports-themed ad creative will perform best during peak seasons? If you’ve traditionally run slow, manual A/B tests on every headline and image combination, you lose that window of opportunity during a short sports season, like football. To identify top-performing creatives during these peaks, use predictive vertical insights and rapid multivariate testing to learn whether fans respond to high-action lifestyle shots or data-heavy odds. By using “breaking news” overlays and monitoring real-time publisher trends, you can skip the discovery phase and scale winning creatives while the season is still hot. This approach lets you pivot fast based on real-time data, ensuring the creatives in your performance campaigns on the open web remain as dynamic as the season itself. ### Can I repurpose my sports-focused UGC for the open web? Yes, but here’s the caveat: Raw, handheld social video often feels cheap and amateurish or out of place on a high-authority news or sports site. The last thing you want is to damage your credibility or the trust required for someone to make a financial transaction, like betting. Repurposing content for the open web should extract the successful DNA of UGC — the hook, headline, player stat, or confetti celebration — and then pivot. By repackaging those elements into high-impact lifestyle stills and snappy motion ads, you provide the authentic social proof needed to build trust in a high-scrutiny environment. Now, like a well-balanced offense and defense, you have the best of both worlds: the authenticity that drives social engagement and the professional polish that preserves brand authority. --- ### 3 Strategies to Navigate Finance and Crypto Ad Policies (and Bypass CRT Rejections) URL: https://www.taboola.com/marketing-hub/align-with-finance-ads-policies/ Last Modified: 2026-04-28 09:37:52 Running finance and cryptocurrency campaigns can be one of the most lucrative opportunities in performance marketing, but it also comes with a potentially significant obstacle — the Content Review Team (CRT). Instant rejections can be triggered by a specific crypto token being featured in your ad, or an official government seal appearing in a campaign. The good news, though, is that compliance and profitability can work hand-in-hand. With the right structural and creative adjustments, you can work within platform guidelines without stalling your campaign’s overall effectiveness. ## Why the CRT Rejects Single Tickers and Government Imagery Finance and crypto ad policies across major networks like Google, Meta, and Realize are built around two core concerns: market manipulation and consumer deception. When a CRT sees an ad pushing single-stock or crypto assets, the immediate association is with pump-and-dump schemes, which are coordinated efforts to artificially inflate the asset’s price through misleading promotional content. Even if your intentions are good, an ad leading with one specific ticker looks like unverified financial advice at best, and market manipulation at worst. Government imagery in ads can also trigger a separate but equally serious concern in impersonation. Ads featuring official seals, logos from agencies like the IRS, or landmarks strongly associated with government authority are treated as potential attempts to deceive consumers into believing that the content is officially endorsed. This kind of potentially misleading framing is explicitly prohibited across all major ad networks. Because finance content falls into the YMYL (Your Money Your Life) guidelines, platforms apply an especially high bar to anything that could mislead consumers about financial decisions or the credibility of the source. Repeated violations risk escalating beyond individual ad rejections toward ad account bans that are far harder to have removed. ## 3 Ways to Bypass CRT Rejections with Financial and Crypto Ads ### Strategy 1: The Educational Listicle Approach The most effective way to promote a specific stock or crypto asset without triggering single-stock ticker restrictions is to reframe the content as education rather than promotion. This is the core of the educational listicle approach. Rather than creating a campaign around one asset, you build it within a curated list such as “5 Crypto Assets Analysts are Watching This Quarter.” Your primary target asset is included, but it sits alongside several other credible securities or well-established cryptocurrencies. This works well because it genuinely shifts the editorial framing of the asset. A listicle covering multiple assets with context and market analysis is more educational than a single-asset pitch. It distributes the promotional weight so no one asset is being pushed more than the others in a way that triggers Content Review Team (CRT) rejections. These also tend to perform better with audiences who engage more with comparative content than direct calls to action. ### Strategy 2: Using an Email Gate Pre-Lander for Specific Assets When restricted content is specific enough that it can’t be diluted into a listicle, the email gate pre-lander is your most reliable crypto ad compliance tool. The structure here is straightforward enough: Your ad and its landing page stay entirely compliant, promoting a broad financial concept. To access the specific asset recommendations, visitors must opt in with their email address on the landing page. The ticker or crypto name is then delivered via email directly to the consumer and completely outside the ad network’s platform. This works because the ad network only reviews what’s on the pre-lander. As long as that page contains no single-ticker promotions or restricted content, it clears CRT review. The email follow-up works under its own regulatory framework: CAN-SPAM in the United States and GDPR in Europe. This way, everything you want to actively promote falls outside the ad platform’s content policies. This structure is also a textbook example of funnel segmentation, where you deliberately split your marketing funnel into distinctive stages, each tailored to a different audience’s engagement level and the compliance environment. The top of the funnel (your ad and pre-lander) is built for a broad reach and regulatory safety. The bottom of your funnel (your email sequence) is built for specificity and conversions. Keeping these stages separate isn’t only a compliance workaround, but funnel architecture that improves targeting precision across your audience. Beyond compliance, the opt-in also creates a qualified list of high-intent subscribers that you can use beyond your initial campaign as part of ongoing marketing efforts. ### Strategy 3: Substituting Restricted Visuals with Thematic Imagery The visual layer of CRT rejections is often underestimated. Advertisers who carefully assemble their copy for compliance sometimes still get flagged because their creative assets include imagery that triggers government impersonation or deceptive practice violations. The fix is thematic imagery substitution, where you replace the imagery tied to specific, identifiable government entities with visuals that convey the same conceptual message, but without the compliance risks. In practice, this could look like: - Government buildings: Swap recognizable federal buildings for generic neoclassical or civic architecture with columns and stone facades, that communicate “official” without being tied to a specific government body. - Agency logos and seals: Replace IRS, SEC, or Federal Reserve seals with generic financial iconography like scales, abstract graphs, or stylized dollar signs, to signal regulatory themes without impersonating an agency. - Authority figures: Ads implying endorsement from real politicians or regulators are almost universally rejected. Generic professional imagery like a suited individual reviewing a chart conveys a sense of authority, without attaching a protected individual. The underlying principle here is that compliance reviewers, both human and algorithmic, are pattern-matching for specific, identifiable signals of impersonation. Generic thematic imagery that conveys the same concept rarely triggers those patterns. You’re not changing what the ad communicates, just the visual shortcut used to convey that message. ## Key Takeaways Financial ad approval in the finance and crypto space doesn’t require compromising campaign effectiveness. Instead, it requires an understanding of where compliance lines are drawn and building your creative around that. The three strategies outlined here address the most common rejection triggers from different angles. The educational listicle reframes single-asset promotion as a multi-asset analysis, reducing pump-and-dump signals. The email gate removes restricted content from the ad network’s review environment entirely, routing it into a private, high-intent email funnel. Thematic image substitution eliminates visual patterns that trigger government impersonation red flags, replacing specific imagery with compliant equivalents. Used individually or together, these approaches allow finance and crypto advertisers to maintain strong return on investment (ROI) while staying on the compliant side of platform policies and far from the consequences of repeated violations. ## Frequently Asked Questions (FAQs) ### Why do ad platforms reject ads featuring single-stock or crypto tickers? Ad networks restrict single-stock tickers promotions to prevent market manipulation, pump-and-dump schemes, and unverified financial advice. Because finance falls under YMYL content categories, platforms apply heightened scrutiny, and aggressively pushing one specific asset raises immediate flags for both automated systems and human reviewers. ### How does an email gate help with crypto ad compliance? An email gate pre-lander keeps restricted content off the page that ad networks review, placing it instead inside a private email funnel triggered by an explicit opt-in. This means the ad and landing page remain fully compliant while still delivering targeted recommendations to interested subscribers. ### What happens if I use a government seal, like the IRS, in a financial ad? Using official agency logos or seals will typically trigger an immediate rejection for deceptive practices or government impersonation, and violations that appear explicitly in the restricted content policies of Google, Meta, and most major ad networks. Repeated violations can escalate toward ad account bans. ### What is a thematic imagery alternative for government buildings? Thematic imagery substitution means swapping specific, identifiable government visuals, like official seals, recognizable federal buildings, or real political figures, for generic equivalents that carry the same conceptual meaning, such as columns and stone facades instead of the Capitol Building, or abstract financial iconography instead of an agency seal. --- ### How to Choose the Right Bidding Strategy for Your Campaign Goal URL: https://www.taboola.com/marketing-hub/choose-the-right-bidding-strategy/ Last Modified: 2026-04-28 08:55:21 In the fast-paced world of digital advertising, the difference between a campaign that merely “spends” and one that “scales” will often come down to a single, invisible factor: the bidding strategy. As we move away from the era of manual lever-pulling and into a new age of strategic orchestration, the most successful advertisers aren't just media buyers, but technical architects who understand how to feed the machine learning algorithms that power modern ad auctions. To help us navigate this complex landscape, we turned to Lauren Wint, advertising account manager for Realize, who has spent years optimizing performance campaigns across some of the most competitive verticals in the industry. In this comprehensive guide, we’ll explore five battle-tested bidding strategies to help you stabilize your CPA, maximize your ROAS, and unlock the scale your brand deserves. ## 5 Ways to Effectively Choose the Right Bidding Strategy for Your Performance Goals ### 1. The “Broad-to-Narrow” Launch: Trusting the Machine When launching a brand within the e-commerce and apparel vertical, particularly one that has no historical platform data, the instinct for many marketers is to target too specifically, eager to pick every interest, every demographic, and every specific publisher site on day one. According to Wint, though, this is the quickest way to starve a campaign of the data it needs to succeed. Instead, the most effective strategy is a “Broad-to-Narrow” approach. By launching “Run of Network” — essentially a broad reach strategy — you allow the platform’s algorithm to discover where your audience actually lives. During this initial phase, the system gathers a clean signal, identifying high-performing pockets that a human analyst might never have predicted. “You have to resist the urge to over-engineer on day one,” says Wint. “We start broad to gather the data, then we use those learnings to dictate where to take the campaign next. If you restrict the algorithm too early, you’re essentially flying blind without a map.” By giving the algorithm the freedom to explore the vast network of over 11,000 publishers, you earn the data required for long-term scaling. Once the system identifies which users are clicking and converting, you can begin to siphon spend into those specific, proven pockets. ### 2. Stabilizing the Surge: Transitioning from Volume to Target CPA For financial services and lead gen verticals, costs are notoriously volatile. One week you’re hitting your lead goals, and the next, your cost per lead (CPL) has crept up by 20%. This is the primary use case for shifting from a volume-focused bidding strategy to a Target CPA (tCPA) model. While automated “Maximize Conversions” strategies are excellent for driving rapid growth and capturing as much of the auction as possible, they don’t always respect strict margin requirements. Once a campaign has established enough baseline data, transitioning to a target-based bid acts as a vital stabilizer: it tells the algorithm that growth is great, but only at this particular price point. “Target CPA is a powerful tool, but it can be a double-edged sword,” warns Wint. “If you set it too restrictively before the system has ‘learned’ the auction, you’ll see your scale disappear overnight. It’s about finding that mathematical sweet spot where the algorithm has enough room to breathe while still hitting your margin.” The key is to wait until the algorithm has learned the lower price point. Wint recommends having a healthy daily budget — ideally 10-15x your target CPA — to ensure the transition doesn’t result in a total loss of delivery. ### 3. The Signal and the Noise: Funnel-Based Event Optimization For high-intent offers within the investment and fintech vertical, not all conversions are created equal: A lead might simply be someone who entered an email address to see a stock tip, whereas a purchase or a verified SMS lead represents a user with true intent. Optimizing solely for a basic lead capture form, then, is often a “noisy” signal. If you tell the bidding engine to find as many email addresses as possible for $5, it will find them, but they might turn out to be the lowest quality users on the web. The strategy here is to move optimization further down the funnel: By focusing the bidding engine on deeper actions, you ensure that your spend is allocated toward users with actual lifetime value. “When you have a deep funnel, the algorithm can get confused if you’re asking it to find everything at once,” explains Wint. “We look at the relationship along the funnel to see where the real value is. Sometimes, removing noisier top-of-funnel events from the total conversion count — once you have sufficient insight from high-value downstream events — actually sharpens the bidding engine’s focus on what makes you money.” This shift in focus allows the machine learning models to ignore the window shoppers and zero in on the investors. ### 4. The Power of Isolation: Dominating High-Value Inventory In the highly competitive home improvement vertical — specifically for high-ticket services like bathroom remodeling — the auction can become incredibly expensive. Average CPCs often skyrocket as competitors bid for the same premium placements. However, savvy advertisers look for outliers. Occasionally, certain top-tier news feeds or premium mobile operating systems offer significantly lower costs compared to the broader market. When you identify these high-performing placements, you shouldn’t leave them in the general campaign pool, where their performance might be averaged out. The winning move in cases like this is High-Value Inventory Isolation, i.e., the grouping together of multiple high-performance publishers. By siloing these placements into their own dedicated campaigns, you can allocate a specific budget to blow them out, ensuring you capture 100% of that efficient inventory. (It’s important to note that your priority consideration here should still be supply diversity and reach.) “When we find a placement pulling a sub-$10 CPA while the rest of the market is at $30, we don’t just leave it in the general pool, we isolate it,” says Wint. “By siloing high-value inventory, you can dominate that specific auction without your average CPC being dragged up by less efficient placements.” ### 5. The Silent Manager: Implementing Systematic Custom Rules During volatile high seasons — think tax season, or insurance enrollment within the personal finance and insurance vertical — market conditions change by the hour. Competitors flood the market with cash, and a publisher that worked yesterday might become prohibitively expensive today. Human managers, no matter how skilled, cannot react fast enough to these micro-shifts over thousands of publishers. This is where systematic custom rules come into play: By setting automated guardrails — such as automatically blocking publishers that hit a certain spend threshold without a conversion, or pausing ads with a low click-through rate (CTR) — you can handle the dirty work of cost control automatically. “I'm a big believer in systematic automation that keeps running while you’re sleeping,” says Wint. “By setting custom rules as a baseline per account, we can simulate the control of manual bidding without the manual labor. It allows us to be more scientific about which metrics lead the charge, whether that’s CTR, CVR, or pure CPA.” ## Key Takeaways As Lauren Wint has demonstrated through these five strategies, modern performance marketing is no longer about “set it and forget it.” Rather, it’s a discipline of constant refinement, data hygiene, and technical strategy. Whether you’re launching a new apparel brand or scaling a complex financial funnel, the principles remain the same: - Give the machine enough data to learn. - Isolate the winners to protect your margins. - Use automated guardrails to maintain efficiency at scale. By adopting the mindset of a technical architect, you can move beyond simply buying ads and start engineering a sustainable growth engine for your business. ## Frequently Asked Questions (FAQs) ### How does machine learning improve ad bidding? Machine learning processes millions of data points across the network (including user intent, placement context, and historical performance) at a speed impossible for people. It adjusts bids in real-time for every individual auction, ensuring you only pay the right price for a user likely to convert. This eliminates the guesswork of manual bidding and allows for much higher precision in hitting CPA targets. ### What is the checklist for setting up a new performance campaign bid strategy? Before you launch, ensure you have the following in place: - Identify Your Seed: Ensure you have enough data. A baseline of 100 conversions in a 30-day window is typically required to fuel predictive models. - Budget Buffer: Set daily budgets at 10-15x your target CPA so the algorithm has enough fuel to test the auction. - Creative Refresh: Prepare at least 6-8 ads per campaign to avoid creative fatigue and give the algorithm variety to test. - Tracking Audit: Verify that your Server-to-Server (S2S) connections or pixels are firing accurately to provide the machine with a clean signal. ### What are the key metrics for measuring performance campaign success? While CPA and ROAS are the ultimate goals, leading indicators are essential for long-term health. Wint emphasizes looking at click-through rate (CTR) and average CPC. A high CTR indicates that your creative is resonating with the audience, while the ad auction rewards this engagement by lowering your CPC, which in turn drives down your overall acquisition costs. --- ### How to Measure ROI on CTV Campaigns in Performance Marketing URL: https://www.taboola.com/marketing-hub/measure-roi-on-ctv-campaigns/ Last Modified: 2026-04-27 13:16:05 Performance advertisers have spent years avoiding TV for a simple reason: it didn’t close the loop. Search gives you clicks, social gives you pixel-tracked conversions — regular TV gives you reach and a prayer. Connected TV (CTV) gives advertisers a way to treat TV more like a performance channel. Nearly 90% of US households now own at least one internet-connected TV device, and US CTV ad spend is projected to hit $37.95 billion this year. With that scale has come real measurement infrastructure. For brands building D2C TV advertising strategies, the pressure is to connect CTV exposure to conversions, revenue, and return on investment (ROI). Here are four concrete strategies for measuring hard ROI and attributing bottom-of-funnel conversions to your CTV spend. ## Implement Cross-Device Identity Resolution The fundamental challenge with CTV is the “clickless” problem. A viewer sees your ad on a 65-inch screen but can’t click it. If a conversion happens, it usually occurs later on a phone, tablet, or laptop. Reliable CTV attribution models must connect those events across devices, which is why cross-device tracking CTV efforts are so central to measuring CTV ROI. Cross-device identity resolution bridges that gap. An identity graph links a CTV impression, typically tied to a household IP address, to a conversion event on another device through two primary methods: - Deterministic data uses confirmed signals, like a user logged into the same app across devices. It’s accurate, but limited in scale. - Probabilistic data infers matches using shared IPs, timing, location, and behavioral patterns to expand scale. Most identity resolution marketing solutions use a blend of both. Cross-device identity resolution is what makes big-screen exposure measurable at the lower end of the funnel. It’s also why many brands rely on mobile measurement partners (MMPs) or specialized measurement partners, rather than relying solely on standard web analytics. ## Run Incrementality Tests (The Gold Standard) Cross-device attribution tells you that a conversion happened after a CTV impression. Incrementality testing tells you whether the ad actually caused it. That difference is important because attribution shows correlation, while incrementality gets closer to causation. The basic setup is straightforward. Split your target audience into two groups: The test group sees your CTV ads, while the holdout group, or control, does not (either through suppressed delivery or ghost bidding, where bids are placed but intentionally not won). To produce a reliable result, the test needs a clean control group and enough time to capture the full conversion window. After the campaign, compare conversion rates across the two groups. The difference is your incremental lift. From there, you can calculate incremental ROAS, i.e., the revenue generated only because of the ad, divided by spend. That’s very different from standard ROAS, which counts all revenue against spend, whether or not those customers would have converted anyway. In CTV, standard ROAS often overstates performance because it includes all post-exposure revenue, not just the conversions the ad actually drove. To measure ROI accurately, CTV incrementality testing is the clearest way to separate influence from true lift. ## Adopt a Multi-Touch Attribution (MTA) Model Last-click attribution doesn’t work for CTV. If a customer sees your streaming ad on Tuesday, searches for your brand on Thursday, and buys on Friday, last-click gives all the credit to the search click and none to the CTV impression that helped start the journey. These blind spots are becoming harder to ignore as budgets span both digital and traditional channels. Yet, only 32% of global marketers currently measure media spending holistically across those environments. Multi-touch attribution distributes credit across all contributing touchpoints. Within that framework, view-through attribution allows CTV impressions to be counted even without a click. For many D2C brands, a seven- or 14-day view-through attribution window is a good starting point because it captures the natural lag between seeing a TV ad and acting on it. Longer purchase cycles may justify extending to 30 days. Be deliberate about window length: too short, and you will undercount CTV’s contribution; too long, and you will over-attribute conversions that had nothing to do with the impression. For the model to work, CTV impression data must be integrated into the same attribution system as your other channels, so it can receive credit for assisted conversions. This is where CTV measurement partners play an important role, helping bring impression data into the broader attribution framework. ## Deploy Direct Response “Pixel” Mechanics Not every team has the bandwidth for identity graphs and full incrementality infrastructure. There’s a simpler approach that’s been around since early direct-response TV (DRTV), though, and it still works. Three tactical methods can provide more direct attribution: - QR codes displayed on screen give viewers a path to act while the ad is running. A scan creates a clean, direct signal: this person, this ad, this moment. - Vanity URLs — short, campaign-specific addresses like “yourbrand.com/tv” — capture intent from viewers who don’t scan but remember the URL. Visits to that page provide a much stronger signal that the response came from your CTV campaign. - Unique promo codes work especially well for e-commerce. A code tied to a specific campaign, such as “TV30,” can link checkout conversions back to the ad without relying on identity resolution software. These methods have limits — QR codes require a phone nearby, vanity URLs depend on viewer memory — but their biggest advantage is providing direct evidence. They connect exposure with action, without depending entirely on probabilistic matching or longer attribution windows. They offer a practical starting point for connected TV conversion tracking and can run alongside more sophisticated measurement approaches as your stack matures. ## Key Takeaways CTV is now a measurable performance channel, not just an awareness play. The infrastructure is in place: identity resolution, incrementality testing, multi-touch attribution, and direct response tactics are proven ways to measure CTV ROI. The key is recognizing that CTV measurement looks different from search and social. It relies on household attribution instead of individual clicks, view-through windows instead of last-click models, and incremental ROAS instead of raw return. For D2C and lead-gen brands, CTV is a far more accountable part of performance TV marketing than traditional TV ever was. Start with one method, validate it, and build from there. Brands that develop those habits early will be in a much stronger position as the channel scales. ## Frequently Asked Questions (FAQs) ### What is the best attribution window for CTV campaigns? Most D2C brands should consider starting with a seven- or 14-day view-through attribution window. It accounts for the lag between seeing a TV ad and visiting the site on another device. Brands with longer purchase cycles may extend to 30 days. ### Can I track CTV conversions in Google Analytics? Not directly. CTV is impression-based, so Google Analytics has no native way to attribute that traffic. Visitors driven by a streaming ad often show up as Direct or Organic Search, leaving CTV with no credit. To track it properly, you need a specialized CTV measurement partner or a direct-response mechanic, like a vanity URL or QR code. ### What is the difference between ROAS and incremental ROAS on CTV? ROAS measures total revenue divided by ad spend. Incremental ROAS measures only the revenue that would not have happened without the ad, usually by comparing an exposed audience against a holdout control group. On CTV, that difference can be significant because broad reach often includes people who would have converted anyway. --- ### How the Modern Performance Marketer Can Maximize Time Efficiently URL: https://www.taboola.com/marketing-hub/modern-performance-advertising-maximize-efficiency/ Last Modified: 2026-04-28 09:26:21 Performance marketing has entered a new phase, one where time rather than tactics separates brands that scale efficiently from those that stall. To discuss how advertisers can best allocate their time, energy, and internal resources in this shifting landscape, we sat down with Nadim Batista-Kuttab of Xevio, one of the world’s largest native advertisers. Nadim’s team manages millions in daily open web spend, giving them a unique vantage point into what actually drives growth, stability, and long-term profitability at scale. Today, the marketers who win are the ones who invest their time where it creates leverage. Platforms like Realize automate much of that mechanical labor, freeing performance teams to focus on creative iteration, funnel optimization, and lifetime-value growth. This shift has fundamentally redefined what “performance marketing work” should look like, and why the highest ROI comes from redirecting time away from platform manipulation and toward improving the user experience. ## The Shift in Focus: From Platform Settings to User Experience Ten years ago, scaling advertising required a relentless commitment to manual intervention. Marketers would toggle bids on individual sites dozens, sometimes hundreds, of times per day. Every hour brought new CPCs to adjust, new publisher patterns to evaluate, new pacing issues to resolve. Growth required constant watch, while stability required constant pressure. As Nadim describes it, “I was micromanaging bids on a site level hourly — thousands of adjustments a day just to squeeze out a couple percent more performance.” Since then, platforms like Realize have evolved from demand-side engines into predictive systems capable of handling auction-level decisions in real time. Automated bidding strategies like max conversion and target CPA dynamically determine optimal CPCs, freeing advertisers from tactical execution. Tools like Realize add an additional layer of intelligence by providing predictive signals, performance risk indicators, and automated policy clarity, reducing the time advertisers once spent manually troubleshooting campaign issues. “With max conversion and target CPA, you don’t have to set a CPC anymore,” says Nadim. This shift has significant implications: When AI automates the mechanical tasks such as bidding, pacing, and allocation, marketers gain the freedom to focus on the areas that provide compounding gains like the creative asset and the post-click journey. These are areas no algorithm can own entirely, because they rely on a brand’s voice, product value, psychological understanding of the user, and the ability to build persuasive digital experiences. ## The Two Core Leverage Principles As AI takes over low-value tasks, time becomes a strategic asset. The modern performance marketer must now invest their hours where the highest leverage sits: the advertiser’s own ecosystem. From Nadim’s perspective, the most leverage is going to be ad performance and post-click performance. Everything else, like bidding rules, micro-segment targeting, and publisher-level adjustments, should be deprioritized in favor of what directly shapes user behavior. Yes, AI can adjust bids, but it can’t improve your creative resonance. It can allocate budget, but it can’t rewrite your landing page. It can predict performance, but it can’t create emotional connection. ### Ad Performance and CTR The first leverage point is the ad itself — in other words, your creative. Every headline, image, and layout decision logically compounds downstream. A fractional improvement in CTR may not just increase traffic — it lowers effective CPC, increases audience reach, and improves the platform’s ability to learn faster. This is why Xevio never stops testing, even when performance is strong. Performance advertising is a creative-led ecosystem, which means that the algorithm may deliver traffic, but the click comes down to the ad. Nadim shares a simple but powerful operational rule when it comes to testing: “If you’re below five ads on top, always upload more — even iterations of winners.” Why? Because a 0.1% increase in CTR at scale can meaningfully shift the economics of the entire campaign. This is also where Realize provides additional value: Its creative-level attribution signals and predicted performance insights help identify which ads have breakout potential earlier, allowing operators to spend more time iterating on winning themes and less time sifting through raw data. When marketers allocate time here through testing more formats, challenging their assumptions and introducing new emotional angles, they unlock efficiencies the algorithm alone can’t create. ### Post-Click Performance The second leverage point, and often the most underestimated, is everything that happens after the click. This is where the deepest, longest-lasting ROI gains occur. Pre-click optimizations deliver incremental improvements, while post-click optimizations reshape the outcome of the entire model. When Xevio invests time into post-click performance, they focus on: - Content quality and depth. - User intent alignment. - Landing page clarity. - Frictionless form or checkout design. - Higher AOV opportunities. - Long-term LTV improvements. A small improvement in conversion rate can be more impactful than a similar improvement in CTR. As Nadim puts it, “We invest time where we have the most leverage — ad performance and post-click performance.” Improving post-click experience decreases CPA, increases LTV, and strengthens the algorithm’s ability to find high-value users. When Realize predicts which traffic segments are more likely to convert or flags potential drop-off patterns, it helps advertisers focus their time on the parts of the funnel most likely to generate returns. When marketers spend their time rewriting headlines, restructuring product pages, improving form flow, or rethinking the user journey, the downstream impact compounds across every future campaign. ## Optimizing the Post-Click Funnel for LTV Performance marketing used to be evaluated primarily on acquisition metrics: CPA, ROAS, CPC. But, in a world where AI bidding compresses short-term differences, long-term value (LTV) becomes the true differentiator. The post-click funnel is where LTV is created. Unlike traditional media, performance advertising on the open web provides extensive data on user behavior. You don’t have to guess whether landing-page changes improve profitability — instead, you can measure the effects in real time. With advertisers like Xevio seeing over two minutes of average time on page from Taboola content placements, optimizing this long-form engagement becomes one of the most valuable investments a performance team can make. ### Landing Page and Content Great landing pages do three things — provide clarity, create narrative momentum, and support decision-making. Open web audiences differ from social audiences because they arrive with the intent to consume content. They’re ready to read, evaluate, and think, not mindlessly scroll. This creates an opportunity to engage at a deeper level than traditional paid media channels. The goal is to maximize the long dwell time this delivers by structuring content that highlights value early, anticipates questions, demonstrates authority, reduces uncertainty, and makes conversion feel natural. Realize’s analytics help identify which content patterns correlate with higher possible conversion potential. By providing greater visibility into user behavior, it helps advertisers understand where to expand or refine their content flow. ### Lead Forms and Product Pages Forms and product pages represent the final step to conversion, and small improvements here can have a large impact, because at this point, potential revenue can turn into actual, tangible revenue. Performance teams should test areas such as: - Form length: Short vs. long, single step vs. multi-step. - CTA copy: Value-driven vs. action-driven. - Checkout layout: Streamlined vs. information-rich. - Page hierarchy: Benefits first vs. features first. - Trust indicators: Badges, reviews, guarantees. Ultimately, any improvements made come down to: Revenue = ( Traffic × Conversion Rate ) × Average Order Value The marketer’s job is to increase the multiplier effect of conversion rate and AOV. When those numbers shift upward, they increase the ROAS of the entire campaign. Realize contributes here by identifying which traffic segments produce stronger back-end performance, enabling advertisers to align creative, landing pages, and product experiences with the audiences most likely to convert at high value. ## The Importance of External Support and Knowledge Performance marketing may be more automated today, but it is not simpler. The platforms evolve quickly, policies shift, and competitive dynamics vary drastically by market. Nadim’s guidance for brands scaling from six to nine figures is direct: “It’s going to be a little bit biased, but get help. We’ve seen a lot of brands come and go because they didn’t have the right tools or the right people or the right knowledge.” ### A Faster Learning Curve Experienced account managers and specialized agencies see patterns that individual advertisers may never encounter. They understand the saturation point of a location and the realistic daily spend potential. They can also identify the early signs of data degradation, how quickly ad fatigue emerges, and when performance changes are caused by creative vs. competition vs. inventory shifts. The difference between spending $500 per day and $5,000 per day often comes down to whether you scale too quickly in a small geographic market, or misinterpret a temporary signal as structural decline. When you work with experts, the learning curve that creates many of these issues is significantly reduced. ### Validation Performance marketing is full of false positives, like ads that look strong early but collapse at scale, landing pages that convert well but suppress LTV, and audiences that click but don’t buy. Nadim points out that Taboola account managers can access predictive data on CPC, CPM, CTR, and spend ceilings. They can tell you how far a campaign can scale before diminishing returns set in. “Talk to your account manager to see what scale potential they have for a campaign,” he advises. This isn’t just operational advice, but risk mitigation: Brands avoid hundreds of thousands in wasted spend by verifying their assumptions before scaling aggressively. Realize helps here too, offering earlier detection of performance anomalies and clearer visibility into how policy or inventory constraints might affect campaigns. ## Key Takeaways Performance marketing is no longer defined by the advertiser who makes the most bid adjustments and is instead defined by the advertiser who allocates their time to the highest-leverage activities. This is where Realize helps support creative teams, by eliminating guesswork, improving predictive understanding, and reducing the need for manual platform diagnostics. Success is not determined by who spends the most, but by who spends the most intentionally, making time the ultimate performance lever. If you want to understand where your time is best invested, consult your Taboola account manager. They can help you evaluate your growth ceiling, validate your strategic approach, and ensure you’re putting your effort in the places that generate the highest long-term return. --- ### 8 Best Performance Platforms for Bid Optimization for 2026 URL: https://www.taboola.com/marketing-hub/best-performance-platforms-for-bid-optimization/ Last Modified: 2026-04-27 13:10:26 Bidding manually across campaigns at any meaningful scale is a losing game. By the time you’ve adjusted bids in one campaign, conditions in three others have already shifted. That’s why you need a bid optimization platform. These platforms — whether through rule-based automation, machine learning, or full predictive AI — help you make decisions faster and with more signals than any manual process can match. With so many options, choosing the right tools for your business can be overwhelming. This guide covers eight platforms worth considering, including what each one does well, and where it might not meet your specific needs. ## Best Performance Platforms for Bid Optimization Compared Platform Why It’s Essential Core Use Cases and Features Best for Pricing Model 1. Realize Performance-first bidding with integrated outcome-based pricing. Automated bid and budget optimization, performance goals, real-time adjustments. Performance advertisers focused on return on investment (ROI)-linked outcomes Performance-based model; campaigns billed on CPC basis, or CPM for programmatic. 2. Google Ads Largest search and display auction with automated bid strategies. Smart Bidding (Target CPA/ROAS), real-time auction bidding, search, video, and display. All advertisers seeking broad reach. Pay per click (PPC)/CPC/CPA. 3. Microsoft Advertising Secondary search channel with lower CPCs and Google import. Automated bidding, audience targeting, and LinkedIn profile data. Advertisers expanding beyond Google. PPC/CPC/CPA. 4. Optmyzr Rule-based automation and optimization for PPC accounts. Custom automation rules, bidding logic, audit, and reporting. Agencies and experienced PPC teams. From about $209/month. 5. Acquisio AI-powered bid and budget management with machine learning. Auto bidding, campaign automation, predictive analytics. Multi-account PPC managers and agencies. Custom/subscription. 6. Marin Software Cross-channel campaign optimization with AI bidding. Unified dashboard, automated bidding, performance forecasting. Mid-large advertisers and enterprise teams. Custom starts at about $500+/month. 7. Skai Enterprise-grade cross-channel bid and budget automation. Unified cross-channel management, predictive bidding, and budget pacing. Large advertisers and agencies. Custom enterprise pricing. 8. Birch Flexible automation rules for social and search ads. Custom triggers, budget automation, alerts, and reporting. Small and medium-sized businesses (SMBs) and agencies scaling cross-platform. From about $99/month. ### 1. Realize Why it’s essential: Realize is a performance-first advertising platform that utilizes deep learning and predictive AI to automate the bidding process across the open web. It’s designed to replace manual, labor-intensive bid adjustments with an outcome-based model, where the platform’s AI calculates the precise value of every impression in real time. Advertisers use Realize to bridge the gap between their conversion goals and the complex supply landscape of premium publishers, ensuring they remain competitive in auctions without overpaying. In practice, Realize is used to optimize bids toward specific performance KPIs such as CPA (cost per acquisition). By analyzing billions of real-time signals — e.g., device type, time of day, location, operating system — the platform dynamically adjusts bids for each individual user interaction. This allows brands to scale their media spend efficiently, while ensuring that budget is always flowing toward the opportunities with the highest probability of conversion. Showcased features: - Performance Simulator: Allows you to model different budget and bid scenarios to forecast potential conversion outcomes, before you commit any actual spend. - SpendGuard: An automated optimization algorithm that protects your budget by instantly blocking underperforming sites or creatives that don’t meet the efficiency benchmarks. - Maximize Conversions: Automated bidding strategy that prioritizes spend on users with the highest intent. - In-line recommendations: Delivers proactive AI-driven suggestions within the dashboard, to help you adjust your bid caps and targeting to unlock more scale. - Abby (AI performance expert): An integrated assistant that monitors campaign health and provides real-time fixes for delivery issues caused by restrictive bidding. Best for: Realize is best for performance marketers in high-competition industries like financial services, insurance, and e-commerce, who need to scale beyond social media without increasing their manual workload. It offers enterprise-grade automated bidding, handling the complexities of the open web auction environment out of the box. Pricing model: Performance-based model; campaigns billed on CPC basis, or cost per mille (CPM) for programmatic. Pros: - Automatically adjusts bids for every single impression, based on billions of data signals, to maximize conversions. - The Performance Simulator removes the guesswork from scaling, by using algorithmic modeling to demonstrate how bid increases directly drive changes in conversion volume. - SpendGuard ensures your bids are never wasted on non-converting placements. Cons: - The predictive bidding algorithms require an initial “warm-up” period and a baseline of conversion data to reach peak optimization. - The platform is heavily weighted toward automation, which may feel restrictive for traditional media buyers who prefer manual, granular control over every bid. - To optimize for high-value offline or deep-funnel conversions, a technical server-to-server tracking integration is required. ### 2. Google Ads Why it’s essential: Google Ads Smart Bidding is the most widely used automated bidding system in digital advertising. It has access to more auction-time signals than any third-party tool — including device, location, time, search query, audience membership, and browser — and it applies those signals to every single auction in real time. For advertisers whose customers search actively, no other platform can match the combination of intent data and bidding precision that Google brings to that moment. Smart Bidding works by setting bids automatically toward a defined goal: target CPA, target ROAS, maximize conversions, or maximize conversion value. The algorithm learns from conversion history and adjusts in real time, with no bid caps or manual adjustments required once campaigns are properly configured. Showcased features: - Target CPA bidding: Sets bids to get as many conversions as possible at or below your target cost per acquisition. - Target ROAS bidding: Optimizes toward conversion value, not just conversion volume, weighted by your historical revenue data. - Maximize conversions/value: Fully automated strategies that spend toward the most valuable actions within your daily budget. - Enhanced CPC: A hybrid approach that adjusts manual bids up or down based on the likelihood of conversion, for advertisers who want more control. - Bid simulator: Shows estimated performance impact of adjusting your Target CPA or ROAS targets before you commit to changes. Best for: Advertisers with strong conversion tracking in place and sufficient conversion volume for Smart Bidding to learn effectively. Campaigns with fewer than 30 conversions per month will see slower optimization. Best suited for search-intent campaigns where the signal quality of Google’s data has a clear performance advantage. Pricing model: PPC/CPC for search. CPM and Target CPA/ROAS across display, video, and Performance Max. Pros: - Unmatched auction-time signal depth across search, display, video, and shopping. - No minimum spend — accessible at any budget level. - Native integration with Google Analytics and first-party data tools. Cons: - Smart Bidding requires clean conversion data and enough volume to function well. - Performance Max limits visibility into how the budget is being allocated by channel. - CPCs in competitive categories have risen steadily, reducing efficiency for smaller advertisers. ### 3. Microsoft Advertising Why it’s essential: Microsoft Advertising runs on the same automated bidding logic as Google — target CPA, target ROAS, maximize conversions — but operates in a less competitive auction environment. That means the same Smart Bidding strategies often achieve lower CPCs with comparable intent signals, particularly for high-value demographics. Microsoft’s user base skews older, more affluent, and more desktop-heavy, which makes it a strong complement for campaigns where those demographics convert well. Showcased features: - Automated bidding (target CPA/ROAS): The same smart bidding strategies available in Google Ads apply to Bing’s search and audience network. - Enhanced CPC: Adjusts manual bids based on conversion likelihood and is compatible with Google Ads import. - LinkedIn profile targeting: Layer job title, company, and industry data from LinkedIn onto search campaigns for B2B targeting. - Google Ads import: Import existing Google campaigns directly, reducing setup time for advertisers expanding beyond Google. Best for: Advertisers looking to extend search coverage cost-efficiently, and B2B or fleet advertisers who benefit from professional audience targeting layered onto search intent. Pricing model: CPC and CPM auction bidding. Pros: - LinkedIn data integration is unique to this platform for B2B bid layering. - Google Ads import shortens campaign setup significantly. Cons: - Smaller search volume limits the ceiling on lead generation at scale. - Younger demographics are underrepresented in the user base. ### 4. Optmyzr Why it’s essential: Optmyzr sits between the automation of Google’s native Smart Bidding and the complexity of enterprise bid management platforms. It’s built for PPC professionals who want to layer their own logic on top of Smart Bidding — adjusting targets, applying custom rules, and automating the optimization work that happens between campaign launches and major strategy shifts. The Rule Engine is the core differentiator. It lets advertisers build conditional automation that fires based on any combination of performance metrics, external data, time triggers, or business signals. That’s a level of customization that Google’s native tools don’t offer, and that most enterprise platforms require a larger budget to access. Showcased features: - Rule Engine: Build custom bidding automation using if-then logic across any performance metric — CPA, ROAS, impression share — device, or external data sources like customer relationship management system (CRM) feeds. - Smart bidding optimization: Adjusts target CPA and target ROAS at the ad group level to improve bid efficiency without overriding Smart Bidding entirely. - Spend projection: Forecasts monthly budget pacing and flags overspend or underspend before it becomes a problem. - Hour of the week bid adjustments: Identifies top-performing time slots and automates bid adjustments based on historical conversion patterns. - One-click optimizations: Pre-built optimization strategies for common scenarios — pausing low-quality score keywords, adjusting bids for high-converting locations — applied across accounts in seconds. Best for: PPC agencies managing multiple Google Ads and Microsoft Ads accounts simultaneously, and experienced in-house teams that want automation with more control than Smart Bidding alone provides. Pricing model: Tiered subscription starting from approximately $209 per month, scaling with ad spend and account count. Pros: - Rule Engine allows a level of bid customization that native platforms don’t support. - Works across Google, Microsoft, and Amazon Ads from a single interface. - Reduces manual optimization time while keeping the advertiser in control of strategy. Cons: - Steep learning curve — the full feature set takes time to configure effectively. - Less suited for advertisers who want full AI automation, rather than rule-based control. - Cost scales with account complexity, which can get expensive for large agencies. ### 5. Acquisio Why it’s essential: Acquisio’s core product is its Turing AI — a machine learning engine built specifically for bid and budget management that’s been running and improving since 2012. It operates at high frequency, analyzing campaign data multiple times per day and making micro-adjustments to bids and budgets that manual management can’t match at scale. The platform is built for agencies managing large numbers of PPC accounts across Google and Microsoft. Its strength is automating the operational layer of bid management — the constant small adjustments that consume hours of manual work — so account managers can focus on strategy, rather than execution. Showcased features: - Turing AI: Acquisio’s proprietary machine learning engine runs more than 30 algorithms simultaneously to optimize bids, budgets, and pacing across campaigns throughout the day. - Budget distribution: Automatically reallocates budget across campaigns based on real-time performance, preventing overspend on underperformers. - KPI builder: Set custom performance targets at any level and let the platform optimize spend against them. - Unified account editing: Audit and edit multiple client accounts simultaneously from a single dashboard. - Predictive analytics: Forecasts campaign performance based on historical patterns and machine learning models, with no minimum historical data requirement. Best for: Digital agencies and local search engine marketing (SEM) resellers managing high volumes of accounts across Google and Microsoft Ads, who need 24/7 bid optimization without proportional headcount increases. Pricing model: Custom and subscription pricing; requires a sales conversation for current rates. Pros: - Turing AI has a long track record, with mature algorithms built on over a decade of PPC data. - High-frequency optimization catches performance changes faster than manual processes. - White-label reporting makes it practical for agencies presenting results to clients. Cons: - The platform interface has a learning curve, and onboarding can be slow. - Some user reviews note that overly aggressive bid automation can restrict impression volume. - Less relevant for advertisers who want transparency into every bid decision. ### 6. Marin Software Why it’s essential: MarinOne is built for cross-channel advertisers who need bidding that accounts for more than what Google or Meta sees in their own data. The platform’s Ascend suite uses machine learning to optimize bids using over 75 signals, including CRM data, offline conversions, seasonality, and third-party revenue signals that publisher-native bidding systems can’t access. The key differentiator is full-funnel bidding. Marin connects bid optimization to downstream events — not just the lead, but the closed deal — by integrating directly with CRM systems. For advertisers with long sales cycles or complex revenue attribution, that’s a meaningful advantage over platforms that optimize only to the first conversion event. Showcased features: - MarinOne bidding (Ascend): Incorporates more than 75 signals (including offline conversions, CRM data, seasonality, and competitive signals) to calculate optimal bids across channels. - Full funnel bidding: Optimizes bids toward downstream CRM events — qualified leads, opportunities, or closed revenue — not just initial form submissions. - Dynamic spend allocation: Machine learning reallocates budget across campaigns and channels based on marginal return, rather than just current performance. - Promo calendar bidding: Analyzes historical performance during sales periods and automatically adjusts bids for upcoming promotions. - Forecasting: Models the relationship between spend levels and predicted conversions or revenue, enabling informed budget planning. Best for: Mid-to-large advertisers running cross-channel campaigns with meaningful offline conversion data, particularly those in financial services, retail, or B2B with longer sales cycles. Pricing model: Custom pricing, typically starting at $500 per month. Pros: - Full funnel CRM integration allows bid optimization toward actual revenue, not just leads. - Transparent bidding calculations — you can see how bid decisions are made. - Strong track record for accounts with high keyword volume and complex attribution needs. Cons: - MarinOne’s interface has been consistently cited as less intuitive than the legacy platform. - The bidding algorithm can lag when adapting to rapid platform changes from Google or Microsoft. - High cost relative to feature value for smaller advertisers. ### 7. Skai Why it’s essential: Skai (formerly Kenshoo) is designed for enterprise advertisers and agencies managing campaigns across search, social, retail media, and app channels simultaneously. Its strength is unifying data from more than 100 publishers into a single interface and applying predictive bidding logic across all of them from one place. The platform’s Celeste AI assistant — built on Amazon Bedrock agents — allows users to interact with campaign data through natural language, surfacing insights and recommendations without navigating complex dashboards. For large teams managing multi-million dollar budgets across channels, that kind of analytical accessibility reduces the time between data and decision. Showcased features: - Cross-channel budget pacing: Monitors and adjusts spend allocation across search, social, and retail media to hit targets across all channels simultaneously. - Predictive bidding: Applies machine learning to real-time performance signals for automated bid adjustments across Google, Microsoft, Amazon, and social platforms. - Celeste AI assistant: A generative AI analytics agent that answers natural language questions about campaign performance, surfaces anomalies, and recommends optimization actions. - Retail media integration: Connects Amazon, Walmart, and other retail network campaigns with search and social data for full-funnel retail performance management. - Automated actions: Set conditional rules to trigger bid changes, budget adjustments, or campaign pauses based on any combination of performance metrics. Best for: Enterprise brands and large agencies running high-spend campaigns across multiple walled gardens simultaneously, particularly those with significant retail media investment alongside search and social. Pricing model: Custom enterprise pricing; standard plan reported at approximately $114,000 per year. Pros: - Best-in-class cross-channel data unification across retail, search, and social. - Celeste AI reduces the analytical workload for large teams without technical backgrounds. - Strong customer support quality. Cons: - Pricing makes it inaccessible for mid-market advertisers. - Interface complexity has a significant learning curve for new users. - Skai simply isn’t built for native discovery or content amplification like some of the other options on this list. ### 8. Birch (Formerly Revealbot) Why it’s essential: Birch is a rule-based automation platform for social and search advertisers who want more control over bid and budget logic than native platforms provide. Rather than full AI automation, it lets you define exactly when and how bids should change, by building conditional rules that run as often as every 15 minutes across Meta, Google Ads, and TikTok. It’s the right choice when you know your optimization logic and want to automate it precisely, without handing decision-making over to an algorithm you can’t inspect. Showcased features: - Rule engine: Build conditional bid and budget rules using AND/OR operators, nested conditions, and custom metrics — more flexibility than native automated rules allow. - Custom metrics: Create your own performance formulas and use them as rule triggers to enable automation based on business-specific KPIs, rather than platform defaults. - Auto-boosting: Automatically promotes top-performing organic posts to paid ads when they meet defined performance thresholds. - Bulk creation: Launch multiple ad variations simultaneously across campaigns and ad sets with centralized management. - Rule libraries: Pre-built automation templates for common scenarios — scaling winning ads, pausing budget drains, managing dayparting — applied in a few clicks. Best for: SMBs and agencies managing social and search campaigns that want precise, rule-based automation without a full AI handoff. Strong for teams with defined optimization logic that they want running continuously, without manual execution. Pricing model: Subscription starting from approximately $99 per month, scaling with ad spend. Pros: - Rule logic runs every 15 minutes — faster than native platform automation. - Custom metrics allow automation based on proprietary business data, not just platform metrics. - Accessible price point relative to enterprise alternatives. Cons: - Rule-based systems require manual updates when market conditions shift — they don’t adapt automatically. - No AI-driven audience discovery or creative optimization beyond what the rules define. - Most powerful for Meta campaigns; Google and TikTok integrations are less robust. ## More About Bid Optimization for Performance Advertisers ### How Does Automated Bid Optimization Work? Automated bid optimization replaces manual bid setting with algorithms that evaluate available signals at each auction and calculate the optimal bid for that impression. At the platform level, those signals include device, location, time of day, audience membership, search query, and conversion history. Third-party tools often add additional signals: CRM data, offline revenue, cross-channel performance, and custom business rules. The key variable is what the algorithm is optimizing toward, e.g., optimizing toward clicks produces different bids than optimizing toward purchases or toward closed revenue. The closer the optimization target is to the actual business outcome that matters, the better the bidding performs. That’s why full-funnel bidding — which connects bid decisions to CRM events downstream — consistently outperforms lead-only optimization for advertisers with longer sales cycles. ### Key Features of Bid Management Tools Automated bid strategies: Target CPA, target ROAS, and maximize conversions are table stakes. The more sophisticated platforms add custom goal types, multi-touch attribution inputs, and cross-channel target coordination. Budget pacing: Ensures spend is distributed correctly across the day, week, or month. Prevents early budget exhaustion that leaves campaigns dark during peak conversion windows. Forecasting: Shows predicted performance at different spend levels before you commit budget. Removes guesswork from scaling decisions. Rule-based automation: Conditional logic that fires bid or budget changes when defined thresholds are met. Most useful for applying business-specific logic that AI bidding can’t account for on its own. Cross-channel coordination: Manages bid targets across Google, Meta, Amazon, and retail media from a single interface. Prevents the inefficiency of optimizing each channel in isolation. ## Key Takeaways The right bid optimization platform depends entirely on where you’re advertising and what you’re optimizing toward. For open-web performance advertising outside search and social, e.g., Realize’s predictive AI and SpendGuard infrastructure are purpose-built for the auction complexity of premium publisher inventory. No single platform is the right choice in every scenario, though. The most efficient bid management strategy uses platforms where their data advantage is genuine — not where their marketing says it is. ## Frequently Asked Questions (FAQs) ### What is a bid optimization platform, and why do advertisers need it? A bid optimization platform automates the process of setting and adjusting bids in digital advertising auctions. Instead of manually managing bids, advertisers define performance goals — a target CPA, a target ROAS, or a maximum budget — and the platform calculates the optimal bid for each impression in real time. Advertisers need them because manual bidding at scale is both time-intensive and mathematically inferior. Algorithms can process more signals, react faster to performance changes, and run continuously without the resource constraints that limit human management. The difference becomes most significant at high account volume and in competitive auctions where bid precision directly affects both cost and conversion rate. ### How do pricing models differ across bid optimization tools? Platform-native tools like Google Smart Bidding and Meta Advantage+ are included at no additional cost, you pay only for ad spend. Third-party tools like Optmyzr and Revealbot charge a flat monthly subscription starting around $99 to $209 per month. Enterprise platforms like Marin Software and Skai use custom pricing, typically scaled to ad spend volume, often starting at $500 to $1,000 per month or more. Realize uses a performance-based model in which campaigns are billed on a CPC or CPM basis, with no separate platform fee. ### Can bid optimization tools replace manual campaign management? They can replace the execution of bid adjustments, but not the strategic judgment behind them. Setting the right optimization targets, structuring campaigns correctly, managing creative quality, and interpreting performance data still require human decision-making. Bid optimization tools handle the labor-intensive implementation layer: the constant micro-adjustments that manual management can’t keep up with. The most effective setups treat automation as a co-pilot, not a replacement for strategy. --- ### How Mail Domain Targeting Unlocks the New 'Search' on the Open Web URL: https://www.taboola.com/marketing-hub/how-mail-domain-targeting-improves-performance/ Last Modified: 2026-04-27 13:05:12 For performance marketers, the search mindset has always been the gold standard: reaching users exactly when they’re looking for a solution. But, over the past few years, search auctions have become significantly more expensive, pushing media buyers and CMOs to look for new environments where user intent is just as strong. One of the most overlooked places where this intent naturally exists is the inbox environment. Platforms like Realize give advertisers access to these kinds of high-intent placements. As founder and CEO of LeaderPrivate, a full-cycle affiliate marketing agency working with 400+ clients across mainstream dating, lead generation, and search arbitrage, I’ve seen mail-based display placements emerge as exactly that kind of environment. Especially within premium web environments, they combine two things that rarely exist together: high consideration intent and massive scale. When executed correctly, they can behave very similarly to search traffic, while offering the reach and creative flexibility of the open web. This is why many performance teams are starting to treat mail-based contextual bundles as a serious alternative channel, alongside traditional search. ## 3 Reasons Mail-Based Contextual Placements Can Rival Search Intent ### 1. From “Scroll Mode” to “Consideration Mode” Most advertising environments today operate in scroll mode. In social feeds, users are browsing for entertainment, distraction, or social updates. Ads interrupt that experience, which means conversion intent is often low or inconsistent. Inbox environments are different. When users open their email, they are typically managing tasks and decisions: reading invoices, confirming bookings, responding to offers, or organizing their schedules. In other words, they’re already operating in a consideration mindset. In this context, well-placed display formats, such as mid-page banners on desktop or tablet, feel less like interruptions and more like relevant offers appearing at the right moment. What makes platforms like Realize particularly effective here is that they don’t treat the open web as one big bucket. The ability to target specific contextual bundles means that for dating and search arbitrage offers, it’s possible to reach 600M daily active users in premium environments where they’re already in consideration mode. For performance advertisers, this shift from passive scrolling to active decision-making can translate into significantly stronger engagement and conversion behavior. ### 2. Hyper-Relevance Through Dynamic Localization Search advertising has always benefited from extreme relevance: A user searches for something specific, and the ad reflects that exact need. Modern display environments can now replicate part of this effect through dynamic localization and contextual personalization. Instead of running hundreds of manually segmented campaigns, advertisers can automatically insert location-specific messaging into headlines and creatives. In practice, the difference is significant. A single campaign can automatically localize headlines like “Find a Date in ” across hundreds of locations simultaneously. Combined with real-time compliance checking of dynamic headlines — a capability Realize offers natively — this approach helps stabilize campaigns that might otherwise underperform due to inconsistent creative relevance. This type of dynamic relevance dramatically increases the perceived personalization of the ad, while keeping campaign management scalable. For performance teams managing multiple geographies, it removes operational friction while maintaining search-like relevance at scale. ### 3. A More Stable Alternative to Social Volatility Another reason performance marketers are exploring inbox environments is auction stability. Traffic on social media has become increasingly unstable: In the past, performance was more evenly distributed throughout the day, but now, algorithm changes, audience saturation, and shifting competition often lead to: - Sudden CPM increases. - Inconsistent delivery. - Performance swings within the same day. Mail-based contextual placements tend to behave differently. Because the environment is tied to daily user habits, like checking email, managing communications, and reviewing notifications, traffic patterns are more predictable. For advertisers, this often results in: - More stable CPMs. - Consistent traffic flow. - Predictable performance trends. For teams trying to diversify away from heavy dependence on a single platform, this stability becomes strategically valuable. ## Why Native Email Inventory Is Still Undervalued Despite these advantages, many media buyers still underutilize inbox-based display inventory. One reason for this is operational complexity, since running campaigns in these environments requires a deeper understanding of: - Audience targeting. - Creative adaptation. - Timing and frequency. - Contextual alignment. Without the right targeting strategy, it’s easy to burn the budget quickly. Successful advertisers typically approach the channel differently: Instead of launching small tests, they allocate meaningful experimentation budgets and run structured creative testing across multiple formats. Once the right combinations are identified, campaigns can scale efficiently. ## Creative Formats That Work Best in the Inbox Environment Since users in inbox environments are in a more attentive mindset, creative quality matters more than volume. Several formats consistently perform well: ### Motion Display Ads Subtle animated elements within standard display units (such as 300×250 or 300×600) help attract attention without disrupting the browsing experience. Motion creates visual contrast while keeping the ad lightweight and native to the page. ### Vertical Video Carousels Short-form vertical videos originally designed for social platforms can be repurposed into display carousels on the open web. This allows advertisers to tell a deeper story while maintaining the visual language users are already familiar with from mobile platforms. ## Scaling with API-Based Automation To fully unlock the potential of high-intent contextual placements, performance teams increasingly rely on automation and real-time optimization. Instead of manual campaign adjustments, advertisers can use tracking integrations and rule-based bidding to: - Increase bids on high-performing contextual bundles. - Pause underperforming segments quickly. - Reallocate budget dynamically as conversions appear. This approach allows marketers to treat contextual display inventory more like search campaigns, where data continuously drives optimization decisions. ## Key Takeaway: Intent Is About Context, Not Just Keywords For years, performance marketing equated intent with keywords typed into a search bar. But, intent also exists in environments where users are actively managing their lives and decisions. Inbox ecosystems are one of those environments. By combining contextual targeting, localized creative relevance, high-impact display formats, and scalable open-web reach, mail-based placements offer a powerful alternative for marketers looking to expand beyond traditional search and social. As acquisition costs continue to rise across major platforms, exploring these high-intent contextual environments may become one of the most effective ways to diversify traffic sources while maintaining strong performance outcomes. --- ### 10 Best Performance Platforms for Motion Ads in 2026 URL: https://www.taboola.com/marketing-hub/best-performance-platforms-for-motion-ads/ Last Modified: 2026-04-23 11:22:06 Today’s digital marketers must battle for a prospect’s attention, which is won or lost in milliseconds. While static display ads have long been an industry standard, we’ve seen how the shift toward motion ads — dynamic, short-form video and looping visuals — is redefining performance. Motion ads are now a must-have strategic tool that can break through banner blindness and simplify complex solutions through visual storytelling. Performance-driven marketers face a serious challenge beyond creating the content, however: choosing the right environment to host it. Walled gardens like Meta and Google? High-intent retail networks? Independent programmatic giants? The choices abound, and the platform marketers choose dictates their brand’s reach, data quality, and cost per acquisition (CPA). This guide explores the top performance platforms for motion ads, breaking down the essential features, costs, and metrics needed to drive meaningful conversations in a movement-first digital landscape. ## 10 Best Performance Platforms for Motion Ads: At a Glance Platform Why It’s Essential Core Use Cases and Features Best for (Performance Advertisers) Pricing Model (Indicative) 1. Realize Performance‑aligned motion and video activation and optimization. Key performance indicator (KPI)‑tied motion/video campaign execution, automated optimization, real‑time performance alignment. Performance teams focused on motion/video ROI. Performance-based model; campaigns billed on CPC basis, or CPM for programmatic. 2. Google Display and Video 360 (DV360) Enterprise‑grade programmatic buying of video and motion formats across screens. Omnichannel motion video ads, real‑time bidding, advanced targeting. Large performance advertisers and agencies. Programmatic spend‑based. 3. The Trade Desk Independent demand-side platform (DSP) with strong video/motion inventory and analytics. Programmatic motion/video ad buying on the open web plus connected TV (CTV) and mobile. Performance marketers needing deep control and cross‑channel delivery. Custom enterprise contracts. 4. Amazon DSP (Video) Programmatic video display leveraging shopping and behavioral signals. Motion video ads on Amazon properties and third‑party inventory. E‑commerce advertisers and motion‑centric campaigns. Cost per one thousand impressions (CPM)/spend‑based. 5. SpotX (Video Ad Platform) Global video advertising platform for programmatic motion ads. In‑stream and out‑stream video delivery, VAST (video ad serving template) compliance. Advertisers needing scalable video placement across inventory. Custom/spend‑linked.6. 6. Vibe.co Self‑serve and managed video ad platform with performance focus. Streaming and CTV motion/video ad inventory. Small to medium-sized businesses (SMBs) to large advertisers targeting premium video environments. Custom/spend‑based. 7. MGID Native performance and programmatic platform, including motion/video formats. Targeted motion ads and rich media in native placements. Performance advertisers wanting alternative reach. CPM/performance‑linked. 8. Epom (Video/Display Ad Server) Video ad server and rich media delivery with templates. VAST‑compliant video ad serving, analytics, multi‑format support. Advertisers and networks managing motion ad fleets. Subscription/template‑based. 9. Meta Video ad tools (via its Ads platform) Massive reach for short‑form motion ads on social feeds. In‑feed video ads, Stories, Reels, dynamic motion creatives. Performance advertisers focused on social video. CPC/CPM/CPA via platform 10. TikTok for Business  Short‑form motion ads with strong engagement and creative formats. Branded motion videos, dynamic creative kits, performance metrics. Brands targeting user‑generated motion content. CPC/CPM/CPA via platform. ### 1. Realize Why it’s essential: Realize is an AI-powered performance advertising platform that utilizes short, looping motion-based creatives to capture user attention and drive engagement across native environments. It’s designed specifically for consideration and conversion goals, allowing advertisers to stand out in a user’s feed through dynamic, visually-driven storytelling that blends seamlessly into premium publisher content. Advertisers use Realize to transform static imagery into dynamic native formats, such as short videos or GIFs, which run without sound and continuously loop. This approach is used to enhance campaign performance by adding subtle movement to creatives, which has been shown to deliver measurable lifts in click-through rates (CTR) and conversion rates (CVR) without requiring a completely new campaign setup. Showcased features: - GenAI Motion Ads: Utilizes short looping video or GIF assets to capture attention in-feed and drive significant improvements in both click-through and conversion rates. - GenAI AdMaker: Creative suite that includes AI image generation and more, to create and optimize both static and high-quality motion assets tailored to specific advertiser goals. - Social Importer: Quickly repurposes existing video creatives from social platforms like Facebook and Instagram into optimized display formats for the open web. - Vertical ads: Supports mobile-first video formats that allow paid social marketers to reuse existing vertical assets to reach incremental audiences in premium environments. - Carousel ads: Allows for the inclusion of multiple interactive cards in a single ad unit, which can feature flexible creative options including GIF and video assets. Best for: Realize works well for performance-driven brands in highly visual or competitive sectors that need to increase user engagement without high production costs. It should be of high interest to marketing teams with minimal video resources, as the platform’s AI tools can automatically generate motion from static images, making it a powerful solution for brands that currently have limited internal creative support. Pricing model: Performance-based model; campaigns billed on cost-per-click (CPC) basis, or cost per one thousand impressions (CPM) for programmatic. Pros: - Proven performance lifts: Motion ads on the platform can drive an average 20% higher conversion rate compared to static formats. - Automated creative generation: GenAI Motion Ads swiftly generates original motion assets directly within the ad creation flow. - Easy campaign integration: Motion assets can be added to any existing campaign and are optimized in the same way as standard static ads. Cons: - Muted video: Motion ads run without sound to maintain a non-disruptive user experience, which encourages brands to prioritize strong, visual-first storytelling. - Fallback image requirement: Each motion asset requires a static fallback image to ensure your campaign achieves maximum reach across all global publisher placements and device types. - Impact-focused duration: Assets are designed for high-impact engagement, and so keep a cap of 15 seconds’ duration for the ads to ensure rapid load times and better performance on mobile networks. ### 2. Google Display & Video 360 (DV360) Why it’s essential: As the enterprise-tier arm of the Google Marketing Platform, DV360 provides a centralized command center for programmatic motion and video campaigns. It’s essential for marketers who need to manage the entire media-buying funnel within a single ecosystem. The platform’s integration with the Google tech stack allows advertisers to do sophisticated audience modeling, and also allows advertisers to buy premium video inventory across nearly every digital screen. DV360 provides a centralized gateway to premium publisher inventory and exclusive private marketplace (PMP) deals, with an inventory that includes 70+ exchanges, private marketplaces, and premium publishers. The platform enables marketers to move beyond basic bidding toward a more nuanced, data-driven approach. Its value lies in its advanced audience frameworks that use first-party data and solid integration between media buying and measurement tools. Showcased features: - Omnichannel motion support: Seamlessly manage video formats across social, web, connected TV (CTV), and digital out-of-home (OOH) from one interface. - Real-time bidding (RTB): Advanced algorithms evaluate millions of video impressions per second to secure the most relevant placements for your budget. - Automated bidding: Uses Google’s machine learning to optimize motion ad delivery based on specific performance goals, like completed views or conversions. - Marketplace and deals: Direct access to premium publishers and curated video bundles to ensure high-quality, brand-safe environments. Best for: DV360 works best for large enterprise performance advertisers and global agencies requiring high-level transparency, scale, and consolidated reporting across multiple regions and channels. Pricing model: Programmatic spend-based; typically involves a platform fee plus media costs. Pros: - Wide reach: Access to Google’s exclusive inventory, including YouTube, and a vast network of premium publishers. - Data integration: Native connection with Google Analytics 4 (GA4) for a closed-loop view of the customer journey. - User review: “I like the breadth and precision of the targeting.” Cons: - Complexity: The steep learning curve may require dedicated specialists or agency management. - Minimum spend: Generally carries higher entry barriers and spend commitments compared to self-serve platforms. - User review: “This can be very expensive for smaller teams, a budget that could be invested in other strategies.” ### 3. The Trade Desk Why it’s essential: The Trade Desk is the leading independent demand-side platform (DSP), offering a transparent alternative to walled gardens. It’s essential for performance marketers who prioritize data sovereignty and want to reach audiences across the web. The Trade Desk excels at cross-channel motion ad delivery, allowing advertisers to maintain a unified view of prospects as they follow potential customers from a mobile video clip to a long-form CTV ad, then to a desktop conversion. The platform uses Unified ID 2.0 to maintain a cohesive view of the customer journey across all devices. An automated optimization engine analyzes data in real time and shifts budgets toward specific groups most likely to convert. Showcased features: - Kokai™ AI: This powerful AI engine analyzes data in real time to suggest optimizations for video campaigns, helping to lower CPAs. - Planner tools: Enable marketers to map out optimal reach and frequency across different video formats, before dipping into ad spend. - Identity alliance: Advanced cross-device tracking that identifies users across screens, without relying solely on third-party cookies. - CTV and premium video: Diverse inventory of high-definition streaming content that facilitates big-screen storytelling with digital performance tracking. Best for: The Trade Desk is best for performance marketers and mid-to-large agencies that need granular control, deep analytics, and the ability to buy video inventory outside of Google and Meta. Pricing model: Custom enterprise contracts; typically a percentage of ad spend. Pros: - Objectivity: The Trade Desk’s independence means it doesn’t favor its own inventory and focuses on what performs best for the advertiser. - Advanced attributions: Strong capabilities in measuring how video views influence offline or cross-device conversions. - User review: “I like the platform's data-driven targeting and reporting capabilities. The precise targeting and detailed reporting help me reach the right audience, track performance in real time, and make quick optimizations that improve ROI.” Cons: - High barrier to entry: Typically requires significant monthly minimum spends and long-term contracts. - Self-managed: Requires a high level of expertise to navigate and optimize effectively. - Lower variety: The Trade Desk doesn’t own its own media inventory or consumer data, putting it at a disadvantage to Google, Meta, and Amazon. ### 4. Amazon DSP (Video) Why it’s essential: Amazon DSP offers a unique advantage that no other platform can match: closed-loop shopping data. By leveraging real-time behavioral signals from millions of Amazon customers, performance marketers can serve motion ads to people based on what they buy, not just what they browse. This capability makes Amazon DSP an essential tool for e-commerce and brands that want to target professional buyers based on their historical purchasing patterns and intent. Showcased features: - Exclusive inventory: Run motion ads across Amazon-owned properties like IMDb, Twitch, and Fire TV, plus a wide network of third-party publishers. - First-party audience insights: Target users based on in-market segments, like those currently shopping for enterprise software. - Streaming TV ads: Access to high-quality motion placements on Prime Video and live sports backed by Amazon’s robust attribution. - Responsive e-commerce creative: Automatically generates video ad layouts that can include product details, ratings, and Add to Cart buttons right in the player. Best for: E-commerce brands, manufacturers, and advertisers who sell products or services directly related to intent-based shopping behaviors. Pricing model: CPM spend-based: available via self-service or Amazon-managed service. Pros: - Purchase attribution: Directly link video views to actual sales on the Amazon platform. - High-intent data: Access to the world’s most powerful database of consumer and professional buying habits. - Supports two video ad types: In-stream video ads play before, during, or after content; out-stream video ads play within non-video content (social feed or in an article). Cons: - Walled garden: Analytics are largely confined to the Amazon ecosystem. - Creative standards: Amazon has strict guidelines for video quality and content that may require more production effort. - Steep learning curve: Easy to make mistakes if you don’t have training; may need to hire a partner. ### 5. SpotX (Acquired by Magnite) Why it’s essential: Following its acquisition by Magnite, SpotX has become one of the world’s largest independent video ad exchanges. It’s essential for advertisers who need a specialized, video-first infrastructure. Unlike generalist platforms, SpotX was built to handle the technical complexities of programmatic motion ads like VAST/VPAID compliance. It ensures your video assets play perfectly across thousands of different publisher environments and devices. Showcased features: - Global video exchange: A massive marketplace of premium in-stream (within video content) and out-stream (between paragraphs of text) placements. - Advanced CTV solutions: Tools specifically designed to navigate the fragmentation of the connected TV market. - Server-side ad insertion (SSAI): Delivers a seamless, TV-like experience for motion ads by eliminating buffering and ad-blocker interference. - Audience management: Allows advertisers to layer their own first-party data over SpotX’s premium video inventory for precise targeting. - Best for: SpotX is best for scalable video campaigns where advertisers must reach a broad but high-quality audience across diverse digital environments. Pricing model: Custom/spend-linked; programmatic bidding via CPM. Pros: - Technical excellence: Superior ad delivery tech that reduces broken ads and slow loading times. - Inventory quality: Strong emphasis on brand safety and direct relationships with top-tier publishers. - Customer support: Good customer support that includes 24/7 live access via phone or online. Cons: - Video only: Not a one-stop shop for other formats, like display or search. - Platform fragmentation: Since it’s now part of the larger Magnite ecosystem, navigating SpotX's specific toolset can sometimes be confusing for new users. - User review: “The Reporting feature still lacks precision and takes a very long time compared to other platforms. Usually, every month-end or start, we face issues in scheduled reports, which delay our reporting plans.” ### 6. Vibe.co Why it’s essential: Vibe.co is democratizing the CTV landscape by making streaming video ads accessible to brands of all sizes. It’s essential for performance advertisers who want the benefit of TV advertising and the agility of social media ads. The platform removes traditional barriers to entry in TV, like huge minimum spend requirements and complex negotiations. Marketing teams can launch streaming motion ads in minutes. Showcased features: - Self-serve CTV platform: An intuitive interface allowing users to upload video assets and set live campaigns on premium channels like ESPN or Hulu. - Real-time reporting: Track performance metrics like CPA and return on ad spend (ROAS) in real time, just like you can on Meta or Google. - Granular targeting: Target audiences by zip code, interest, or specific TV apps and genres. - Budget flexibility: No minimum spend requirements; marketers can test CTV with smaller performance budgets. Best for: Vibe.co is great for small-to-medium businesses (SMBs) and agile performance teams who want to test the effectiveness of high-impact streaming video without a massive upfront commitment. Pricing model: Custom/spend-based; generally operates on a CPM model with a user-friendly bidding system. Pros: - User-friendly: It’s perhaps the easiest-to-use platform for CTV, requiring almost no programmatic knowledge. - Low entry barrier: Makes big screen advertising accessible to brands with modest budgets. - User review: “The most helpful thing about Vibe is the ease to upload, execute, and report on campaigns.” Cons: - Inventory scope: While premium, the inventory focuses more on streaming apps than the broader open web display. - Limited creative services: Unlike some competitors, advertising teams must generally provide their own finished video assets. - User review: “Limited customization, control, and reporting.” ### 7. MGID Why it’s essential: MGID is a leader in native performance advertising. Because the platform offers a natural way to introduce motion ads to users, it’s perfect for advertisers who find that consumers are ignoring their traditional banner ads. By placing motion ads within the recommended content sections of high-traffic news and lifecycle sites, MGID captures users in a discovery mindset. The result? Higher engagement rates for informational or B2B content. Showcased features: - Native video placements: Motion ads that look and feel like part of the publisher’s editorial content, reducing ad blindness. - AI self-serve platform: An easy-to-use dashboard with built-in AI tools to help optimize headlines and motion assets for performance. - Contextual intelligence: Automatically matches your motion ad to the most relevant article content to ensure high user interest. - Smart widget technology: Dynamic ad units that can adapt their layout and motion behavior based on the user’s device and behavior. Best for: MGID is best for performance advertisers seeking alternative reach outside of search and social, especially those in top-of-funnel lead generation or content marketing. Pricing model: CPM or performance-linked (CPC); budget-friendly for testing. Pros: - High engagement: Native formats often see much higher click-through rates (CTRs) than traditional display ads. - Global reach: Access to a vast network of international publishers; great for global B2B campaigns. - Customer support: 24/7 tech and account support. Cons: - Traffic quality: Like many native networks, it requires diligent monitoring to filter out lower-quality clickbait-style sites. - Conversion friction: Since users are often in a reading mode, converting them immediately to a sale can prove more challenging than on high-intent platforms. - Hidden costs: Uses fixed bidding, which means you could miss premium traffic. ### 8. Epom Why it’s essential: Epom serves as a robust backbone for advertisers who want to build and manage their own ad ecosystem. It’s essential for those who need a high degree of customization in how they serve their motion ads. This ad server gives marketers the tools to host, serve, and track complex rich media and video formats across any chosen platform or partner. Showcased features: - VAST/VPAID compliance: Supports all industry-standard video serving protocols, ensuring compatibility with any external video player. - Rich media templates: A library of ready-made motion ad formats (like video sliders or interactive banners) that don’t require heavy coding. - White-label capabilities: Allows agencies to rebrand the platform as their own internal ad-serving solution. - Cross-channel frequency capping: Manage how many times a user sees your motion ad across different websites and apps to prevent ad fatigue. Best for: Epom is best for ad networks, large-scale advertisers, and agencies that manage a vast fleet of motion assets and need total control over the technical delivery. Pricing model: Subscription-based (flat fee) or volume-linked (based on impressions). Pros: - Total control: Marketing teams own the data and delivery logic. - Format flexibility: Excellent support for unconventional motion formats beyond the standard MP4 video. - User review: “The platform helps our team effectively manage retargeting campaigns, increasing our conversion rate.” Cons: - Technical management: Requires more hands-on technical work to set up and maintain, compared to a simple DSP. - No built-in audience: Unlike Amazon or Meta, Epom is a tool to serve ads; marketers must bring their own media placements or publisher relationships. - User review: “Creating a platform profile might take some time for newcomers.” ### 9. Meta Video Ads (Facebook/Instagram) Why it’s essential: Meta remains the undisputed king of short-form social motion. It’s essential because of its sheer volume and the passive intent data it collects. Meta’s ability to serve a 15-second motion ad in anyone’s Instagram Stories or Facebook feed, tailored precisely to user interests and behaviors, is one of the most effective ways to drive quick, measurable conversions. Showcased features: - Advantage+ Creative: An AI suite that automatically crops, brightens, and optimizes your video assets for every placement (feed, Reels, Stories). - Reels ads: High-engagement, vertical motion formats that tap into the fastest-growing consumption trend on the platform. - In-feed video: Seamlessly integrates motion into the user’s social scroll; ideal for short explainer clips. - Lead forms: Video ads that can trigger an instant, pre-filled lead generation form without the user leaving the app. Best for: Meta Video Ads is best for performance advertisers focused on high-speed testing, social engagement, and driving direct leads or sales through short, impactful motion. Pricing model: Auction-based (CPC/CPM/CPA); highly flexible daily budgeting. Pros: - Ease of use: The most advanced self-serve tools in the industry, making it easy for one person to manage a global campaign. - Massive scale: Reach nearly any demographic in the world through a combination of Facebook and Instagram. Cons: - Ad fatigue: High-frequency social environments mean motion assets can burn out quickly, requiring constant creative refreshing. - Privacy shifts: Changes in mobile tracking (like Apple’s ATT) have made attribution slightly more challenging than in years past. - Rising costs: Robust competition on the platform has driven up ad costs, with 2025 benchmarks showing double-digit increases in CPM and lead-generation expenses. ### 10. TikTok for Business Why it’s essential: TikTok is the primary home of sound-on motion. It’s essential because it has redefined how users consume video. What’s out? Polished commercials. What’s in? Authentic, high-energy, user-generated content. TikTok provides an opportunity for marketing teams to use motion ads that don’t feel like ads, which leads to engagement levels and viral potential nearly impossible to replicate on other platforms. Showcased features: - Spark Ads: Allows brands to boost existing organic video content, maintaining the authenticity of a native post while adding performance tracking. - TikTok Creative Center: Provides real-time data on trending songs and motion styles to help marketing teams build ads that fit the current vibe. - Interactive add-ons: Features like voting stickers or gift code stickers can layer over motion ads to drive engagement. - Dynamic creative optimization: Automatically tests different combinations of video clips and music to find the best-performing hook. Best for: TikTok for Business is best for brands targeting younger professionals (think Gen Z and millennials) and for those who can produce high-energy, authentic, and sound-driven motion content. Pricing model: Auction-based (CPC/CPM/CPA); low minimums for starting. Pros: - Unmatched engagement: Users spend more time on TikTok than on almost any other app, and they’re conditioned to watch videos with the sound on. - Cultural relevance: Allows brands to appear modern and human through less formal motion content. - Social commerce integration: A wide array of creative formats complemented by integrated shopping features shorten the customer journey from discovery to checkout. Cons: - Production demand: Content that works on LinkedIn or YouTube often falls short on TikTok; you must create specific, high-energy assets for this platform. - B2B maturity: While growing rapidly, the professional targeting segments aren’t yet as granular as those on LinkedIn or Google. - Brand safety: The platform’s evolving content moderation landscape requires advertisers to carefully manage placements to prevent creative authenticity from sacrificing brand safety or reputation. ## More About Motion Ads in Performance Advertising Motion ads, ranging from short-form loops and cinemagraphs to full-scale video explainers, have shifted from nice-to-have creative assets to essential performance drivers. Whether you have long sales cycles or complex concepts to convey, motion provides a sensory layer that static imagery can’t match. The primary advantage of motion today is its ability to compress information into high-impact, scroll-stopping moments that qualify leads before someone even clicks. ### What Are the Key Features of a Performance Ad Platform? A platform is only as good as its ability to optimize for the bottom line. If you want to incorporate motion ads into your marketing strategy, a top-tier platform should offer: - AI-driven creative optimization: The ability to automatically assemble different assets (video clips, headlines, and CTAs) to find the highest-converting combination for specific audience segments. - Precise targeting: Granular filters — like job title, seniority, and company size — ensure that you won’t waste expensive video impressions on non-decision-makers. - Cross-device attribution: Tracking that follows a user from a mobile view on a morning commute to a desktop conversion at the office. - Dynamic asset scaling: Support for multiple aspect ratios (9:16 for mobile/social, 16:9 for desktop/YouTube) to ensure the motion remains high quality, regardless of where it appears. ### The Average Cost for Video Advertising on Top Platforms Costs vary significantly based on industry and target audience, but 2026 benchmarks for performance advertising show the following trends: Platform Average CPC (cost per click) Average CPM (cost per 1,000 impressions) Primary B2B Strength LinkedIn $5.26 $30.00 - $50.00 Decision-maker targeting. Google $2.00 $2.00 - $10.00+ High intent and awareness. YouTube $0.18 $15.00 - $30.00 High intent, awareness, problem solving. Meta (FB/IG) $1.86 $10.00 - $23.00 Retargeting and reach. ### Key Metrics for Measuring Motion Ad Performance Instead of focusing on vanity metrics like total views, performance marketers should focus on: - Video completion rate (VCR): Tells you if your content is actually holding viewers’ attention, or if the hook is failing. - Conversion rate by view-through: Measuring how many users converted after watching the ad, even if they didn’t click immediately. - CPA: How much it costs in ad spend to secure a single lead or sale. - Scroll-stop ratio: The percentage of people who saw the first three seconds of your motion ad, versus those who scrolled on past. ## Key Takeaways Motion ads have transitioned from a cool, fun-to-have gimmick, to strategic tools that counteract banner/ad blindness, elevate brand awareness, and simplify complex concepts more effectively than static imagery. Modern platforms, like Realize and Meta, use Generative AI to automatically transform static assets into high-converting motion loops, enabling even teams with limited video resources to compete at scale. Choosing the right platform depends on your data needs and goals, but whichever you choose, performance marketers must prioritize the first three seconds (scroll stop ratio) and view-through conversions to understand how motion influences the long-term customer journey. ## Frequently Asked Questions (FAQs) ### What exactly are motion ads in performance marketing? Motion ads include any digital advertisement using motion to convey a message. Unlike traditional brand videos, motion ads in performance marketing are specifically designed to trigger an action. They often use captions, fast-paced editing, and clear CTAs to move a prospect through the funnel. They focus on metrics like CTRs and lead generation, as well as raising brand awareness. ### How do motion ad platforms differ from static display platforms? The primary difference lies in engagement and data. Motion ad platforms can handle heavy data loads and provide metrics like play time and interaction rate. Motion-first platforms use AI to read the video content and identify which specific scenes drive the most conversions, which facilitates more sophisticated A/B testing than a single static image. ### Should performance advertisers use motion ads on all channels? While motion is powerful, it’s best to use it strategically rather than universally. Text still dominates high-intent channels like Google Search, while discovery channels like Meta, LinkedIn, and YouTube benefit from motion. A balanced strategy uses motion at the top and middle of the funnel to build desire. Static ads work well for final retargeting reminders once the prospect has familiarity with the brand. --- ### A Modern Advertiser's Guide to Open Web Attribution URL: https://www.taboola.com/marketing-hub/attribution-measurement-on-the-open-web/ Last Modified: 2026-04-12 14:48:01 Today’s consumers take complex journeys across multiple channels, with limited exposure to any one particular ad. For marketers, that means that the conversion path is fragmented, and it can be hard to connect the dots. Since users engage through various offline and untrackable online touchpoints (like billboards, word-of-mouth recommendations, or shared links), marketers cannot get a full view of the journey. Without that data, it’s impossible to accurately measure campaign impact or assign a reliable score for their contribution to the final purchase. Performance marketers and advertisers are navigating these shifts and looking for new ways to capture insights into attribution to understand which tactics and strategies are working. Open web advertising offers lots of new possibilities, but requires a novel way of approaching attribution measurement. Here, we’ll take a deeper look at the trends affecting performance marketing attribution, and how modern performance advertisers can adapt and succeed. ## Attribution Challenges for Performance Advertisers Attribution isn't just a technical challenge — it's a critical business necessity. Marketers need accurate attribution models to know which campaigns are actually working and to effectively allocate their budget toward the most promising target audiences. Without a clear view of outcomes, or at least a serious effort to create a clear view, you risk wasting money on ineffective campaigns or, worse, misinterpreting data and funneling funds into activities that appear successful, but are ultimately meaningless. Here are the key challenges marketers are facing today: ### 1. Fragmented Customer Journeys Customer journeys are no longer linear, and they vary widely across touchpoints and devices. Consumers have more options than ever for researching and purchasing products, which leads to complex paths that can be very challenging to track. Without a full view of the journey, attribution can be incorrect, which can hurt the bottom line. ### 2. Privacy and Data Restrictions With the adoption of GDPR and other new privacy standards, performance marketers must pay close attention to regulations when capturing and using customer data. While first-party data — information collected directly from your own users — is generally considered the most future-proof foundation for high-performance targeting, it isn’t the sole path forward. Marketers can also still leverage broader data ecosystems for personalization by ensuring they have a valid lawful basis, e.g., consent, under frameworks like GDPR. As major players like Google continue to maintain support for third-party cookies (for now, at least), balancing owned data with privacy-compliant third-party signals is still a viable route for reaching new audiences on the open web. ### 3. Cross-Device and Cross-Platform Measurement In addition to regulatory changes, the future of third-party cookies continues to flip-flop, so it’s become much more challenging to plan effective tracking of user interactions across multiple devices and platforms. It’s common to have gaps in conversion tracking, making it harder for marketers to analyze data and, ultimately, to prove ROI. ## Attribution on Open Web vs. Walled Gardens The “walled gardens” of Google, Meta, and others are digital ecosystems run by a single company, which controls data measurement and access. Walled gardens offer precision within their own ecosystem due to deep first-party data, but they create silos and attribution conflicts. In addition, walled garden attribution models tend to over-attribute and may skew the credit toward a single platform. These platforms differ quite a bit from open web options for marketers, so their attribution models also differ dramatically. Open web attribution brings more opportunity to unify data points across various digital touchpoints, which is essential as customer journeys become more complex and user fatigue grows. Here’s an overview of the differences between open web and walled-garden approaches: Characteristic Open Web Walled Gardens (Google/Meta) Ecosystem Diverse, open. Closed loop. Data Third-party cookies. First-party data reliance. Transparency and flexibility Using various tools and platforms is possible. Limited visibility into raw data. ## Characteristics of Open Web Attribution Models ### Diverse Ecosystem The open web consists of countless publishers and ad tech partners, making it a fragmented environment for tracking. ### Reliance on Third-Party Data Historically, the open web has relied on third-party cookies for cross-site tracking, but there remains a lot of indecisiveness about their future, making them tough to build future plans around. ### Transparency and Flexibility Advertisers can choose the attribution tools and platforms they prefer on the open web, instead of relying on the proprietary tools from closed systems. That allows for more flexibility and transparency in how data is collected and analyzed. ## Characteristics of Walled Garden Attribution Models ### Closed-Loop Systems These platforms have their own logged-in user bases, allowing them to track user behavior and conversions within their ecosystem with high accuracy. ### First-Party Data Reliance Walled garden platforms rely heavily on their extensive first-party data to attribute conversions, which is less affected by third-party cookie restrictions. ### Limited Visibility Advertisers using these platforms have limited visibility into the raw data and must rely on the platform's reported metrics, which can make it difficult to compare performance across different channels. ## What to Prioritize to Ensure Robust Attribution on the Open Web Like any other digital channel, the success of open web projects ultimately hinges on performance. Every business should define its own key performance indicators (KPIs), but the top priority is always the tangible business outcome, such as generating leads, sign-ups, or actual revenue. The most frequent error marketers make is over-relying on a single attribution model, which risks skewing results and leading to flawed conclusions. When you’re building an attribution model for the open web, consider combining that with marketing mix modeling (MMM) and incrementality testing for best results. These three methodologies — advertiser-owned model; MMM; and incrementality testing — are complementary, and when you combine them in a unified framework, they provide the most comprehensive, accurate, and actionable view of marketing performance. Each methodology answers different questions and compensates for the others’ limitations. The open web offers a ton of opportunity, but you’ll need a solid strategy and the right technology to quickly test, learn, and optimize. As you’re planning for this new paradigm, both in strategy and choosing a platform provider, make sure to consider these key areas and what’s needed for your business: ### Flexible Conversion Tracking Mechanisms Without third-party cookies, there are a few ways to track user movement and behavior to understand where they’re converting. First-party pixel solutions put a small snippet of code (the pixel) onto a website to directly capture information on what pages a user visited, time spent on page, purchases made, and more. Server-to-server (S2S) capabilities offer another option to track and gather data in lieu of third-party cookies. These tools share data about app or web activity between two servers, eliminating the need for cookies or other embedded measurement tools. ### Comprehensive Attribution Models Open-web attribution requires the full picture of user behavior, so consider your attribution model options carefully. Ideally, your campaigns can support various models, such as click-through and view-through conversions, depending on your goals. Timing data will also be important; use customizable attribution windows, such as 30-day click-through or 24-hour view-through. These metrics are often much richer and more useful to performance marketers than what legacy systems can offer. ### Seamless Third-Party Integration With the wealth of options on the open web, it’s possible to capture data from many sources. Make sure the platform you choose can integrate with a wide variety of partners, particularly if you plan to use multiple tools. Data transparency will make a big difference in quickly capturing and acting on data. ### Segmentation and Targeting With open web attribution becoming more common due to privacy concerns and complex customer journeys, advertising platforms have developed new features for efficiency and better results. Look for automated audience generation, which automatically creates audiences for remarketing and lookalike targeting. Also explore the option of exclusion capabilities, which give marketers the ability to use conversion data to exclude existing customers from campaigns. This is a key feature for efficiency and budget management. ## Getting Started With Open-Web Attribution The open web may seem like a wild world to explore, but it can provide greater control and more effective ad campaigns for marketers, plus more opportunities to capture first-party data and use it wisely. --- ### How Gambling Brands Can Capture Attention and Drive FTDs URL: https://www.taboola.com/marketing-hub/drive-first-time-deposits-performance-campaigns/ Last Modified: 2026-06-28 05:05:25 If you watched the Super Bowl — or really, any major sporting event these last few years — you’ve encountered an advertising blitz of gambling apps, websites, and services. It’s not just during the actual game, either, as searching for anything sports-related, like stats and scores, will bring up banners, text-based ads, and more. This means that the World Cup will be an absolute whirlwind: not just for the players on the pitch, but for gambling brands trying to capture a few seconds of attention in a sea of digital distractions. Winning that “share of screen” is a high-stakes game of its own. To help you cut through the noise, we’ve tapped into the expertise of our lead performance pros at Realize, advertising account manager Ana Ligeiro and advertising sales manager Rafael Saboia. They live and breathe the iGaming and sports betting world, helping global brands grow by using smart predictive modeling and fresh creative ideas. By breaking down their work with successful partners like Brazilian iGaming platform Bet7k, we’ve identified four proven ways to help your brand take the lead during the big tournament. ## 4 Ways Gambling Brands Can Capture Attention and Drive Sign-Ups During the World Cup ### 1. Leverage GenAI to Break the “Static” Noise During the World Cup, sports fans are scrolling through a mountain of content. If your ad is just a flat, static image, it’s likely to get ignored (that’s what we call “banner blindness.”) To win the click, you need to disrupt that scrolling pattern with a bit of movement. The Strategy You don’t need a massive production budget to introduce action: By using Realize’s GenAI AdMaker, you can take your existing “raw” photos — like a simple shot of a stadium, or a player — and add subtle motion. Think flickering stadium lights, or grass turf kicking up. “You can use this quick fix and it will work perfectly,” advises Ligeiro. “In fact, there are massive betting companies that rely solely on these adjustments. You simply extend the image and let the AI work — you don’t even need to put a request in with your designer.” In terms of the creative itself, Marllon Angelis, of Bet7k, agrees that authenticity beats high-end editing, saying that, “The ‘raw’ creative is what sells: The more Photoshop you have, the worse it performs.” Why It Works Using AI-enhanced motion ads acts as an instant boost to primary performance, quickly and easily making imagery more dynamic while ensuring it’s still “authentic” enough to bypass ad blindness. ### 2. Own the Auction with “Championship” Campaign Structures If you want to grow, you need to go where the action is. Setting up dedicated campaigns specifically for the tournament (also known as Championship/Campeonatos structures) is like putting your best players on the field. Data shows these specific setups can drive nearly 20% of a brand’s total sign-ups during the tournament. A single flagship tournament creative featuring a brand mascot can drive approximately 10% of total sign-ups alone. The Strategy Use Maximize Conversions bidding. Think of this as an automated assistant that ensures your ad is always visible when fans are most active. “100% of our betting clients are running Maximize Conversions,” explains Ligeiro. “It ensures you’re always inside the auction — it’s the magic formula for betting brands.” While this may sometimes seem more expensive at first, it pays off over time. “The cost-per-acquisition was higher, but the user lifetime value was much better,” Angelis notes. “The Realize user is different.” Why It Works When you stay competitive in the auction during those peak moments, you’re doing more than just getting seen — you’re connecting with a higher tier of users across the open web. These aren't just “one-and-done” casual players, they’re the kind of high-value fans who stick around for the long haul, which is exactly what keeps your long-term return on ad spend (ROAS) growing. ### 3. Maximize Efficiency Through Aggressive Remarketing The World Cup moves fast. Someone might click your ad during halftime, get distracted, and forget to finish their sign-up. Remarketing is your way of giving them a friendly nudge to come back. The Strategy Use remarketing to catch this “lost” traffic with lower-friction messaging. A clear, direct “PLAY NOW” call-to-action (CTA) works well. In top-performing setups, this specific button drives virtually all conversion volume. “Remarketing is a campaign type that delivers excellent CPAs,” Ligeiro points out. “If acquisition costs are high, we apply Target CPA to significantly improve efficiency.” Indeed, Bet7k saw massive success here, sharing that, “When I ran remarketing, the FTD (first-time deposit) cost was approximately 11% of the average lead cost. It’s an extremely positive result for the house.” Why It Works Giving a direct “Play Now” nudge to warm leads — those folks who’ve already checked you out — is a total efficiency win. It’s much cheaper to get them across the finish line than starting from scratch with a brand-new audience, plus it helps you bring in a much higher volume of sign-ups overall. ### 4. Drive Relevance with Dynamic “Smart” Personalization Fans want to feel like you’re talking directly to them. By using dynamic macros, your ads can automatically update based on where the user is or what day it is. The Strategy Integrate dynamic parameters such as “The favorite in ${city}$” or “${dayofweek}$ is your lucky day!” in your ads. It’s simple to create this withicn Realize. When building your campaign, insert specific placeholders into your headlines. Realize then does the heavy lifting, swapping those tags for the reader’s real-time info the second the ad appears on their screen. It’s an easy way to turn a broad message into a local, timely invitation without any extra manual work. “The user who comes through Realize is much more qualified — we’re a constant in the user journey,” says Saboia. “Sometimes, a user sees an ad on Meta, but Realize is always there in their journey.” Why It Works Think of personalization as a way to close the gap between a user and your brand. When your ad feels local and hits at exactly the right time, it puts the user in the right frame of mind. This makes the move from catching up on sports news to actually making a deposit feel like a natural next step, rather than a jarring interruption. ## The Strategy In Action In mid-2025, Bet7k, a leading Brazilian online casino, made the strategic decision to shift from focusing on first time deposit cost per acquisition (FTD CPA) to optimizing their campaigns towards return on advertising spend (ROAS). While this change aligned better with long-term user value, the campaign’s early results didn’t meet expectations, prompting the team to explore new creative approaches. Using Realize GenAI AdMaker, Bet7k tested subtle motion to specific creatives within their existing Max ROAS campaign, making the imagery more dynamic and capturing attention more effectively in placements across the open web. The impact was immediate: The AI-enhanced motion ads drove the strongest performance within the entire campaign on both FTD CPA and ROAS, demonstrating the important role creative innovation can play even in already performance-optimized setups. ### Why This Worked Motion drives attention: Subtle animation helped the creative to stand out on busy publisher sites and increase engagement in competitive open web environments. AI-enhanced creative testing: Realize’s AI tools enabled rapid experimentation and enhancement of existing creatives without heavy production lift. Creative as a performance lever: Even in the mature campaigns, fresh creative formats unlocked incremental gains, while pairing motion creative with a Max ROAS campaign amplified performance-focused outcomes. ## Key Takeaways For World Cup success, remember the four following rules: - Motion is mandatory: Static ads get lost in the World Cup noise. Use AI to add subtle motion to “raw” photos. Authenticity captures more eyes than heavy editing. - Bid for visibility: Use Maximize Conversions bidding within tournament-specific campaigns to stay competitive when traffic spikes. - Retarget aggressively: Don’t let interested users slip away. Use remarketing with a bold “PLAY NOW” CTA to win back users who dropped off mid-game. - Personalize at scale: Use dynamic macros to make your ads feel timely and relevant to each specific user, fostering a stronger connection and driving higher-quality sign-ups. ## Frequently Asked Questions (FAQs) ### Why should I prioritize motion-based creatives over static banners for high-traffic events? During big tournaments, fans are so locked into scores and news that they often tune out plain, static banners, a phenomenon known as banner blindness. Motion, on the other hand, acts as a pattern interrupt that snaps their attention back to your ad, which usually has a higher click-through rate (CTR) and lower cost-per-click (CPC). For high-performance campaigns on the open web, tools like Realize GenAI AdMaker make this a breeze by letting you quickly add movement to your already successful images. It’s an easy way to see an immediate performance lift without the long wait of traditional video production, ensuring your ads stay as current as the daily World Cup updates. ### How do I maintain a stable CPA when competition and bidding costs spike during the World Cup? It’s common for advertisers to experience auction shock during the World Cup, where sudden spikes in competition send ad prices through the roof. To keep your budget in check, the best move is to split your spending between reaching new audiences and using high-efficiency remarketing to bring back previous visitors. This balance helps average out your costs so you aren’t overpaying for every sign-up. When running high-performance campaigns with Realize on the open web, we recommend using the Maximize Conversions bidding strategy, along with Target CPA guardrails. Think of this as a smart assistant that knows exactly when to lean in: It bids aggressively for fans who are most likely to convert, but automatically holds back on less promising impressions. This keeps your costs stable and your efficiency high, even during the most action-packed moments of the tournament. ### Is it better to target sports fans directly, or broader lifestyle audiences? Targeting sports fans directly seems like a no-brainer, but it’s also the most expensive way to play. That’s why savvy advertisers search for “lookalike” audiences or specific interest groups: think news, finance, or even gaming. In these pockets of the web, competition is usually much clower, but the desire to place a bet is still very much alive. For more advanced campaigns, Realize uses Predictive Audiences to find high-value users based on their actual reading habits. This allows you to reach potential bettors on premium news and entertainment sites, catching them in a headspace where they aren’t already being bombarded by ads from every other brand in the business. --- ### Audience Targeting in CTV: Solving Fragmentation Challenges for Performance Marketers URL: https://www.taboola.com/marketing-hub/targeting-and-fragmentation-in-ctv-campaigns/ Last Modified: 2026-04-12 14:19:55 Connected TV, or CTV, is a big opportunity for performance marketers to find and engage with new audiences. CTV is video content delivered through internet-enabled devices like smart TVs or streaming devices, and includes streaming services like Netflix, Hulu, and Apple TV. It offers the precision of digital matched with the impact of television, and it’s a huge market: 70% of US consumers stream content across multiple platforms. That said, connected TV targeting challenges can be daunting. Advertisers today are used to the streamlined ecosystems of search and social, but CTV is very fragmented: it’s essentially a system of walled gardens, disparate devices, and data siloes. Measuring return on ad spend (ROAS) for CTV can easily become a nightmare for performance marketers. Since connected TV is closely related to OTT (or over-the-top advertising, which refers to video ads delivered directly to viewers over the internet on streaming platforms), the advertising challenges are similar for both, although OTT ads can also include mobile or desktop ads. This guide explores the targeting and fragmentation challenges facing performance teams, and identifies the strategies needed to turn CTV into a measurable growth channel. ## Why CTV Fragmentation Kills Performance Campaigns (And How to Fix It) The connected TV marketing segment offers a lot of promise and can be lucrative for advertisers, but CTV audience fragmentation presents a big challenge. Fragmentation shows up as a series of data siloes across original equipment manufacturers (OEMs), publishers, and operating system (OS) providers, making it different from other siloes, e.g., simply a high volume of apps. For marketers and advertisers working in CTV, these siloes can easily lead to budget waste through unmanaged frequency. Instead of setting and forgetting ad buys, teams should use a CTV frequency capping tool to limit the number of times a user or household sees a specific ad on streaming services over a set period of time. This helps to prevent ad fatigue and optimize budgets. ## Identity Crisis: Targeting Households Without Cookies When adding CTV to a performance marketing strategy, there’s a technical shift from the 1:1 pixel tracking that has traditionally captured user behavior, to IP-based household targeting. Household-level targeting is a digital ad capability that matches postal addresses with IP addresses to deliver ads to a specific home across phones, TVs, and laptops. It’s a way to send digital mail directly, without using cookies. To bridge the gap between TV exposure and mobile conversions, marketers can use identity graphs and deterministic data. This CTV identity resolution is essential for gaining a unified view of households or viewers without cookies. CTV identity graphs are databases that create a cohesive view of a user or household profile using devices, apps, and other fragmented user identifiers. This enables advertisers to match their customer email list with the identity graph to target users on streaming platforms. Deterministic data for CTV is first-party, accurate data that users provided, like email addresses, login IDs, or device IDs. This information allows advertisers to identify individual users or households for CTV advertising with privacy-compliant targeting and accurate attribution. ## Overcoming Measurement Blind Spots in a Fragmented Ecosystem As you’re exploring any new advertising channel, unified measurement and metrics should be top of mind, and that’s no different with CTV. But, CTV performance marketing measurement brings unique challenges. You may know that a majority of consumers are watching connected TV, and the ads that pop up, but it isn’t as straightforward to understand how users are engaging with or acting on those ads. This contrasts with walled garden options like Google and Amazon, which make cross-screen attribution easy for performance marketing teams. Connecting big-screen views to small-screen actions, also known as cross-device attribution, is the primary job of CTV performance marketers. Cross-device attribution in CTV relies on two solutions: automatic content recognition (ACR) and incrementality testing. ACR technology samples content at the screen level to identify viewing habits, using audio and visual fingerprints. Advertisers can measure viewership as well as target or retarget specific audiences and manage ad frequency. CTV incrementality testing, also called lift testing or causal impact analysis, puts users into test and control groups to see how many actual conversions were driven by ads. It quantifies the lift by comparing conversion rates to demonstrate whether CTV ads led to new sales, or simply captured organic ones. ## Consolidating Your Ad Buy: Programmatic vs. Direct IO In addition to tackling the measurement challenges of CTV, it’s important to understand programmatic CTV vs. direct IO — the buying methods for CTV advertising. Direct IO is the more traditional method, a direct agreement between an advertiser and publisher to buy premium inventory at a fixed price and time. Because it isn’t flexible and works only with one publisher or service, direct IO increases fragmentation. A programmatic DSP, or demand-side platform, is a more modern method of buying streaming ad inventory, and generally better suited for CTV performance marketers. A programmatic DSP software platform automates the buying of streaming ad inventory for smart TVs and apps with real-time bidding that’s informed by data. Advertisers can target specific audiences instead of buying broad linear TV spots. These platforms also allow for unified frequency control and dashboard management, which marketers will find similar to social platform dashboards. Leading CTV providers like Paramount Advertising and LG Ad Solutions have adopted Realize’s performance marketing tools to drive performance, by retargeting users across smart TVs and personal devices using audience matching. This functionality helps marketers use their budgets more wisely and move beyond awareness advertising into performance-driven marketing. ## Key Takeaways CTV is a newer market for many performance marketers, but modern ad platforms and strategies can help marketers get control over the frequency and attribution of placements, allowing for household level targeting. While it’s a fragmented market, with multiple services and publishers, marketers can shift to household identity graphs and programmatic consolidation for better results. ## Frequently Asked Questions (FAQs) ### How does audience fragmentation affect CTV campaign performance? Audience fragmentation in CTV creates data siloes, making it harder to see if you’re reaching the same person across different apps. Without a unified view, it’s easy to overdo the frequency of ads, wasting spend and complicating measurement. It’s difficult to calculate a true cost per acquisition (CPA) or return on ad spend (ROAS) without trustworthy attribution data. ### Can I target specific audiences on CTV like I do on Facebook? It’s possible to target specific audiences on CTV, though the method differs from targeting on a platform like Facebook. Instead of tracking user pixels, use identity graphs to target households. These databases target new audiences based on IP addresses, first-party data, and third-party segments, like auto-intenders. ### What is the best way to track conversions from CTV ads? Tracking conversions and measuring ROAS on CTV requires different methods than other types of performance marketing campaign measurement. Users can’t click on a TV ad, so you should track CTV performance with cross-device attribution. This is the best way to track conversions from CTV ads, since it links the IP address of the TV that showed the ad to the mobile device or laptop where the purchase eventually occurred. Deterministic data can also help build a more comprehensive view of CTV ad performance. --- ### 9 Best Performance Platforms for Ad Spend Optimization in 2026 URL: https://www.taboola.com/marketing-hub/best-platforms-ad-spend-optimization/ Last Modified: 2026-04-12 13:08:45 Performance advertisers don’t lose money because they “picked the wrong channel.” They lose money because budget decisions lag behind reality. Yesterday’s winners keep spending, today’s losers keep leaking, and the team can’t see the shift until the invoice hits. That’s why ad spend optimization tools have evolved in two directions: - Native, channel-specific automation (e.g., Meta’s Advantage+), where the platform reallocates spend inside its own ecosystem. - Independent optimization layers (e.g., Realize, Madgicx, Optmyzr, Revealbot, Trapica, Adzooma, Wrench.AI, Fluency), which help marketers forecast, enforce rules, or automate decisions across accounts and sometimes across channels. Below is a practical guide to the leading platforms for ad spend optimization, including what they’re best at, what to watch out for, and how they map to real performance workflows. ## Best Platforms to Optimize for Ad Spend Compared Platform Why It’s Essential Core Use Cases and Features Best for (Performance Advertisers) Pricing Model (Indicative) 1. Realize Predictive performance and spend optimization beyond basic rules. AI‑driven budget optimization, performance forecasting, and publisher placement insights. Advertisers needing AI‑powered spend decisions with broader inventory. Performance-based model; campaigns billed on CPC basis, or CPM for programmatic. 2. Adzooma Free and all‑in‑one PPC management with spend insights. Campaign audits, budget tracking, performance alerts. Small to mid‑sized advertisers optimizing spend across Google, Meta, and Microsoft. Free tier plus paid upgrades. 3. Fluency AI operating system that unifies spend optimization cross‑platform. Centralized ad campaign oversight integrated with major platforms. Advertisers and agencies seeking fully automated cross‑platform optimization. Custom enterprise (emerging). 4. Meta Advantage+ Native AI budget optimization inside Meta ecosystems. Automated budget distribution and creative optimization. Facebook and Instagram campaign spend optimization. Included with ad spend (no separate software fee). 5. Optmyzr Rule‑based and automated spending workflows. Automated bid management, budget pacing, one‑click bulk optimizations. Agencies and advertisers with complex multi‑account pay-per-click (PPC) needs. Subscription (starts higher tier). 6. Revealbot Cross‑platform budget automation and alerts. Rule‑based spend optimization and automated campaign actions. Advertisers needing 24/7 automated budget management. Subscription-/usage- based. 7. Trapica AI-automated bidding with budget allocation across channels. AI targeting, bidding, and spend allocation reducing CPAs. AI‑light optimization for mid‑market campaigns. Custom-/usage- based. 8. Wrench.AI Predictive audience and spend insights. Predictive analytics and audience segmentation to inform budget shifts. Multi‑platform ad spend insights and optimization. Custom tiers. ### 1. Realize Why it’s essential: Realize is an AI-powered performance advertising platform designed to help brands scale beyond the walled gardens of search and social. It serves as a strategic command center for the open web, using predictive algorithms to identify high-intent users across thousands of premium publishers. By shifting from manual bid management to an automated, outcome-based model, Realize ensures that every dollar is directed toward the placements and audiences most likely to convert. You can use Realize to eliminate inefficient spend. The platform is used to automate complex bidding tasks, simulate budget impacts before they are implemented, and dynamically adjust creative delivery based on real-time performance signals. This allows advertisers to maintain a profitable ROAS even as they scale their media buy into new, less saturated environments. Showcased features:  - Performance Simulator allows you to test different budget and bid scenarios to forecast potential outcomes before committing actual spend. - SpendGuard, an automated optimization algorithm, continuously monitors campaign health and blocks underperforming sites or creatives to prevent wasted budget. - Pacing Health Score provides a real-time visual indicator of how well your budget is being utilized, helping you identify and fix delivery bottlenecks instantly. - Maximize Conversions uses deep learning to adjust bids for every single impression in real-time, focusing spend on the users with the highest probability of conversion. - Delivers AI-driven suggestions directly within the dashboard to help you optimize bids, targeting, and creative assets for better efficiency. - Integrated AI assistant (Abby) that provides proactive guidance on budget allocation and helps troubleshoot campaigns that are not meeting spending targets. Best for: Realize is best for performance-driven brands and agencies — particularly in sectors like e-commerce, finance, and insurance — that have reached a point of diminishing returns on Meta and Google. It’s the ideal solution for growth teams that need sophisticated AI to manage large-scale open web campaigns without the overhead of a massive internal media-buying team. Pricing model: Performance-based model; campaigns billed on CPC basis, or CPM for programmatic. Pros: - Automated Waste Reduction: Proprietary tools like SpendGuard proactively identify and eliminate low-value traffic to protect your margins. - Data-Driven Forecasting: The Performance Simulator removes the guesswork from scaling by showing exactly how budget changes will impact your CPA. - Superior Open Web Reach: Accesses exclusive, first-party data signals from premium publishers that aren’t available through standard programmatic exchanges. Cons: - Minimum Data Thresholds: The most powerful AI optimization features typically require a consistent baseline of conversion data to function at peak efficiency. - Full Tracking Adoption: Reaching maximum visibility on user journeys and feeding high-quality data to the algorithm requires an initial setup of both the Taboola Pixel and S2S (server-to-server) tracking; however, once established, it becomes a powerhouse solution for long-term spend efficiency. ### 2. Adzooma Why it’s essential: Adzooma is a practical, accessible layer for auditing accounts, tracking budgets, and surfacing optimization opportunities across Google, Microsoft, and Meta, especially for small-to-mid teams. Showcased features: - Campaign audits to identify waste and structural issues. - Budget tracking and performance alerts. - Cross-channel visibility without heavy setup. Best for: Small and medium-sized businesses (SMBs) and lean teams that want quick wins and guardrails across major PPC platforms. Pricing model: - Free: Features include monthly PPC performance reports; monthly opportunity analysis; monthly SEO/web metrics reports; one SEO profile; one web metrics profile; unlimited ad accounts; budget tracking; education; and one user seat. - Silver: Features include weekly reports; visibility into all available opportunities; five SEO profiles; five web metrics profiles; unlimited user seats; performance report branding; and priority support for $69 a month. - Gold: Features include all the features of free and silver tiers; daily reports; unlimited SEO profiles; unlimited web metrics profiles; unlimited ad accounts; budget tracking; education; unlimited user seats; performance report branding; and priority support for $179 a month. Pros: - Fast time-to-value: Great “sanity checker” for accounts. - Low barrier to adoption: Useful when resources are limited. Cons: - Not a replacement for a senior media buyer: It surfaces issues; it doesn’t define strategy. - May feel lightweight for complex enterprise portfolios. ### 3. Fluency Why it’s essential: Fluency frames itself as a “digital advertising operating system” that centralizes and automates workflows across major channels by reducing operational drag and enabling scalable budget and campaign management from one interface. Showcased features: - Unified workflow layer across channels. - Automation for launches, bulk changes, notifications, and operational execution at scale. - Roadmap includes more agentic/AI-driven optimization positioning (per recent coverage). Best for: Agencies and enterprise teams where operations is the bottleneck: many accounts, many locations, many campaigns, lots of repetitive work. Pricing model: Pricing is based on a percentage of ad spend or a fee per individual ad account. It’s tailored to the organization’s size, scaling from pilot projects to enterprise-wide solutions, and requires a personalized consultation for a quote. Pros: - Massive operational leverage: Particularly for bulk management and cross-channel execution. - Centralization reduces error and time cost: Useful when campaign volume is huge. Cons: - Enterprise onboarding reality: OS-level changes take time, training, and stakeholder buy-in. - May be more ops-centric than “pure performance AI,” depending on your use case. ### 4. Meta Advantage+ (Native) Why it’s essential: Advantage+ is Meta’s native automation suite for budget distribution and optimization inside Facebook/Instagram. In short: Meta reallocates spend across ad sets to maximize results using its internal performance signals. Showcased features: - Automated budget distribution across ad sets. - Optimization tied directly to Meta’s delivery system and signals. Best for: Anyone running meaningful spend on Meta who wants simpler campaign management and algorithmic budget allocation without extra software. Pricing model: An automated, performance-based model where AI optimizes budget, audience, and placements in real time to maximize results. Included with ad spend (no separate tool subscription fee). Pros: - Native advantage: Uses Meta’s deepest internal auction and delivery signals. - Low friction: No vendor onboarding — just configuration. Cons: - Ecosystem limited: It can’t optimize your Google/open web spend. - Control trade-off: You give up some manual allocation precision for algorithmic efficiency. ### 5. Optmyzr Why it’s essential: Optmyzr is built for advertisers and agencies who need structured, rule-based PPC workflows at scale, especially across multiple accounts and complex Google Ads portfolios. Showcased features: - Automated bid management and budget pacing workflows. - One-click bulk optimizations for repetitive account work. - Tooling geared for large PPC operators managing many accounts. Best for: Agencies and in-house teams with multi-account Google Ads complexity. Pricing model: - Essentials: Up to 25 accounts, includes keyword, search query, and ad optimizations; Performance Max (PMax) optimizations and insights; account audits and performance reports; budget monitoring; performance monitoring and key performance indicator (KPI) alerts; data insights and custom strategies for $209 a month. - Premium: Offers unlimited ad accounts, and includes all the features above, plus shopping/PMax retail campaign management; Campaign Automator (one free account); PPC vertical benchmarks; multi-account/cross-platform budget management, dashboards and reports; smart placement exclusions; daily automations for shopping, reports and optimizations; two 30-minute onboarding sessions; and one personalized video training session every six months for $272 a month. - Enterprise: Offers unlimited ad accounts and all of the features above, plus custom data integrations and solutions; access to Optmyzr API on request; Okta Single Sign-On (SSO); a dedicated account manager; and monthly training sessions and check-ins. Pricing is available upon request. Pros: - Operational leverage: Turns senior strategy into repeatable automation templates. - Great for governance: Rule-based controls reduce “random walk” account changes. Cons: - Set-up overhead: You need to define rules, thresholds, and exceptions well. - Less “black-box AI magic,” more “power tools”: That’s a pro for some teams, a con for others. ### 6. Revealbot (Birch) Why it’s essential: Revealbot (now branded as Birch) is purpose-built for 24/7 rule-based automation that offers the kind of always-on guardrails performance teams use to prevent overspend, catch anomalies, and enforce pacing discipline. Showcased features: - Cross-platform rules and automated actions (pause, scale, adjust budgets, notify). - Alerting for performance shifts and delivery issues. - Automation logic that mirrors how media buyers actually operate. Best for: Advertisers who want always-on budget protection, especially when campaigns run continuously or across time zones. Pricing model: - Essential: Features include workspaces with overview; post boosting; reports; activity page; integration with Slack; and support via email for $45 a month. - Pro: Includes all the Essential features, plus automated rules and strategies; Explorer; Launcher; top audiences; custom metrics and timeframes; custom and lookalike audience builder; integration with Slack, Google Sheets, AppsFlyer, Hyros, etc.; support via live chat for $91 per month. - Enterprise: Includes all the Pro features, plus onboarding help; tech setup help; premium support; no limits and no overages apply. Pricing available upon request. Pros: - Reliable guardrails: Great for preventing “silent waste.” - Easy to align with standard operating procedures (SOPs): If your team already has playbooks, rules map cleanly. Cons: - Rules are only as smart as your inputs: Bad thresholds can lead to bad automation. - Requires ongoing maintenance: New offers, new funnels, new creatives mean new logic. ### 7. Trapica Why it’s essential: Trapica positions itself as an AI automation layer for targeting, bidding, and budget allocation, reducing CPA while scaling across multiple paid channels. Showcased features: - AI-driven targeting and bidding automation. - Budget allocation across channels. - “Autopilot” approach for teams that want lighter operational lift. Best for: Mid-market advertisers who want AI-forward optimization without building complex in-house automation. Pricing model: Custom/usage-based (typically quote-driven). #### Pros: - Automation-first posture: Built to reduce hands-on bidding and targeting work. - Cross-channel intent: Designed to operate across multiple ecosystems. #### Cons: - Less transparent pricing: Quote-based models can slow evaluation. - Requires clean conversion data: As with any AI optimizer, measurement quality dictates results. ### 8. Wrench.AI Why it’s essential: Wrench.AI is positioned around predictive analytics and audience segmentation, helping teams decide where to shift budget by improving the quality and actionability of audience insights. Showcased features: - Predictive analytics for segmentation and performance insights. - Multi-platform data ingestion options (varies by implementation). - Audience intelligence to inform allocation and personalization. Best for: Teams who are strong on execution but want a sharper decision layer, especially where segmentation and customer relationship management (CRM)-like understanding can improve spend efficiency. Pricing model: Pricing is primarily volume-based, generally costing between 3 cents and 6 cents per output for services like segmentation, data appending, and analytics, with base subscriptions starting around $500 per month. It’s designed to scale with usage, offering tailored, higher-tier, or enterprise-level pricing. Pros: - Stronger audience-to-budget logic: Helps move beyond “last-click winners.” - Good complement to activation platforms: Especially if you’re aligning spend with segments. Cons: - Not purely a bidding tool: You still need execution systems to act on insights. - Integration effort varies: Depends on your data environment and goals. ## More on Optimizing Ad Spend on Performance Campaigns ### What Is ROAS and How to Improve It ROAS measures revenue generated per dollar spent on advertising, commonly calculated as revenue attributable to ads ÷ ad spend. Here are some best practices: - Fix measurement first: If conversion value is missing or inconsistent, ROAS optimization becomes guesswork, especially for automated bidding systems. - Segment by intent, not just audience size: High-intent cohorts can support higher bids without wrecking efficiency. - Refresh creative deliberately: When creative fatigue hits, ROAS often erodes before click-through rate (CTR) collapses. Automation helps, but only if you feed it new variants. - Use targets appropriately: Platforms like Google Ads and Meta both support ROAS-oriented optimization constructs. Use them when the conversion value is meaningful, not just “lead count.” ### Best Practices for Performance Campaign Budget Allocation - Separate testing from scaling budgets: Testing needs stability; scaling needs speed. Mixing them blurs your signal. - Reallocate based on marginal returns: The question isn’t “what’s best?” Instead, it’s “where does the next dollar perform best?” - Watch pacing like a hawk: Underpacing can be as damaging as overspending, because you lose learning time and miss windows of demand. - Plan for conversion delay: ROAS/CPA can look worse in the most recent window simply due to delayed attribution. Many platforms explicitly warn about this when evaluating ROAS. ### Common Mistakes in Performance Campaign Budget Management - Chasing yesterday’s winners: Over-allocating to segments that have already saturated. - Overreacting to noise: Making big budget moves off small sample sizes. - Ignoring incrementality: A low CPA isn’t always incremental value, especially when retargeting eats the budget. - Letting automation run without guardrails: Even AI needs constraints on parameters like spend caps, anomaly alerts, and quality blocks. ### Automated Bidding vs. Manual Bidding for Performance Campaigns - Automated bidding is best when you have consistent conversion tracking, enough volume, and clear optimization events. It’s faster than humans at reacting impression-by-impression and adjusting to micro-shifts in auctions. - Manual bidding is useful when data is sparse, events are noisy, or you’re intentionally steering delivery. In reality, the best programs use automation for execution and humans for strategy with tools like Revealbot and Optmyzr enforcing playbooks, and platforms like Realize and Meta handling real-time optimization inside their ecosystems. ### Key Metrics for Measuring Performance Campaign Success - ROAS (value efficiency). - CPA (efficiency per conversion). - Conversion rate (funnel quality). - Incremental lift/holdout performance (true business impact). - Pacing/budget utilization (delivery health). - Creative performance indicators (fatigue and variant learning). ## Key Takeaways The best spend optimizer for you depends on where you buy media and how mature your measurement is. It’s possible to use native automation for fast wins inside a single ecosystem, and independent tooling when you need governance, cross-account scale, or decision support across channels. The most important features that will optimize your ad spend are forecasting and putting in solid guardrails. Modeling budget changes and enforcing rules and alerts prevents the two biggest spend killers: guesswork and lag. ## Frequently Asked Questions (FAQs) ### How do I optimize ad spend for lead generation campaigns? Start by validating lead quality and not just lead volume. The next step is to connect downstream outcomes (think booked calls and funded accounts) to your tracking so automation optimizes toward real value. Then, use pacing controls and rule-based guardrails to prevent spend from concentrating on cheap-but-low-quality placements or audiences. Keep creative testing running continuously so you don’t “optimize” into fatigue. ### CPA vs. ROAS: Which is better for performance campaigns? CPA is best when every conversion has roughly similar value. ROAS is stronger when conversion values vary meaningfully because it optimizes toward value per dollar, not just cost per action. ROAS is generally defined as revenue attributable to ads divided by ad spend. ### Do I need separate tools if I already use Google and Meta? Not always. If you’re mostly in one ecosystem and your goal is basic automated allocation, native tooling may be enough. You add separate tools when you need to: first, cross-account governance; second, rule-based guardrails and alerts; third, operational automation at scale; fourth, forecasting; and fifth, deeper audience/decision intelligence that platforms don’t provide natively --- ### Codeless Conversion Tracking: Accelerating AI-Driven Performance on the Open Web URL: https://www.taboola.com/marketing-hub/codeless-conversion-tracking/ Last Modified: 2026-04-12 12:39:13 Performance advertisers scaling their campaigns beyond search and social have to make the most of the open web’s potential. Success on the open web requires agility, flexibility, and real-time data, and new codeless conversion tracking technology can meet these needs. Codeless conversion tracking cuts out the need for developer help in setting up conversion tracking for marketing campaigns, so marketers can set up and track events themselves, feeding high-quality signals into their bidding systems. It’s fast, simple, and easy to use. The result is that marketing teams can use AI’s predictive power to optimize their algorithms, perform better in new ad placements, and drive improved return on ad spend (ROAS). ## What Is Codeless Conversion Tracking? In early 2026, Google Ads launched updates to its Google Tag Manager (GTM) that allow users to track events like page loads and form submissions, without interacting with website source code. Marketers can use codeless conversion tracking by setting up events directly in Google Ads. This brings a simple interface for marketers to define a conversion URL or form by clicking a few checkboxes. Previously, marketers relied on developers to make these changes and capture related data. With this power and independence, performance marketing teams can have a more direct impact on ROAS optimization and avoid tag conflicts that can arise when using Google Tag Manager form tracking. This new codeless option allows for faster testing and iteration without the need for specialized skills or complex setup. ## Why Open Web Advertising Demands Agile Tracking Tracking is more important than ever to performance marketing teams. Open web advertising is an entirely different game than working in the walled gardens of search and social, requiring speed and agility throughout the process, and that includes tracking results. When advertisers are testing new open web publishers, native placements, and various ad networks, they have to spin up new landing pages quickly. Codeless tracking is a big step toward collecting data at that same speed, without slowing down rapid experimentation. ## How AI-Driven Bidding Relies on Real-Time Conversion Data The AI-driven bidding of modern performance marketing goes hand in hand with modern conversion tracking. AI predictive bidding algorithms look for patterns among users who convert, to find similar audiences on the open web. Adding codeless conversion removes tracking implementation delays, so the AI engines get a continuous stream of high-quality data — an important part of the machine learning (ML) optimization step. ## Eliminating Developer Bottlenecks to Speed Up Execution Performance marketing is all about speed, and waiting on web developers to deploy tracking scripts can be anything but fast. This step has slowed down advertising teams for years, so codeless event detection promises a big improvement in executing AI-driven performance campaigns. Codeless conversion tracking lets media buyers launch campaigns, track new interactions, and tweak the strategy in real time, all of which represents a huge competitive advantage. ## Setting Up Codeless Tracking When you’re getting started with this no-code event setup in Google Ads, you can either create conversion actions manually, or use website events within the interface: ### Using Google Ads Go to Goals>Conversions>Summary and add a new conversion action. Once you enter your web domain, click scan, then choose the option to create a codeless event, or add a manual event and select the page load option. You can then configure the URL parameters, set your goal category, value, and count, then save the action. This option is easy to use and offers a lot of flexibility across clicks, forms, and scrolling. ### Using Google Tag Manager With the new codeless conversion tracking option, you don’t have to use Google Tag Manager, but you can use it to make sure the core Google Tag is installed. You can also create a Google Ads Conversion Tracking tag here using the conversion ID and conversion label to trigger events, like clicks or forms, within the Tag Manager. This option is less flexible, but very easy to set up and good for simple conversions, like a “Thank You” page. Whichever you choose, as long as the base Google Tag is installed, enabling codeless event tracking or GTM form submissions takes only a few clicks. ### Using Realize for Open Web Campaigns For advertisers scaling on the open web, the Realize platform provides a streamlined environment to manage these no-code setups. To ensure your campaign is optimized for success, you should use the Tracking Test Tool found under the Tracking tab in Realize. This tool allows you to simulate a user journey to confirm that your conversion events are firing correctly and being captured in real time. Additionally, you should implement UTM parameters at either the campaign or ad level (but not both) to avoid data conflicts. By using macros like {site} and {platform}, you can gain granular insights into which open web placements are driving your conversions, ensuring your tracking setup provides the visibility needed to scale effectively. ## Tracking Form Submissions and Lead Generation Events B2B and lead generation-focused advertisers can get a lot of value out of codeless conversion tracking. The technology can automatically track diverse types of forms and multi-URL submissions without custom scripts or broken tracking. Cleaner lead data helps AI tools identify which open web placements are actually driving prospects instead of cheap clicks, and this feature shifts more power into the hands of advertisers who need to move fast. ## Fueling Value-Based Bidding (VBB) With Codeless Signals Good AI capabilities need the highest possible quality of data. Not every conversion is equal on the open web, and codeless tracking is able to capture the initial lead, which can be matched with offline CRM data to attribute revenue accurately to an open web campaign. This then fuels improved value-based bidding, because the full-funnel visibility allows the AI to bid based on anticipated customer lifetime value (CLV) versus a cost per acquisition (CPA) number that doesn’t have the full context. ## Empowering First-Party Data Collection in a Privacy-First World Along with the other benefits I’ve mentioned — simplicity and speed — codeless tracking integrates with enhanced conversions to securely match user-provided data. That’s huge in this changing landscape of data privacy, with stringent regulations and the uncertain future of third-party cookies. Using consented, reliable, first-party data is better for marketers and improves your AI’s predictive modeling capabilities. ## Knowing When to Escalate to Developer-Led Tracking With these new do-it-yourself tools, there may still be times when you need to call on an expert. Codeless tracking handles page loads and standard forms easily, but some areas where codeless tracking falls short include highly complex on-page interactions, or dynamic e-commerce revenue variables. In these scenarios, ask for developer help in setting up a robust data layer. ## Best Practices for Scaling Open Web Campaigns With No-Code Tech As you’re embarking on your own no-code tracking journey, try these tips for success: ### Track everything Include macro- and micro-events — everything that leads up to a conversion — to give the AI the most data points possible. ### Set up accurately Ensure you get all the details right up front, such as using specific matches (not “contains” matches), using server-side (S2S) tagging, and doing a live test of the conversion flow at the beginning to make sure it shows up accurately in the dashboard. ### Monitor and optimize As with data hygiene more broadly, make sure to monitor and optimize conversion tracking after it’s set up, to make sure the algorithms are acting on accurate signals. ## Key Takeaways Codeless conversion tracking closes the gap for performance marketers who need to scale quickly without losing time to development overhead. Now available in Google Ads and Google Tag Manager, no-code tracking also plays an important role in feeding high-quality data into AI algorithms. Access to this feature removes another barrier for performance marketers to publish, test, and iterate advertising campaigns on the open web. ## Frequently Asked Questions (FAQs) ### What is codeless conversion tracking? Codeless conversion tracking lets marketers set up conversion events, like page loads and form submissions, without needing help from developers to manually add code to a website’s source. This feature is available in Google Ads and Google Tag Manager along with Microsoft, and offers a simple interface for users, saving time and providing a seamless connection to AI tools. ### How does codeless tracking improve AI campaign performance? Codeless conversion tracking offers many benefits to marketers, including improved AI campaign performance. AI and ML algorithms only work well when they can access large amounts of accurate data to identify patterns of which users are most likely to convert. Codeless tracking ensures error-free data collection to feed the AI in real time. This improves bidding efficiency along with better user information. ### Can I use codeless tracking for campaigns outside of search and social? Yes, you can use codeless tracking for campaigns outside of search and social. Codeless tracking operates on your own site and landing pages, so it captures user behavior driven by traffic from any open web advertising channel. This makes it easier and more accurate to evaluate the performance of new traffic sources. ### What are the limitations of a no-code tracking setup? When using a no-code or codeless conversion tracking setup, remember that it’s ideal for actions like page views, link clicks, and standard form submissions. If you’re tracking less straightforward actions, like complex dynamic variables, granular e-commerce data, or custom user interactions, get help from a developer to build a data layer. --- ### Beat Meta Diminishing Returns with Yahoo Mail Premium Inventory URL: https://www.taboola.com/marketing-hub/beat-meta-diminishing-returns-with-yahoo-mail-premium-inventory/ Last Modified: 2026-04-12 12:20:19 Every growth-focused home improvement brand eventually runs into the same problems: performance plateaus, budgets increase, and lead costs follow. Audience expansion may generate more clicks, but fewer qualified inquiries. The issue isn’t visibility: it’s context. Home services are rarely impulse purchases. They’re planned, researched, and evaluated. That means brands need environments where homeowners are thinking about real-life decisions, not just scrolling for inspiration. Premium inbox inventory, particularly Yahoo Mail placements accessed through the Realize network, creates exactly that context. With access to more than 900 million logged-in users, advertisers can reach homeowners in a task-oriented setting primed for action. In this guide, our performance experts (advertising sales manager Jeremy Bade, advertising account management team lead Victoria Proença, and Growth Advertisers advertising account manager Jason Poulos) walk us through how a concrete floor coatings company broke through its scaling limits, by expanding beyond social and into Yahoo’s premium community. The brand in question had hit a performance ceiling on Meta: budget increases led to rising costs per lead (CPLs), audience expansion introduced lower-intent users, and creative performance was flattening. To continue scaling, they needed diversified exposure on the open web — specifically Yahoo Mail’s premium, high-trust inventory — to generate qualified leads for a specialized polyurethane flooring solution. What followed wasn’t a simple channel swap, it was a strategic shift in how campaigns were structured, segmented, and creatively optimized, using performance marketing platform Realize. ## The Problems Faced When Trying to Scale on Meta Scaling on social platforms tends to follow a predictable arc: early efficiency followed by optimized targeting and marginal budget expansion, then diminishing returns. For this particular brand, five core issues were stifling performance: - Plateaued performance on social media: When spend scaled, cost efficiencies deteriorated rapidly, resulting in wildly escalating CPLs. - Low-intent traffic: High click volume did not translate to consistent consultation bookings. - Platform fatigue: Repeated exposure across search and social reduced responsiveness. - Creative fatigue: Static ads stopped commanding attention. - Broad targeting inefficiency: Third-party marketplace audiences delivered expensive clicks with poor downstream conversion. In short, performance stalled. The “walled garden” environment had been maximized. This is what many advertisers refer to as Meta’s scaling ceiling: the point at which increasing budget no longer produces proportional lead growth, but instead drives up your CPL as frequency rises and audience quality thins out. Breaking through that ceiling requires accessing new environments with different user behaviors and intent signals. The shift to Yahoo Mail offered this brand incremental reach and a different psychological environment. Unlike a social feed, the inbox is built for decision-making. That distinction is critical when selling high-cost services like concrete floor coatings. When someone is reviewing household communications, renovation estimates, or neighborhood updates, they’re more primed to engage with practical service offerings. This behavior alignment often produces stronger lower-funnel metrics than interest-based browsing alone. ## 3 Strategies to Overcome Diminishing Returns with Home Service Performance Campaigns ### 1. Segment by Device and Platform One of the most immediate structural changes involved device segmentation. High-ticket home improvement services, like polyurethane floor coatings, require careful planning. Desktop users tend to research more thoroughly, complete longer forms, and schedule consultations. Mobile, on the other hand, offers volume. “We always recommend splitting out mobile from desktop and tablet,”says Bade. “This helps us remain competitive. If you bundle them with too low a budget, you simply won’t be competitive enough to win the premium Yahoo Mail placements.” This operational detail is more important than it might at first appear. When advertisers merge device types under a single modest budget, bids struggle to clear premium inventory auctions, particularly within Yahoo Mail. By segmenting devices, you allow: - More aggressive bidding for high-intent placements for desktop users. - Mobile campaigns to drive affordable traffic volume. - Independent optimization based on device-specific conversion rates. In many high-consideration home service campaigns, desktop traffic often shows stronger lower-funnel engagement. Segmenting devices can help your team optimize bids and budgets more precisely, instead of relying on blended metrics. ### 2. Pre-Qualify with Creative Excellence Audience targeting isn’t the only way to filter quality. Creative can serve as the gatekeeper. In home improvement, narrowing audiences through expensive third-party data often inflates costs per click (CPC) without proportionate conversion lift. Instead, strategic messaging can naturally filter disinterested users. “We prefer to pre-qualify users through the creative funnel,” says Proença. “Especially in home improvement, we move away from marketplace audiences to keep CPCs lower, focusing instead on headlines that call out the specific pain point — like ‘Epoxy vs. Polyurethane’ — to ensure only interested homeowners click.” This approach is deceptively simple but highly effective. By clearly highlighting a specific decision point or service comparison in the headline, you can: - Attract homeowners actively evaluating solutions. - Deter casual browsers. - Maintain lower CPCs by avoiding premium data overlays. - Improve form completion rates through aligned expectations. In Yahoo Mail’s environment, where space is limited and attention is finite, specificity wins. Creative that addresses a real household frustration feels less like an ad and more like a helpful suggestion within a task-driven setting. ### 3. Boost CTR with Motion Ads and Dynamic Keywords Standing out in an inbox often requires more than static imagery. That’s why the team introduced motion-enhanced assets — subtle animations derived from still images — to increase scroll-stopping power. “We see a significant lift in CTR (click-through rate) when we use images with people, like an installer at work,” says Bade. “Even better, we use the GenAI Motion Ads tool in Realize to add subtle movement, making a static floor transformation look like a living process.” The psychology behind this is fairly straightforward. Human presence signals trust and professionalism, and motion indicates relevance and urgency. Combined, they elevate engagement rates without requiring expensive video production. “Don't ignore the power of localization,” adds Poulos. “Using dynamic city insertion in your headlines — like ‘Best Garage Floors in ${city}$’ — makes the ad feel like a local recommendation, which is gold for lead gen.” Dynamic city insertion increases relevance while preserving scale. In performance environments where users manage regional service needs, seeing their own city name creates immediacy. This is where open web campaigns differentiate themselves. Instead of relying solely on algorithmic audience expansion, you gain more control over how and where your message connects with decision-makers. ## Key Takeaways Transitioning from social-only acquisition to open web performance campaigns requires structure, discipline, and carefully honed creative strategy, but the payoff can make it well worth it. Home improvement brands hitting Meta’s scaling ceiling should segment campaigns by device, use messaging to pre-qualify prospects instead of overpaying for narrow audiences, incorporate motion and localization to increase engagement, and diversify into premium inbox environments where homeowners are actively managing real-life decisions. Success also depends on rigorous, ongoing optimization to protect lower-funnel efficiency. By moving beyond walled gardens and into publisher environments like Yahoo Mail inventory accessed through the Realize network, brands can tap into high-intent environments that support more stable, scalable growth. ## Frequently Asked Questions (FAQs) ### Why should I use Yahoo Mail ads instead of just Google or Meta? Yahoo Mail provides a premium, less cluttered environment where users tend to be more focused and task-oriented compared to social feeds. Unlike platforms built for browsing, inbox environments support more intentional behavior, including reviewing bills, confirming appointments, and managing household decisions. From a broader performance perspective, campaigns on the open web allow you to capture attention within this high-intent, personal setting while reaching incremental audiences outside traditional search and social. Because these placements leverage authenticated, first-party publisher data, advertisers can drive lower-funnel outcomes with greater behavioral alignment and reduced over-reliance on a single channel. ### Do I need a professional videographer to run Motion Ads on Yahoo? High-quality video can certainly help, but it’s not required. Many modern performance platforms allow you to repurpose existing creative — including static images, short clips, or simple transformations — into engaging motion-style assets. Strategically, open web performance campaigns often include built-in tools that automatically convert still images into subtle, looping animations. This enables rapid creative testing without large production budgets. The result is increased attention and stronger engagement inside inbox environments, where motion can significantly improve visibility without inflating acquisition costs. ### How do I make sure I’m not wasting money on low-quality sites? Most advertising platforms provide reporting tools that show where ads appeared, allowing you to review placements and exclude sites that underperform. Monitoring conversion rates, bounce rates, and engagement metrics helps ensure spend is directed toward productive environments. Within open web performance campaigns, this oversight becomes even more strategic. Advertisers can implement structured exclusion strategies to block low-performing domains while relying on pre-bid brand safety filters and curated allow lists of premium publishers. This dual approach protects budget, preserves brand integrity, and ensures ads appear in high-trust environments that support stronger conversion intent. --- ### Effective A/B Testing Techniques to Optimize ROI in Insurance Campaigns URL: https://www.taboola.com/marketing-hub/a-b-testing-to-optimize-roi/ Last Modified: 2026-04-12 12:05:02 If you want to understand why many insurance campaigns stall, look at the landing page. It may not be glamorous, but it’s where campaigns are won or lost. Every additional form field introduces hesitation. Every removed requirement introduces risk. Insurance advertisers are constantly negotiating between volume and value, trying to satisfy both marketing key performance indicators (KPIs) and sales expectations. In the insurance quote space, that negotiation can create a bottleneck. Recently, one account team faced this exact challenge with a leading provider. Instead of chasing incremental ad optimizations, they rebuilt the landing page strategy around structured A/B testing with a clear mandate: measure what actually drives profit. Read on to see how this was achieved, with Realize expert advice Lauren Wint, advertising account manager. ## The Challenge: Friction vs. Fulfillment Insurance lead generation has a well-known obstacle: form abandonment. In a typical flow, users are asked to submit Personally Identifiable Information (PII) — such as name, email address, and phone number — before they receive pricing information or coverage options. That data is essential for sales outreach, compliance documentation, and remarketing. But, when a customer is still researching options, being asked for sensitive details too early can feel intrusive. As a result, a significant portion of users exit before reaching the quote page. This creates some core business questions: - Is a smaller batch of higher-quality leads more valuable than a larger pool of partial submissions? - Does removing friction increase net profit, even if some leads lack contact details? - How much upfront information is truly necessary to deliver value? There’s no universal answer. It depends on audience intent, acquisition cost, premium size, underwriting model, and follow-up efficiency. A/B testing becomes the mechanism for answering those questions with data instead of guesswork. ## The Strategy: A 3-Step Landing Page Testing Framework Effective A/B testing in insurance goes beyond button color and headline wording. It must evaluate how the entire user journey supports the traffic source, audience intent, and your own strategy. A structured framework can be broken down into three core components: - Align the landing page with the traffic environment. - Test the funnel architecture. - Adjust friction at the point of monetization. Each step isolates an important variable without disrupting the overall campaign. ### Step 1: Aligning the Page With Premium Environments Before adjusting form fields or testing conversion mechanics, advertisers need to evaluate how the landing page reflects the placement that delivered the click. Not all traffic behaves the same, and users arriving from trusted, editorial-rich platforms typically expect polish, clarity, and accuracy. “When a landing page is receiving traffic from premium environments like Apple News, standard API location passing often doesn’t apply,” explains Wint. “To ensure the best user experience, we recommend running specific ‘clean’ asset versions that lead to a landing page optimized for those environments. This alignment is the foundational step before you even begin testing form fields.” Establishing this foundation ensures that user expectations, creative presentation, and backend functionality are in sync. Without that cohesion, subsequent A/B tests can produce distorted results, because performance variations may stem from audience response to the landing page itself, rather than your form optimization efforts. ### Step 2: The Architecture — Bridge Page vs. Direct Entry Once structural alignment is in place, it’s time to look at how users enter the conversation flow. Entry activity plays a critical psychological role in insurance funnels, where trust and perceived value heavily influence completion. Traffic is usually divided between two pathways: - Direct-to-form: The visitor lands directly on the lead form’s opening question, starting the data collection process immediately. - Bridge page: The visitor first encounters a short explanatory page that reinforces the offer and requests minimal input, such as a ZIP code, before transitioning to the full form. While the direct approach reduces steps, it also accelerates exposure to required disclosures and PII inputs. The bridge approach introduces context before escalation, often reducing hesitation. “For premium environments, a bridge page provides a bit more context,” says Wint. “This can prime the user to convert at a higher rate, because they understand the value proposition before they are asked for sensitive information.” Testing both structures reveals how different audiences respond. In some cases, immediacy increases conversions, but in others, a brief contextual buffer gets better results. The only reliable indicator is performance data segmented by environment and audience quality. ### Step 3: The “No-PII” Parameter Test After evaluating entry structure, optimization moves to the most granular level in the insurance funnel: required information before plan visibility. Instead of redesigning the entire flow, advertisers can deploy dynamic backend adjustments using URL parameters to control how much information is required before users see coverage options. This enables parallel testing without disrupting operations. “By appending a parameter to your ad URLs, you can trigger a streamlined form experience on the backend,” notes Wint. “This version skips the contact info slides and sends the user straight to the plan options. It’s a strategic trade-off: You lose the individual lead data, but you gain a massive increase in users reaching the final insurance plan options. This allows teams to see if that volume surge results in a higher net profit.” This experiment isolates a critical economic question: Does increased mid-funnel participation generate enough additional income to make up for reduced upfront lead detail? Because insurance profitability depends heavily on downstream behavior — not just submission numbers — this test measures success based on net revenue impact, rather than surface-level conversion rates. Over time, these insights inform whether friction reduction strengthens or weakens overall campaign return on investment (ROI). ## Managing Reporting Without Losing Clarity Running multiple landing page versions across different placements can quickly create reporting chaos. To prevent this: - Assign consistent naming conventions to each version. - Use distinct tracking parameters. - Monitor performance at both micro and macro levels. - Evaluate profit-based KPIs, not just lead volume. UTM parameters, unique destination URLs, and structured campaign naming are essential. When set up properly, performance engines can automatically shift budget toward the highest-performing combinations. Insurance advertisers should also segment performance by environment. A variant that wins in premium news placements may lose in broader inventory. Without segmentation, the results can muddy the data. Clarity in tracking enables faster iteration cycles, which is where compounding ROI gains occur. ## Why A/B Testing Is Especially Critical in Insurance Insurance differs from many e-commerce categories in one major way: The lifetime value of a customer is high, but the conversion path is complicated. That means: - Customer acquisition cost tolerance is higher. - Lead quality matters. - Compliance restrictions limit messaging flexibility. - Trust is essential. Because of these factors, optimizing solely for cost-per-lead often brings misleading conclusions. When A/B testing, landing pages should be structured around economic outcomes like: - Close rate. - Policy activation. - Premium size. - Retention probability. The winning variant isn’t always the one that generates the most submissions: In some cases, it’s the one that produces the strongest downstream revenue relative to acquisition spend. Performance campaigns on the open web give advertisers the ability to reach high-intent audiences outside of traditional search channels, but that opportunity demands strategic design. Without structured A/B testing, scaling budget can be risky. ## Key Takeaways Effective insurance landing page optimization begins with managing form friction, but that lever only works when the surrounding environment is properly aligned. Testing should start with traffic source alignment and funnel structure before moving into deeper adjustments around required information and entry flow. Dynamic parameters then allow advertisers to experiment at a granular level without disrupting the entire user experience. When supported by disciplined traffic infrastructure, these tests transform the landing page from a static endpoint into a continuously improving performance asset that drives sustainable ROI. ## Frequently Asked Questions (FAQs) ### What is the most impactful element to A/B test on a landing page for insurance? Form length and the quality of PII requested are often the strongest levers in insurance conversion testing. Reducing required fields can significantly increase completion rates, while adding fields may improve downstream efficiency. For performance campaigns on the open web, the highest-impact variable is often the alignment between headline value proposition and audience intent. When the messaging seamlessly connects the initial hook with the lead capture experience, high-intent users are more likely to complete the journey. ### How do premium environments like Apple News affect my landing page A/B test? Premium environments tend to attract users who expect fast-loading, visually clean, and credible experiences. They’re less tolerant of cluttered layouts or delayed performance. In performance campaigns on the open web, testing authoritative copy against conversational copy can reveal which tone better matches these environments. Additionally, minimizing technical friction and simplifying visual hierarchy can improve conversion rates among these higher-trust audiences. ### How do I track multiple versions when I A/B test a landing page? Most advertisers rely on distinct tracking parameters or separate destination URLs to differentiate between versions. This ensures performance data is accurately attributed. For open web-based campaigns, granular tracking allows optimization engines to recognize which landing page variant drives stronger insurance conversion outcomes. Over time, the system can automatically allocate more budget to higher-performing versions while providing actionable insight into which elements influence ROI. --- ### Five Optimization Levers You Should Be Pulling for Scaling on Performance URL: https://www.taboola.com/marketing-hub/optimization-levers-for-scale/ Last Modified: 2026-06-10 08:39:55 Nowadays, running performance campaigns requires more than pinpoint targeting and persuasive creative: You also need to deliver operational efficiency. Advertisers who manage large portfolios are often juggling dozens, sometimes hundreds, of active campaigns simultaneously. Each one has its own budget, bids, schedules, and targeting rules. Trying to adjust all of these parameters one by one is not only tedious, it also carries unnecessary risk. If you want to stay competitive, you need systems that allow you to control the most important campaign parameters quickly and accurately. This article breaks down five critical campaign elements that should be managed at scale, backed by the expertise of Realize experts. ## What to Optimize to Scale on Performance ### 1. Bulk Budget Adjustments: Portfolio-Wide Budget Allocation and Reallocation Ad budgets are the backbone of campaign management, but they can quickly become a bottleneck when you apply each adjustment manually. Whether you’re increasing spend ahead of a seasonal push or cutting back on underperforming segments, budget changes should never require a campaign-by-campaign sweep. We recommend allocating or reallocating your daily budgets across your entire portfolio in a single step to achieve greater financial discipline. It’s a centralized approach that prevents individual campaigns from drifting out of alignment with your overall goals, especially during fast-moving performance shifts. Bulk updates also eliminate the lag time between spotting a need and making a change, allowing your teams to stay responsive without compromising accuracy. Ultimately, scaling budget adjustments strengthens the advertiser’s ability to maintain a healthy portfolio, even in the face of constant changes. “To see a true proof of concept, you need the right budget allocation across your platforms. We start by casting a wide net to let the algorithm find where the customers are, but the goal is always to follow the performance, pruning out what’s not working and doubling down on what is. By leaning into those insights, you ensure your spend remains efficient and you’re able to scale the account the moment you identify success.” — Ari Del Rosario, Realize Advertising Sales Manager ### 2.Bidding Strategy Management: Ensuring Uniform Bidding Across Campaign Cohorts Effective bidding strategies are a key driver for campaign performance. If you’re operating similar campaigns under different bidding rules, whether on purpose or by accident, your performance can suffer and become difficult to compare or optimize. To maintain consistency across multiple campaigns, set your bidding approaches in a single action. For example, if all conversion-focused campaigns should be optimized using a specific automated bidding strategy, shifting them together will ensure consistency. Likewise, performance teams might want to move an entire group to a manual or fixed-bid approach while testing sensitivities or attempting to control spend. In that case, manage these changes at scale to avoid confusion, wasted time, and errors that occur when you update campaigns one by one. “The game in digital marketing today is played on the creative front, but your bidding strategy is what unlocks the scale. For conversion-focused campaigns, an automated bidding strategy is essential — it allows the algorithm to do the heavy lifting and calibrate based on hard data. By managing these shifts consistently, you allow the algorithm to learn faster and scale budgets without negatively affecting performance, ensuring you hit your CPA goals across the entire portfolio.” — Sofiia Zuieva, Director of Growth Sales & AM, Realize ### 3. Campaign Scheduling (Start/End Dates): Mass Updates for Promotional Pacing and Deadlines Scheduling parameters, such as start and end dates, are among the most critical and error-prone aspects of campaign management. They are especially vital during promotional periods, product launches, and other time-sensitive initiatives, when even the most minor oversight can result in overspend, underspend, or missed opportunities. By updating your campaign schedules in bulk, you can remove these risks. Let’s say you’re a retailer launching a holiday sale: You may need dozens of campaigns to begin at the same hour across multiple time zones. If you set these start times manually, you increase the chances of error, especially when dealing with multiple teams and clients. You also need to ensure your campaigns end on a firm deadline to avoid budget waste after a promotion has ended or an offer has expired. Bulk scheduling tools help you stick to time-based parameters. By setting new start or end dates for all relevant campaigns in a single action, you ensure perfect synchronization during critical moments. “Scheduling is more than just picking a date, it’s about timing your launch with the content review process to avoid missed opportunities. Because every ad goes through a multi-day review, setting your campaigns to ‘Start as Soon as Approved’ ensures you don’t lose valuable momentum. For time-sensitive promotions, managing these parameters at scale is the best way to synchronize your efforts across teams and ensure you’re live exactly when the market is most active.” — Sanket Welankiwar, Client Success Lead, Realize ### 4. Geo-Targeting Refinements: Instant Application of Regional Inclusion and Exclusion Geographic targeting influences everything from cost efficiency to compliance. Whether a business operates in specific regions, must abide by licensing rules, or wants to shift ad spend toward emerging, high-performing markets, updates must be applied comprehensively and without delay. If you attempt geo-targeting adjustments across multiple campaigns manually, you can trigger inconsistencies that alter performance data or fail to comply with regional regulations. Bulk adjustments allow teams to maintain accuracy while responding quickly to market conditions, regulatory updates, or performance patterns. Bulk geo-targeting also supports ongoing optimization. Let’s say certain regions consistently outperform others: You can update location targeting universally, then quickly reallocate your spend to adjust to campaign insights. “The algorithm needs a regional anchor to start, but the real success in performance advertising comes from not limiting yourself too early. We look for the readers, not just the publishers. By applying broad regional targeting and avoiding over-segmentation, you give the algorithm the freedom to find trending articles and high-performing pockets you might not have predicted. Once you have those data points, you can then refine and universally reallocate your spend to where the real conversions are happening.” — Blessing Osadolor, Realize Advertising Sales Manager ### 5.  Site Targeting (Inclusion/Exclusion): Universal Site Boosts and Keyword Blocks Site targeting plays a key role in performance optimization and brand safety. Whether you use whitelists to prioritize high-performing placements or blacklists to avoid poor-quality or non-brand-safe endorsements, you need to control your site targeting across all relevant campaigns. If a new low-performing site or keyword emerges, you need the ability to block it across your entire campaign set immediately. If you were to apply exclusions one campaign at a time, it would create unnecessary delays and increase the risk that some campaigns would continue serving on ineffective or unsuitable inventory. On the flip side, if you identify a valuable opportunity, you want to add it to inclusion lists in a single action to capitalize quickly on high-quality placements. Having unified, scalable control over site boosts, blocks, and keyword-level adjustments gives advertisers confidence that every campaign reflects the most up-to-date intelligence. “I prefer not to work with overly restrictive ‘Approved’ lists. Instead, I go the other way: I block everything at the account level that doesn’t promise quality. Using automated rules, we ensure that campaigns aren’t even served on sites with poor indicators, such as specific notification or lockscreen placements. If you see that a site type is generating impressions but not converting, you need to be able to flip that switch for the entire account immediately to steer the budget efficiently toward premium publishers.” — Patrick Coyle, Sr. Advertising Sales Manager, Realize ## Key Takeaways If you’re an advertiser overseeing large portfolios, it no longer makes sense to manage campaigns individually. You need to be able to adjust budgets, bidding strategies, schedules, geo-targeting, and site targeting in a single step. It’s essential for modern optimization. Thankfully, bulk management reduces the time you need to spend on repetitive tasks, cuts down on manual errors, and ensures your campaigns operate under consistent frameworks. ## Frequently Asked Questions (FAQs) ### Why is it important to manage multiple campaign elements at scale? Managing campaign elements at scale helps advertisers maintain consistency and avoid the mistakes that occur when the same change must be applied repeatedly across many campaigns. It also frees up significant time, enabling teams to focus on strategy and analysis instead of operational tasks. Realize’s Bulk Edit tool allows advertisers to update up to 200 campaigns simultaneously, providing clear visibility into changes and reducing risk by ensuring updates are applied uniformly. ### Which key campaign elements can be managed using a bulk method? Advertisers can typically streamline universal updates across several high-impact areas, including campaign budgets, schedules, geo-targeting, and site targeting. In fact, these components often require synchronized adjustments to deliver consistent performance across a portfolio. Realize lets you automate all these changes in a single action. ### What is the main benefit of adjusting targeting in a bulk operation? Bulk targeting adjustments enable advertisers to rapidly apply whitelists, blacklists, and other targeting rules across a group of campaigns, ensuring consistent brand safety standards and directing ad spend to the highest-performing placements without delay. Using Realize’s Bulk Edit feature, you can optimize both budgets and targeting parameters at scale, applying boosts, blocks, and manual refinements universally. This will help you improve your return on investment (ROI) and maintain consistent performance across all of your campaigns. --- ### 5 Challenges of CTV Campaigns for Performance Advertisers URL: https://www.taboola.com/marketing-hub/ctv-performance-campaigns-challenges/ Last Modified: 2026-06-30 11:21:42 Most performance marketers understand that search and social advertising has a ceiling. Once audience saturation arises, CPMs start increasing, and diminishing ROI is on the horizon, it’s time to look beyond these channels. Connected TV, or CTV, is a great next step. The scale of television, the targeting of digital, and an audience who are actually engaged is a compelling prospect. But, the gap between what CTV promises and what it can deliver is wider than many platforms will admit. This channel operates on fundamentally different rules than click-based, last-touch digital platforms like Google and Meta. Without understanding the CTV performance marketing challenges that you’ll likely encounter, you’ll be spending your way to a very expensive lesson. ## 5 Challenges Performance Advertisers Face with CTV Campaigns ### 1. The “Black Box” of Measurement and Cross-Device Attribution For anyone who’s built a reporting stack around Google Analytics or a Meta Pixel, CTV attribution can feel like a black hole. The core problem is that CTV ads run on shared household devices, can’t carry cookies, and conversions typically happen later, on a personal phone or laptop that has no direct connection to the TV where the ad ran. This is the central CTV measurement attribution challenge: An impression was served, someone may have seen the ad, then days later, someone in that household searches for the product and buys it. Was that the CTV ad working? Retargeting? Organic search? In most situations, it’s impossible to tell for sure. Closing the loop requires different infrastructure than most advertisers have in place. Cross-device attribution with CTV typically relies on IP address matching, connecting the household IP address that received the ad impression to the IP associated with a later web visit or purchase. Shared IPs, VPNs, or mobile networks, though, can all introduce noise to this kind of attribution tracking. More sophisticated approaches involve third-party measurement partners, identity graphs, or data clean rooms that match hashed user identities across different environments. These tools exist and work well, but they’re costly and complex to integrate. For a lean marketing team running across a handful of channels, creating an accurate CTV measure is a challenging task. ### 2. Inventory Fragmentation and the Frequency Cap Nightmare On Meta, frequency capping is quite easy to set up. On CTV, it’s almost impossible, because CTV isn’t a single platform, but a group of streaming apps, smart TV manufacturers, and programmatic exchanges, all of which are operating with their own identity frameworks and inventory logic, resulting in a fragmented ecosystem. If you’re buying ads across Roku, Fire TV, and Samsung, you have no native mechanism to recognize that the same household, and maybe even the same person, is receiving your ad on all of them. The result is a CTV frequency capping problem that’s both frustrating and expensive. A user who’s seen your 30-second ad ten times in one weekend isn’t going to convert at a higher rate; they’re going to develop brand fatigue, and you’re still paying for every one of those impressions. When your ad dollars need to justify themselves, this kind of invisible waste is detrimental to campaign efficiency and the final CPA. The best available solution is to buy ads through a centralized demand-side platform (DSP) with universal frequency logic. While this doesn’t completely solve the problem, it’s the best option as of now to attempt to manage frequency in a system that wasn’t designed to be managed that way. ### 3. High CPMs vs. Strict Performance Goals: CPA and ROAS CTV CPM vs. CPA is an incredibly difficult equation to balance. CPMs regularly run between $20 and $40, with premium inventory on top-tier streamers often going for higher prices. Comparing that to $5-12 CPMs on social display, this leaves CTV advertising with a much smaller margin of error. That cost is somewhat defensible, in that you’re buying space on a full screen, in a non-skippable environment. There is genuine value. But, for a brand trying to hit a $35 CPA on a $60 product, the conversion rate required to make most CPMs work is significantly higher than what many CTV campaigns can initially deliver. Understanding ROAS on CTV means accepting that you’re unlikely to see a direct response, like you would with a well-optimized search campaign. What CTV excels in is brand recognition and intent that makes search and social campaigns more effective later. But, that’s a very different value proposition to “run this campaign and measure the return.” Setting internal expectations with stakeholders is critical. The brands that make CTV work from a cost perspective tend to share a few similar traits — tightly defined audiences that reduce wasted impressions, creative that’s optimized for a living room format, and a measurement framework that captures downstream influence, rather than demanding last-click attribution immediately. ### 4. Granular Targeting Limitations vs. Search and Social Meta targeting is built around knowing exactly who the audience is. CTV is catching up to this, but it’s not there yet. This gap matters to performance marketers, who have become accustomed to precise targeting on social. The targeting layers available with CTV are usually demographic overlays, content categories, household income bands, and purchase intent signals. The ability to target someone here as precisely as on Meta is simply not possible: Contextual and audience targeting can get you close to the right viewer, but close costs money when CPMs are as high as they are on CTV. There’s also the practical friction of CRM onboarding. Matching your first-party customer data into a CTV environment involves extensive identity resolution steps that add lag, data loss, and sometimes a lower match rate. What uploads cleanly into Meta custom audiences can take days and a significant technical investment to activate meaningfully in a programmatic TV-challenged context. This certainly doesn’t make CTV unusable, however. Leaning into contextual alignment that matches your highest buying personas can compensate for some of this audience precision gap. But, it’s important to remember that you need to approach this with a different strategic mindset than with social. ### 5. Ad Fraud and Lack of Transparency Digital advertising has always had a fraud problem, and CTV has a fraud problem with a more complex supply chain and fewer detection measures in place. The architecture of CTV with Server-Side Ad Insertion (SSAI) creates opportunities for cybercriminals to fire ad impressions without a human ever actually seeing the content. Pixel stuffing, app spoofing, and bot traffic that mimics legitimate streaming behavior are all forms of CTV ad fraud that are difficult to detect at an impression level, and even harder to reverse once spend has already been committed to the ads. For performance advertisers, this isn’t just a brand safety concern, but also a direct ROAS problem. If 15% of your impressions are fraudulent, you’re not just wasting 15% of your budget, but also corrupting your attribution data, making your campaign look significantly less efficient than it actually is. This could result in you pulling back on spending that was actually working. Mitigating this requires active efforts, like vetting supply sources, working with DSPs that have a pre-bid fraud filter, insisting on app-level transparency rather than bundled inventory, and using third-party verification partners that specialize in CTV. ## Making the Case for CTV: Incremental Reach and the Full-Funnel Picture The strongest argument for CTV is incremental reach, delivering against audiences you can’t reach anywhere else. Cord-cutters and linear viewers are generally unreachable via traditional broadcast and increasingly resistant to mobile and desktop ad formats. CTV is often the only way to get a video ad in front of them in a premium, high-attention environment. Understanding CTV vs. linear TV for performance also clarifies the value proposition for CTV. Linear TV buys are proxies at best, as you’re buying a time slot and hoping that the right people are watching. CTV offers actual audience data, household-level targeting, and, at minimum, the infrastructure for post-exposure measurement. ## Key Takeaways CTV is a compelling channel for advertisers who’ve hit a ceiling with search and social, but it’s important to remember the challenges that come with it. The measurement gaps can be substantial and frequency management isn’t always easy across fragmented ecosystems. High CPMs can end up being costly, while fraud risk in the supply chain requires active mitigation on an ongoing basis. That said, for advertisers looking to make the shift, and who understand moving from immediate ROAS to downstream influence, CTV can bring real opportunities to expand into new markets beyond search and social. ## Frequently Asked Questions (FAQs) ### Can you track clicks on CTV ads? Not in the traditional sense, as CTV is mostly non-interactive, so there’s no click to measure. Instead, performance is tracked through view-through attribution, identifying users who were exposed to the ad on a CTV device and then visited the site or converted on another device. ### Why is CTV more expensive than Facebook or YouTube ads? CTV inventory runs against professionally produced long-form content like TV shows, movies, and live sports in a non-skip format on a large screen. Advertisers are paying for a higher audience attention with no risk of a below-the-fold placement, like with social and search. ### How do I control frequency across different CTV publishers? The most effective approach is consolidating buys through a single DSP that can enforce a universal frequency cap across its inventory. Layering in identity solutions that recognize the same household across different apps and publishers also helps, but some level of duplication is an accepted cost of using this approach. ### Is CTV good for bottom-of-funnel conversion campaigns? Generally, CTV is better suited for mid-and-upper funnels than direct response. Its real strength is in building awareness and purchase intent among audiences who are harder to reach elsewhere. Trying to use CTV as a standalone conversion channel typically results in a disappointing ROAS. --- ### Building High-Intent Ad Creatives for Home Services URL: https://www.taboola.com/marketing-hub/high-intent-ad-creatives/ Last Modified: 2026-04-12 07:47:21 The digital world we all live in is a busy one: consumers bounce quickly among search, social media, and favorite websites as they’re researching purchases. For home services companies, it can be daunting trying to figure out how to break through with ad creative to earn the attention of a new prospect. At Realize, we’ve found that success in home services advertising is all about getting the details right — attracting attention, engaging your prospect, and closing the deal with tailored images. Here’s how one customer worked with Realize home services advertising expert Jeremy Bade to cut out channels and tactics that weren’t delivering, and instead reach the right prospective customers for their product. ## Ditch the Agency Tactics and Build High-Intent Creatives To help amp up the work of a brand using Realize to boost its reach and conversions, the first step was delving into how elite home services brands scale their digital presence. It started with some marketing detective work: inspecting the tracking codes and landing pages of the U.S.’s top flooring companies to help this top-tier concrete coating company, which specializes in polyurethane and polyaspartic floors. The coating company was already spending heavily on Meta and YouTube ads, but struggled to scale Google Search due to industry categorization issues. Inspecting the tracking codes and landing pages of top flooring companies uncovered a useful piece of data: the biggest players in the industry were quietly winning on native advertising. To replicate the success of these companies, Bade recommended that the coating company move away from so-called black box agency tactics — where it’s often unclear where spending is going — and use the Realize platform to get truly specific with their audience and strategies. In this case, the plan was to build high-intent creatives to convert homeowners over age 40 into coating company customers. Here’s what Bade recommended: ### 1. Leverage Motion Ads to Satisfy Visual Intent To capture attention on premium publisher sites, you must move beyond static imagery. For this coating company, the work they do for their customers was a perfect fit to add some motion. The recommendation: Replace or supplement static finished product photos with motion ads (GIFs) that showcase the application process. The reasoning: In the home improvement vertical, the satisfying nature of seeing a product applied — like a smooth coating being poured — acts as a high-intent hook. “Pouring polyurethane is incredibly satisfying to watch,” says Bade. “You’ll get a much higher click-through rate and higher intent by showing the work in action, versus a static image of a finished floor that the user might just scroll past." ### 2. Use Dynamic Keyword Insertion for Hyper-Localization To make a national brand feel like a local specialist, personalization is key. The recommendation: Utilize dynamic keyword insertion, or DKI, to call out the user’s specific location directly in the headline. The reasoning: Homeowners are looking for local contractors. A headline that asks a question about their specific area feels like a local service announcement, rather than an advertisement. “Go ahead and add in the region, city, or zip code as a dynamic keyword insertion, it does really well on Realize,” suggests Bade. “If you say, 'What does a garage floor cost in ?,' it becomes hyper-targeted and hits closer to home for the consumer.” ### 3. Segment Campaigns by Device to Control Reach and Bid To ensure your budget isn't eaten up by low-intent clicks, you must control the environment. Realize makes this possible with segmented campaigns for better targeting. The recommendation: Duplicate your campaigns to isolate mobile traffic from desktop or tablet traffic, instead of running a catch-all campaign. The reasoning: Mobile cost per click (CPCs) are often lower, causing the ad platform’s algorithm to spend the entire budget there. However, desktop users on premium news sites often represent a high-value demographic for home services. “If you have all device types in one campaign, you’re not going to get any reach with desktop, because mobile is going to eat up all that budget,” says Bade. “For a sophisticated marketer, you can split those to optimize for the specific behavior of each user.” ### 4. Implement Tier 1 Blocklists to Ensure Brand Safety To avoid junk traffic and low-quality leads, aggressive placement filtering is mandatory. The recommendation: Move away from open-exchange blind buying and instead apply specific blocklists for non-converting categories, like gaming or kids’ content. The reasoning: High-ticket home services ($4,400+ average jobs) require a professional context. You want your ad next to a Bloomberg article, not a mobile game. “Realize makes sure you’re only showing up on those sites that are most likely to convert,” confirms Bade. “By putting blocklists in place early, we avoid the nonsense, like kids’ gaming sites, and keep you in premium environments where your demographic actually spends their time.” ## Key Takeaways Home services companies don’t have to start from scratch to build better ad creatives for high-intent audiences. A combination of the right tools and details on what the industry leaders are doing led to better targeting, better use of budget, and more high-intent audiences for a concrete coating company. The use of motion ads, hyper-localization, device segmentation, and blocklists all helped this Realize customer stop wasting money on search traffic. And, they added transparency and control so that native advertising could be a scalable, fruitful channel. ## Frequently Asked Questions (FAQs) ### How many creative variations should I test at once? Creative ad variations can be tested endlessly to find a winner — typically, a business might test as many as they have the time and resources for. Realize experts, though, recommend testing a maximum of 10 high-quality ad variations. More than 17 variations can dilute data, Bade says, and Realize’s capabilities work best when you focus on your top 10 so the algorithm can stabilize and find the winners faster. ### What is a realistic lead cost for home services? Cost per lead (CPL) varies wildly by industry and location, and the home services industry covers a broad range of categories. Estimates show CPL in home services can range anywhere from $45 to $228 in the U.S. That said, for home services like concrete coatings, Realize experts have found that a CPL between $40 and $80 is the sweet spot for scalability. Realize allows you to monitor this number daily, so you can move budget from underperforming social campaigns into high-performing native ones the moment you hit these targets. ### How long does it take to see stable results for high-intent home services ads? Typically, digital campaigns need a few weeks to learn the audience and see performance metrics. With Realize, the goal is a 7-10-day intensive monitoring period. Realize experts monitor the early learning phase swings to ensure that by the 30-day mark, your cost per acquisition (CPA) has stabilized and your set rates (appointments booked) are meeting your ROI goals. --- ### How to Scale High Ticket D2C Products with Predictive Audiences URL: https://www.taboola.com/marketing-hub/scale-high-ticket-d2c-products-with-predictive-audiences/ Last Modified: 2026-04-13 11:48:41 For direct-to-consumer (D2C) brands selling high-consideration hardware priced at $250 and above, the path to purchase is rarely a single click. A shopper might discover you on social, compare you against three competitors, watch a review, read a forum thread, ask a friend, and only then search for your brand name a week later. That’s where many performance programs stall. Social can spark interest, but often drives single-visit drop-off. Users click, skim, bounce, and disappear before they understand what makes your product worth the price. Search captures intent, but it typically captures already-formed intent, and for premium tech products, those auctions can be saturated and expensive. This is the mid-funnel gap, the phase where people are interested, but not yet convinced. In the home cinema and tech hardware vertical, bridging that gap is often the difference between decent ROAS and scalable growth. Below is a practical framework for scaling high-ticket D2C hardware using Realize and Predictive Audiences. Realize advertising sales specialist Ari Del Rosario shows you how to move beyond basic retargeting to find the early adopters and research-driven consumers who are statistically most likely to convert. ## Aligning Creative Strategy With Native Editorial Mindsets If your product costs $250 or more, your first touchpoint shouldn’t feel like a hard sell: It should feel like an invitation to learn. Native placements live inside an editorial environment, and users behave differently there. They’re usually in reading mode, getting ready to shift gears to buy mode. That’s a massive advantage for high-ticket D2C, because education is often the missing ingredient between curiosity and purchase. Instead of sending cold traffic straight to a product page, then, shift your first click to an advertorial landing page (or a video story built like one). The goal isn’t to hide that you’re selling something — it’s to deliver the value proposition in a format that matches the user’s mindset. A strong advertorial for premium hardware usually includes: - The “problem” with the category (cheap competitors, confusing specs, misleading claims). - A clear “why now” (what’s changed, what makes this product a disrupter). - Proof of credibility (testing, reviews, use cases, comparisons, warranties). - A simple path to the next step (email capture, quiz, “check compatibility,” or a buy CTA). “With a $250 price point, you need that mid-funnel education piece,” confirms Del Rosario. “On native placements, users are already in a reading mode. We recommend leveraging advertorials or video stories to guide the buyer through the journey, effectively acting as a digital infomercial that builds brand trust before they hit the ‘buy’ button.” ## Implementing Multi-Layered Audience Targeting Once creative matches the mid-funnel journey, the next step is making sure you’re reaching the right kind of learner — because not all clicks are created equal. For premium tech, broad interest targeting can drive volume, but it often brings in casual browsers who love reading about gadgets and hate paying for them. Scaling requires a smarter progression: start with what you know, then let the platform reveal what you need to know. A strong approach is: - Seed with your existing customer data (CRM) to build lookalike audiences. - Establish a conversion baseline (the “ground truth” for what success looks like). - Once you hit a meaningful threshold, layer in predictive audiences to expand efficiently. “We don’t just throw your money into a black box!” says Del Rosario. “We start with your existing customer data to build lookalikes, but the shiny new tool our partners love is Predictive Audiences. Once we hit that 100-conversion threshold, the system identifies the subtle signals of users most likely to pull the trigger, allowing us to optimize for high-intent buyers automatically.” ## Optimizing Through Full-Funnel Visibility Even with the right creative and targeting, scaling fails when teams can’t see what’s working fast enough, or when they optimize on the wrong signal. High-ticket D2C campaigns need full-funnel visibility because: - Top-line CTR can be misleading (curiosity clicks are common in tech). - CPA can swing based on device, landing page load time, and checkout friction. - Publisher quality matters more than raw volume when the product is expensive. One practical way to bring order to the chaos is to segment campaigns by mobile versus desktop early. High-ticket buyers often research on mobile and convert later on desktop, and performance can differ dramatically by device. Then, to move quickly through the platform’s learning period, use a Maximize Conversions bidding strategy, especially in the first 7–10 days, when the algorithm is trying to map which combinations of placements, users, and creatives produce your desired outcome. “If you win, we win,” says Del Rosario. “By using our Maximize Conversions bidding strategy, we can feed the algorithm the signals it needs to get through the learning phase quickly. This gives us the visibility to see exactly which premium sites — like Apple News or TechCrunch — are driving the highest quality traffic, so we can scale what works and cut what doesn’t.” ## Key Takeaways Scaling a premium D2C hardware brand is more than scaling traffic: It takes the right traffic, delivered in the right educational context, at the stage where decisions are actually formed. By pairing mid-funnel creative with a layered targeting plan, brands can bridge the gap between social awareness and search intent. Through the Realize platform, Realize SMEs bring the technical backing and strategic oversight needed to ensure budgets aren’t just spent, but invested in repeatable, scalable growth. ## Frequently Asked Questions (FAQs) ### What is predictive targeting? Predictive targeting uses machine learning to analyze vast amounts of real-time signals — such as content consumption patterns, device habits, and historical conversion data — to identify and reach users most likely to take a specific action before they even enter a traditional search funnel. Strategically, this allows you to move beyond static audience segments by leveraging AI to find high-value prospects across the open web who exhibit the same digital behaviors as your existing best-performing customers. ### How many conversions do I need to start using predictive targeting? To achieve statistical significance, you generally need a minimum of 50-100 conversions per month for the machine learning model to accurately identify the behavioral patterns that predict future success. Strategically, starting with a robust baseline of data ensures the platform’s AI doesn’t optimize toward ‘noise,’ allowing it to effectively scale your reach to high-probability prospects across the open web. ### Why should I use native ads for high-ticket hardware instead of just Search? Performance ads for high-ticket hardware allow you to educate and build trust through storytelling before the user has even identified a specific product need, effectively warming up cold audiences who aren’t yet active in search. This strategic approach captures a massive, incremental audience on the open web by positioning your hardware as a solution within contextually relevant editorial content, rather than competing solely for the limited volume of high-intent search queries. --- ### Optimizing Google Tag Manager for High-Volume Finance Campaigns URL: https://www.taboola.com/marketing-hub/optimizing-google-tag-manager/ Last Modified: 2026-04-12 06:59:48 In the fast-paced world of microcap stock promotion and financial services, data isn’t just a metric, it’s the engine of liquidity. However, many financial advertisers face a tracking black hole when dealing with complex landing pages, third-party brokerage links, and strict compliance rejections. For many finance advertisers running lead-gen campaigns, tracking breaks down exactly where it matters most: across complex landing pages, strict compliance rules, and the moment a user clicks out to a third-party brokerage site. This guide walks through a real-world scenario: a finance advertiser scaling multiple deals with $50,000 monthly test budgets, aiming to unlock a “sky is the limit” spend phase. Realize experts Brandon Jones (solutions engineer) and Anslynn Capps (advertising sales manager) used Google Tag Manager (GTM) plus Realize best practices to fix the tracking black hole, implement high-intent event signals, and create a clean measurement layer that can survive compliance scrutiny. ## The Pain Points: Ad Rejections for “Unverified Claims” Plus Tracking Blindness on Off-Site Brokerage Clicks In the finance vertical, ad rejections can feel like a black box. This particular advertiser was repeatedly flagged for “negative or unverified claims,” a common issue when landing pages include dense microcap language, performance-adjacent phrasing, and disclaimers that don’t surface clearly during review. At the same time, performance measurement was compromised by a second problem: conversions weren’t happening on the advertiser’s site. Their “success moment” was often a user clicking out to a third-party brokerage link. Once the user leaves the site, a standard browser pixel frequently loses visibility, creating a painful combination: - Compliance friction slows or stops spend. - Measurement gaps prevent optimization. - Lack of proof makes it harder to defend ads during review. - Scaling decisions become guesswork. Realize’s approach was to fix both sides of the problem: tighten the GTM foundation so events are trustworthy, then design a funnel that captures intent, not just page views. ## How to Overcome Financial Ads Rejections With GTM Optimization: 3 Strategies from Realize Experts ### 1. Auditing the GTM Container Hierarchy Before you add a single trigger, you need to confirm you’re working in the container that’s actually live on the site. In this use case, the client was building tags in a GTM workspace that wasn’t deployed in production. “Ghost tags” existed in the interface, but never fired on real traffic. This is surprisingly common with WordPress sites that have multiple GTM plugins installed over time, theme updates that overwrite header changes, staging versus production containers, and old containers hardcoded by prior agencies. #### Troubleshooting WordPress and GTM Integration Realize SMEs recommend a manual audit to confirm the snippet is installed correctly and consistently: - Inspect the WordPress theme header.php (or your site’s header injection method). - Ensure the GTM snippet is placed inside the <head> as intended. - Confirm there aren’t multiple GTM containers firing (yes, this happens). - If you’re using plugins, use something clean and explicit like WPCode (or an equivalent snippet manager) so the container placement survives theme changes. “You can have the most sophisticated tracking strategy in the world,” says Jones, “but if your GTM container ID doesn’t match the script live on your site, you’re flying blind. We always start by inspecting the network tab in DevTools to confirm that the container firing is the one we actually have access to.” ### 2. Implementing a Multi-Layered Conversion Funnel In finance, a page view is almost meaningless. What you really want is intent-to-buy behavior. With this in mind, Realize guided the advertiser to build a funnel that creates optimization leverage before the final off-site conversion. #### Defining “High-Intent” Events They implemented three core triggers: - The 60-Second Timer (deep consumption signal): A timer trigger at 60,000 ms distinguishes drive-by clicks from real attention, especially valuable in microcap flows where serious users actually read. - The Form Submit (lead capture): This is your obvious mid-funnel conversion, but in finance it’s also a quality gate: a form submit tends to correlate with genuine interest and can be used to train optimization. - The Brokerage Click (liquidity intent/off-site handoff): This is the most important bridge event. If you can reliably track the click-out to a brokerage platform, you can optimize toward the behavior that precedes liquidity, even when the final conversion is off-site. “In finance, success is measured by liquidity,” explains Jones. “By setting up a 60,000 ms timer and scroll-depth triggers, we can differentiate between a ‘bounced’ user and someone who is genuinely studying the stock’s potential. This data allows Realize’s algorithm to find more ‘readers’ who eventually become ‘buyers.’” ### 3. Bridging the Gap With Server-to-Server (S2S) Tracking Even with perfect GTM, browser tracking has limits, especially in finance: User journeys often end on brokerage sites you don’t control, call center follow-ups are used, and compliance-driven intermediate pages cause the standard browser pixel to lose visibility. That’s why we recommended moving from “pixel-only” measurement to server-to-server (S2S) for the final conversion signal. #### Using Click IDs for Offline Conversions The usual approach is to pass a unique click ID (from the ad click) through the journey, then post back the conversion from the server side once it happens. In practice, many advertisers use tracking platforms like Voluum or Bemob to capture the click ID, persist it through redirects, tie it to an eventual conversion event, and send a server-side conversion to the ad platform. “Finance ads are heavily scrutinized by content review teams,” says Capps. “By using server-to-server tracking, we don’t just solve the tracking issue, we gain the ability to prove the quality of the traffic to our review teams, showing that the ads are driving legitimate interest in verified financial services.” This matters because S2S does two things at once: - It restores measurement when the browser pixel can’t follow the user. - It creates auditability. You can support approval and review conversions with consistent, traceable data. ## Key Takeaways Setting up Google Tag Manager for a finance campaign requires more than pasting a code snippet: It requires strategic alignment between liquidity goals and the technical architecture of your site. Through Realize, Taboola’s SMEs helped finance advertisers ensure every dollar of a $50k+ test budget was measured with confidence: the right container, the right events, and the right post-click visibility. When tracking is right, optimization accelerates, and as the client put it, “the sky is the limit.” ## Frequently Asked Questions (FAQs) ### Why is my GTM tag not firing even though I see it in my workspace? The most common reason a GTM tag visible in your workspace isn’t firing is that the container version has not been published, meaning the changes are not yet live on your web server. Strategically, you must also ensure your trigger logic aligns with the actual DOM events or URL structures on your landing page, as performance campaigns often utilize dynamic elements that standard page-load triggers may fail to capture. ### How do I track conversions on pages I don’t own, like a brokerage site? To track conversions on third-party sites like brokerage platforms where you cannot place a pixel, you must implement server-to-server (S2S) tracking by passing a unique click identifier from the performance platform to the partner site’s CRM. Strategically, this allows the external site to post back conversion data directly to the platform’s API once a lead is qualified, ensuring your campaign’s AI continues to optimize for actual sales rather than just outbound clicks. ### My finance ads keep getting rejected for “unverified claims.” Can GTM help? While Google Tag Manager (GTM) cannot directly bypass compliance rejections, it can strategically manage the mandatory disclosures and fine print required to substantiate your finance claims without hard-coding changes. By using GTM to dynamically inject localized disclaimers or “Terms and Conditions” overlays based on the platform’s referral source, you can quickly align your landing page with the strict verification standards required to maintain active ad status. --- ### Three Strategies to Optimize Conversion Tracking: Financial Services Use Case URL: https://www.taboola.com/marketing-hub/optimize-conversion-tracking/ Last Modified: 2026-04-12 06:36:59 In the high-volatility world of microcap finance, traffic isn’t the goal — liquidity and investor intent are. That’s a crucial distinction, because the path from an ad click to a meaningful action (reading a thesis, downloading an investor deck, clicking out to a brokerage, or ultimately placing a trade) is rarely direct. If your measurement strategy only tracks a single buy or lead event, you’re likely optimizing toward the wrong users, and feeding your algorithm noisy signals that inflate spend without improving outcomes. That’s also why microcap advertisers run into a familiar set of problems: - Trust is fragile. Landing pages live under strict compliance scrutiny, and even minor copy issues can delay launch. - Bots are persistent. Form fills and low-friction leads can be bot-heavy, making cost per acquisition (CPA) look great, while downstream quality collapses. - Attribution breaks easily. Investor journeys often leave your site to a third-party brokerage or a presentation host, which is exactly where standard pixels lose visibility. That’s why the right performance marketing platform can make a big difference. This guide examines a real-world example in which a finance client partnered with Realize solutions engineer Brandon Jones and advertising sales manager Anslynn Capps, to solve pixel misfires, reduce bot-heavy lead data, and implement a full-funnel tracking architecture using Google Tag Manager (GTM) and server-to-server (S2S) integrations to optimize for real investor behavior, not accidental clicks. ## Architecting a Full-Funnel Conversion Path The most important shift for microcap finance, where transactions are often less than $5 per share, is moving from single-event tracking to a multi-layered conversion framework that measures what’s known as “investor warming.” The goal is to capture a progression of intent signals so the optimization engine has enough high-quality data to distinguish curious readers from likely investors. ### Capturing Soft Conversions to Fuel Algorithm Learning For microcap deals, the distance between an ad click and a stock purchase is wide. Even a high-intent investor may spend time validating claims, reading risk disclosures, or reviewing catalysts before taking a next step. That’s why “soft conversions” matter: they’re early signals that correlate with deeper intent. To that end, Jones recommends tracking behaviors like: - 60 seconds on page (a timer-based engagement event). - 30% scroll depth (a sign of real consumption vs. bounce). - Multi-page depth (e.g., landing page → thesis page → disclosures). These events serve two purposes: - They create more optimization signals (especially early in a campaign when hard conversions are limited). - They filter out a large share of accidental clicks and low-quality sessions that never engage. “We focus on which events drive the most ‘in-the-funnel’ value,” explains Jones. “By setting up a 60-second timer and scroll depth triggers, we give the campaign the signals it needs to learn who the actual readers are, not just the accidental clickers.” Use a timer trigger for the 60-second event, ensuring it only fires once per session/page, and a scroll depth trigger for 30% (or multiple thresholds, if you want finer granularity). Then, fire your conversion tags on those triggers and pass consistent metadata like the event name, page type, and campaign identifiers if applicable. ### Measuring Intent via Resource Downloads If direct purchase or trade data lives behind third-party systems, you can still measure bottom-of-funnel intent using proxy actions that strongly correlate with investor seriousness. Common examples include: - Clicks to investor presentation/deck downloads. - Clicks to brokerage links or “Where to Buy” modules. - Clicks to supporting diligence assets such as research notes, filings hub, and FAQ pages. These are meaningful because they represent a user choosing to stop passive reading and take an action that moves them closer to executing, which is often the strongest signal you can reliably capture on owned pages. Track these as click events with link targeting rules that are specific enough to avoid false positives, e.g., target URLs containing /investor-presentation or outbound domains that match known deck hosts or brokerage partners. Then map these to distinct conversion events in Realize so you can analyze them separately (and, if needed, optimize toward the best-performing proxy). ## Solving Attribution Gaps with Server-to-Server (S2S) The moment a user leaves your owned property, traditional pixel-based attribution becomes fragile. It can be blocked by browser controls, cross-domain limitations, or the simple fact that you can’t place code on the destination site. That’s where server-to-server (S2S) tracking becomes a finance-grade solution. ### Maintaining Visibility Beyond the Landing Page When a user moves from your landing page to a brokerage site or third-party presentation host, a standard pixel loses visibility. The Realize team’s recommendation is to use Click IDs to close the loop. At a high level, the architecture looks like this: - Realize click generates a unique click ID (passed via URL parameters to your landing page). - Your site captures and stores the click ID (cookie, local storage, or server-side session, depending on your setup and privacy requirements). - When the user completes a downstream action (for example, “deck viewed,” “account created,” “application submitted”), your server sends an S2S event back to Realize with that click ID and conversion details. - Realize uses the click ID to attribute the conversion to the originating campaign and optimize accordingly. “Server-to-server (S2S) tracking is the gold standard for finance,” says Jones. “By passing a click ID, we can optimize campaigns based on actual downstream actions — like viewing a deck on a third-party site — ensuring we optimize for lower CPA based on real investor behavior.” You’re no longer optimizing just for who clicked, you’re optimizing for clicks followed by a measurable action, even if that action happened outside your domain. That’s the difference between scaling spend and scaling results. ## Navigating Compliance and Technical Implementation Finance performance isn’t just about measurement design, it’s about execution under real constraints. Finance campaigns often use compliance-heavy templates, multiple tracking layers, and frequent copy revisions. That’s exactly where tracking breaks and launches get delayed, unless you approach implementation proactively. ### Overcoming the Negative Claim Rejection Finance ads are often rejected for unverified claims, especially when language implies certainty, or when disclaimers don’t align with policy expectations. Even negative claims (implying a user will lose out if they don’t act, or implying certainty about outcomes) can cause issues, depending on wording and context. “Finance is difficult because you can’t promise ‘getting rich,’” says Capps. “When a rejection happens, we take the guesswork out of it. We submit tickets directly to our policy teams to get the exact language that needs to be tweaked, so the client can launch without delay.” Realize experts recommend escalating quickly through the proper channels to isolate the root cause, whether it’s: - Boilerplate disclaimer language. - A specific landing page claim. - An implied promise created by a headline and subhead pairing. - A mismatch between ad copy and landing page copy. Treat compliance troubleshooting like a production blocker: The faster you identify exactly what needs to change, the faster you preserve momentum and prevent repeated rejections that can throttle scaling. ### GTM Container Audit One of the most common reasons conversion tracking looks broken is surprisingly basic: the GTM container you’re editing isn’t the one running on the site. This happens a lot in finance because organizations may have multiple agencies, legacy tags, staging containers, or region-specific templates. Realize experts often provide hands-on support to audit: - Header/footer scripts to confirm the live GTM container ID. - Tag firing on all relevant pages, not just the homepage. - Trigger logic (ensuring events fire once, on the right conditions). - Network calls confirming conversion requests are sent successfully. A practical checklist to avoid these problems includes: - Confirming the GTM container ID in your page source matches the container you’re publishing. - Using browser tools (e.g., Network tab) to verify requests are firing on trigger conditions. - Validating that “soft” conversions don’t overfire (inflating counts and corrupting optimization). - Ensuring conversions are mapped correctly inside your performance software, with the right event names and right funnel stage. ## Key Takeaways Success in the finance vertical requires more than budget, it requires technical infrastructure that can separate bots from real investor intent and preserve attribution when users leave your site. When you combine the elements below, you give your platform’s optimization engine what it needs to find higher-quality users, reduce wasted spend and scale with confidence: - A full-funnel conversion architecture (soft + proxy + hard signals). - Clean implementation via GTM (with audited containers and reliable triggers). - S2S attribution (using click IDs to track downstream actions). As the client in this scenario noted: once the tracking is solid, the “sky is the limit” for spend and return on investment (ROI). ## Frequently Asked Questions (FAQs) ### Why is my conversion tracking not showing any data in the dashboard? The most common culprit on the open web is a broken chain caused by aggressive browser ad-blockers, or a missing Google Click Identifier (GCLID) parameter during redirects. The fix usually involves moving from client-side pixels to server-to-server tracking, which ensures conversion data reaches your dashboard even when the user’s browser blocks standard scripts. ### How do I track conversions on pages I don’t own (like a brokerage site)? Since you can’t place a pixel on a third-party brokerage site, you should track the outbound click to the broker as a proxy conversion or, ideally, use a postback URL if the brokerage platform supports affiliate-style attribution. This allows the external platform to ping your server once a trade or sign-up is completed, closing the loop on your media spend. ### What is the best “conversion” to optimize for in a finance campaign? For microcap campaigns, where final “Buy” actions are infrequent, you should optimize for a high-intent micro-conversion like “Stayed on Page > 60 Seconds” or “Clicked ‘Join Mailing List.’” Training the AI on these high-volume signals allows the algorithm to learn much faster than waiting for the rare data of an actual stock purchase. --- ### Repurpose Facebook Ads to Scale Your Performance Without Reinventing the Wheel URL: https://www.taboola.com/marketing-hub/repurpose-facebook-ads/ Last Modified: 2026-04-12 06:19:46 Performance marketers often feel trapped in a cycle of creative exhaustion, the sense that scaling into new channels automatically requires new formats, new messaging, and a brand-new production cycle. For high-growth brands in consumer electronics and health, the path to incremental scale isn’t to start from scratch, but rather to leverage what’s already working well. Industry experts working with performance-driven brands consistently see the same pattern — high converting Facebook and Instagram ads are treated as channel-specific assets, even though the creative insights behind them are largely channel-fluid. By extending those proven assets beyond the social feed and into performance campaigns on the open web, brands can bypass the saturation and rising costs of closed platforms while reaching users in a more attentive, research-oriented moment. The subject-matter experts in this post work most closely with consumer electronics and health and wellness advertisers who are navigating this transition and reflect this strategy — that scaling doesn’t require reinvention, but a smarter distribution of your best creative work. Read on for tips and advice from advertising sales manager Ari Del Rosario, advertising sales team lead Nick Bonanni, and senior advertising account manager Remus de Jesus, all Realize experts. ## Advantages of Repurposing Facebook Ads to Scale Performance: 3 Use Cases ### 1. Extend the Life of High ROAS Social Creative in Consumer Electronics In consumer electronics, success on Meta often follows a predictable arc: A hero product finds traction, creative testing yields a winning combination, and return on ad spend (ROAS) climbs past the 4x mark. At that point, the instinct is to push harder, increase budgets until frequency spikes, and performance begins to erode. Creative fatigue sets in, not because the message stops working, but because the audience has simply seen it too many times in the same environment. Rather than waiting for that decline, experienced performance marketing teams recommend extending those winning assets into new contexts while results are still strong. Electronics purchases, especially those over $250, rarely happen on impulse: Buyers research specifications, read reviews, and compare options across multiple sessions. That makes the open web, where users are actively consuming editorial content, a natural next step. A U.S.-based smart projector brand encountered this exact scenario. Their Meta campaign delivered impressive results early on, but returns lessened as feed saturation increased. Instead of cycling through new creative concepts, the team kept the visuals exactly the same and instead changed only the environment that they appeared in. The brand deployed those same high-engagement assets into a split-funnel strategy, using mobile placements to spark interest and desktop environments to support the final purchase decision. “Our most successful partners don’t reinvent the wheel when they join Realize — they bring their winners with them,” says Del Rosario. “For a product like a projector, we take the video assets that already have high social engagement and place them in front of users reading tech reviews or home improvement blogs. By mirroring your social creative on the open web, you build brand consistency while reaching the 25% of the internet that isn’t on social media.” The result is not just incremental reach, but alignment with user intent. When the same creative appears alongside relevant editorial content, it feels less like an interruption and more like a continuation of the research process, increasing the likelihood that users progress toward a conversion. ### 2. Pivot Health and Wellness Offers Using Existing Lifestyle Assets Health and wellness advertisers operate in a far more volatile environment. Regulatory constraints, platform scrutiny, and competitive saturation can disrupt performance overnight. A campaign that was scaling smoothly can stall due to factors entirely outside the creative itself, from stricter approval processes to increased friction in gated experiences. When this happens, waiting weeks for new creative direction can be costly. Instead, high performing teams should look to their existing asset library for flexibility. Lifestyle imagery that performed well on Instagram, often secondary to more clinical or product-focused ads, can become the foundation for a rapid pivot. One wellness brand faced sudden traffic drops when users abandoned complex medical questionnaires required for their primary offer. Rather than pause spend entirely, the team shifted budget toward a lower-threshold vitamin product that allowed straight-to-purchase conversions. They reused their best-performing lifestyle images, maintaining brand consistency while removing the friction that had stalled growth. “Performance marketing moves too fast for you to wait on a new creative team every time you pivot,” notes Bonanni. “If your medical-grade ads are stalling, grab your top-performing lifestyle shots and immediately point them at a lower-friction supplement offer. It’s about working smarter with the assets you already have in your library.” By pairing familiar visuals with a simplified offer and distributing them through performance campaigns on the open web, the brand was able to stabilize spend and preserve momentum without sacrificing compliance or brand cohesion. ### 3. Hyper-Target Niche Audiences With Social-Verified Assets Another often overlooked advantage of repurposing Facebook creative is that social platforms themselves can be helpful testing grounds for future work. Assets that achieve strong click-through rates (CTR) or engagement have already been validated by these platforms at scale, but treating those outcomes as final, rather than foundational, leaves significant value on the table. This becomes especially important for brands targeting niche or sensitive segments. One health brand working with a polarizing ambassador faced limitations when trying to scale broadly on social. While certain creative performed well, audience expansion risked backlash or inefficiency. The solution was to disconnect creative validation from audience expansion. The team selected product-focused images — simple bottle shots that had already proven effective on social — and applied them to tightly defined first-party segments across the open web. By targeting specific states and high-net-worth demographics, they preserved relevance while expanding reach beyond the confines of a social feed. “We help brands scale by taking the guesswork out of creative,” says de Jesus. “If an asset has a 3% CTR on social, it’s a proven winner. We take that social-verified asset and use Realize’s predictive targeting to place it in front of the exact demographic — whether that’s specific U.S. states or high-net-worth individuals — ensuring your existing creative works harder for every dollar.” The broader takeaway is that creative testing and audience targeting don’t need to happen in the same place: By separating the two, brands can gain more control over how and where their stronger assets perform. ## Key Takeaways Scaling your brand doesn’t have to mean ballooning creative budgets or constant reinvention: The most efficient growth often comes from redistributing what already works. Performance campaigns on the open web offer a way to extend the reach of high-performing social assets into environments that align more closely with research, consideration, and intent. Whether navigating the long decision cycles of consumer electronics or the fast pivots required in health and wellness, the principle remains the same: Your best creative deserves a bigger stage, not a complete rewrite. With the right strategy in place, those assets can carry your brand further. ## Frequently Asked Questions (FAQs) ### What are the best advertising channels to expand to after Facebook? To expand a digital presence beyond Facebook, many advertisers traditionally look toward high-intent search platforms like Google or Microsoft Ads, as well as visual discovery platforms like TikTok and Pinterest. Each offers access to different user behaviors, from active search to inspiration-driven browsing. For performance campaigns on the open web, expanding beyond Facebook to high-intent search environments and native discovery platforms means that users can be reached while they’re still consuming content. This allows advertisers to diversify audience reach across premium publisher environments, while leveraging platform-specific AI to maintain the conversion efficiency already gained on social media. ### Does social creative perform differently on news sites? Users on social platforms are typically scrolling quickly, while users on news and editorial sites are actively reading content. That shift in mindset can influence how ads are perceived, but it doesn’t mean that social creative is underperforming. In reality, social-style ads, especially those using lifestyle imagery or motion, often stand out more in editorial environments because they break up dense content in a natural way. When deployed through performance campaigns on the open web, static social images can even be enhanced with AI-driven motion, helping them feel native to the page while capturing attention, all without disrupting the user’s reading experience. ### What are some common mistakes when scaling Facebook ad campaigns to new channels? A frequent mistake when scaling Facebook campaigns is assuming that higher budgets alone will unlock growth, without accounting for how different environments affect user behavior. On the open web, aggressive scroll-stopping creative can clash with more authoritative, information-seeking mindsets of readers on premium publisher sites. Another common pitfall is failing to adjust attribution and bidding strategies. Unlike social platforms that offer rapid feedback loops, performance campaigns on the open web involve longer consideration cycles and benefit from more advertorial-style content. Brands that account for these differences and adapt accordingly are more likely to see sustainable results when extending their strongest Facebook assets into new channels. --- ### 8 SaaS Marketing Trends for 2026: The Latest Tools to Spur Growth URL: https://www.taboola.com/marketing-hub/saas-marketing-trends/ Last Modified: 2026-04-12 09:28:50 The average business uses 305+ SaaS applications (software as a service, or subscription-based software). Small or medium-sized businesses (SMBs) probably use fewer — maybe 30 to 50 — but even that number can feel unmanageable. Too many tools can lead to technical debt: outdated software, siloed data, and frustrated employees struggling with systems that don’t (or can’t) talk to each other. SaaS was built on a simple premise: to make software easier to manage. For the most part, it’s over-delivered. Not long ago, upgrading a word processing program or contact database created a logistical nightmare. It required sneakernet IT — tired specialists literally walking from desk to desk to manually install new versions on every individual machine. If that software was buggy, the fix was just as painful. You waited weeks for a physical disc to arrive by mail before IT began its cycle of manual installs again. The cloud revolution of the 2000s was only the beginning, and today, the landscape has shifted again. Now, we don’t wait for updates: CI/CD (continuous integration/continuous deployment) and evergreen management facilitate daily background software updates with zero downtime. Modern SaaS is becoming agentic, so instead of you learning the software, the software uses AI agents to understand your goals and execute tasks on your behalf. Deployment strategies like canary releases mean that 1% of new users test new features first. If errors appear, the system automatically rolls back before the other 99% see it. 2026 SaaS is (at its best) bug-free and predictive. AIOps (artificial intelligence for IT operations) can identify and patch vulnerabilities or performance lags before a human user even notices a problem. With all that said, then, what are the top SaaS trends for 2026? I’ve laid out the eight you should know below. ## Trend 1: AI Evolves from Accessory to Infrastructure In 2025, AI was the shiny new toy of the SaaS world: experimental, exciting, and somewhat chaotic. The novelty has evaporated in 2026, leaving behind something even better: invisible infrastructure. AI isn’t a feature you choose to use. Now, it’s the engine powering each stage of the B2B journey, from initial keyword research to lifecycle retention. According to a SaaS Capital survey on AI adoption among SaaS companies: - 76% use AI in their products. - 69% have deployed AI solutions in daily operations. - 88% using AI in daily operations also use it in their product. - 92% planned to increase AI use in 2025 (and likely beyond). ### The Rise of the Agentic Workflow The most significant shift this year is the transition from generative AI (which writes) to agentic AI (which acts). Advanced SaaS teams are now deploying autonomous agents that suggest content and manage entire workflows. These agents can handle onboarding sequences, lead scoring, and partner activation with minimal human oversight. #### Why it Matters for SaaS Marketing This shift creates a massive competitive advantage. Lean marketing teams can scale their operations without a corresponding jump in headcount. The value proposition for customers changes, too. Your software is more than a simple tool; it’s an intelligent partner actively working on their behalf. ### Precision Over Speed: The New Content Standard Early adopters used AI for speed, often resulting in a sea of generic, surface-level content. Today’s winners are using AI for precision. - Content scoring: AI models can predict which assets will resonate with specific buyer personas before they’re published. - Predictive analytics: AI interprets natural language processing (NLP) queries to make platforms more intuitive and accessible, turning static databases into conversational partners. - Data-driven authenticity: The challenge marketers face isn’t whether to use AI, but how to balance automation with human thought leadership. The most successful brands use AI to handle the drudge work of SEO and formatting, freeing human creators to focus on original insights and brand voice. ### Practical Takeaways To stay ahead, B2B marketers must move beyond basic chatbots and integrate AI into the core of their strategy. - Deploy vertical-specific agents: General AI is out; niche AI is in. Choose agents tailored to your specific industry (e.g., fintech or healthcare) to ensure the insights you provide are relevant to your unique buyer intent. - Focus on retention metrics: Use AI to identify at-risk users during onboarding. By predicting churn before it happens and triggering automated, personalized interventions, brands are seeing up to a 5x reduction in acquisition costs. - Human-in-the-loop strategy: Build a workflow where AI handles the data and original drafts, but human editors provide the final authenticity (and accuracy) check to help your brand stand out in an AI-saturated market. ## Trend 2: Hyper-Personalization is the New Baseline Persona-based marketing has gone the way of the dinosaur. SaaS buyers rarely respond to broad categories like “marketing manager in mid-market tech.” Instead, they expect an experience that adapts in real-time to their specific intent, product usage, and current adoption stage. Generic outreach has become invisible. Research shows that personalized CTAs and tailored experiences now convert 202% better than generic ones. The moral of this story? Hyperpersonalization is a mission-critical feature. ### The Move to Predictive Experience Design Forget the {First_Name} era. Today’s hyper-personalization is predictive: your software and marketing anticipate what users need before they request it. How? By real-time data orchestration that analyzes: - Behavioral signals: If a prospect views your enterprise pricing page five times, your website automatically swaps out “Getting Started” guides for “Enterprise Security & Compliance” case studies. - LLM-driven intent: AI agents analyze the why behind a search. Someone looking for retention frameworks receives different nurture content than someone searching for automated onboarding tools, even if they share the same job title. - Contextual awareness: Messaging adapts to external signals, like recent company news, industry-specific market shifts, or even a user’s local time and device. (Source: Azarian Growth Agency) ### Account-Level Orchestration (ABO) For B2B SaaS, an individual is rarely the only decision-maker. Personalization has scaled to the account level this year. AI systems coordinate consistent messaging across the entire buying committee (typically 6-13 stakeholders). If the CTO is concerned about security and the CFO is focused on ROI, your AI-native platform ensures that each stakeholder sees a personalized version of the same value proposition delivered via their preferred channels (e.g., a LinkedIn ad, email, or in-product notification). #### Why it Matters for SaaS Marketing By shifting to signal-responsive journeys, brands are seeing a 70% boost in conversion rates within the first year. This precision reduces marketing waste by only directing ad spend to accounts showing high-intent research behavior. ### Practical Takeaways Hyperpersonalization is only as good as the data feeding it. Audit your data infrastructure: - Consolidate first-party data: Break down silos between your CRM, marketing automation, and product usage data to create a single source of truth. - Implement adaptive web layers: Use tools that allow for dynamic landing pages. Visitors from different industries shouldn’t see the same hero image or headline. - Prioritize next best action logic: Use AI to suggest the most logical next step for each user — perhaps a specific feature tutorial for a new user or an upsell offer for a power user — which will boost retention and lifetime value (LTV). ## Trend 3: The Rise of Value-First, Content-Led Growth In the pre-cloud world, software was a one-off purchase: You bought the disc, used the code, and that was that. Today, SaaS requires an ongoing commitment, and the free trial isn’t enough to win over a cautious market. Today’s most successful brands have moved away from polished, sales-heavy campaigns toward value-led content. Buyers expect tangible proof of value before they consider a subscription. Enter the hidden buyer: non-primary users in finance, legal, and operations who hold significant veto power over B2B deals. To reach them, marketing must prioritize credibility and educational depth over razzle-dazzle. According to the 2025 LinkedIn B2B Marketing Benchmark, 94% of marketers agree that trust is the primary currency for B2B success. (Source: 2025 LinkedIn B2B Marketing Benchmark) ### Educational Content Buyers are bypassing product pages and flocking to authoritative guides that solve actual problems. Whether it’s “How to scale a remote DevOps team” or “Managing HIPAA compliance in 2026,” educational SEO is a magnet for intent-rich traffic. Savvy teams use AI to repurpose one long-form guide into a LinkedIn carousel, a 30-second TikTok explainer, and a series of newsletter deep-dives. ### Thought Leadership Thought leadership isn’t a C-suite ego play, it’s become a cost-effective acquisition tool. The 2025 Edelman-LinkedIn Thought Leadership Impact Report found that 81% of buyers say that high-quality thought leadership helps them recognize previously overlooked business opportunities. Data from FirstPageSage shows that thought leadership SEO has one of the lowest customer acquisition costs (CACs) in the industry, averaging $647 for B2B SaaS, compared to the $982 average for LinkedIn ads. (Source: FirstPageSage) ### Actionable Steps To close complex deals in 2026, arm your internal champions with content that helps them sell for you. - Stop selling, start solving: 95% of hidden buyers say thought leadership is more effective than sales materials at demonstrating a vendor’s potential. - Visual overload: People remember 65% of visual content vs. only 10% of text. Use bite-sized explainer videos and infographics to simplify complex technical concepts for non-technical stakeholders. - Prioritize human credibility: 75% of B2B marketers have increased budgets for partnering with influencers (like industry creators and subject-matter experts) to humanize their brand and bypass the AI-generated noise. (Source: 2025 Edleman-LinkedIn B2B Thought Leadership Impact Report) ## Trend 4: Community as the Primary Growth Engine In 2026, the era of one-way, polished brand broadcasting has ended. As AI-generated noise saturates every digital channel, buyers have developed a “filter” for traditional marketing. Instead, they’re turning to peers. Over 90% read online reviews, and 73% only trust recent reviews (those within the past month). Thus, SaaS marketing is continuing its transition from a support-led model to a community-led growth (CLG) strategy. According to Travis O’Leary, director of digital transformation at Candescent, “The institutions that win over the next decade will not be the ones with the most locations, but the ones with the most intelligent, adaptable, and customer-centric digital foundations.” ### The Rise of Micro-Communities Successful SaaS companies no longer view Slack groups, Discord servers, or niche LinkedIn collectives as “extra” support channels. Instead, these communities have become retention and referral powerhouses. These micro-communities offer space for customers to solve problems in real-time, share industry hacks, and celebrate wins. #### Why it Matters for SaaS Marketing By shifting the focus from user to member, brands create a sense of ownership. Digital advertisers can amplify this change by: - Directing high-intent leads into these private communities, rather than just to a landing page. - Repurposing the most helpful user-generated content (UGC) into performance ad campaigns. - Using community engagement data to identify at-risk users before they churn. (Source: BuddyBoss) ### Activating Your Unpaid Sales Force Happy customers make the best sales reps. Social proof has evolved beyond the static, generic testimonial: Today’s buyers expect unfiltered advocacy, like transparent metrics, data-backed success stories, and open-product roadmaps. Marketing teams are using long-term behavioral data to identify power users (those whose loyalty and activity suggest they’re ready to influence others). By activating these advocates through incentivized reviews, customer-led webinars, and expert Q&As, brands can create programmatic campaigns that feel authentic rather than promotional. #### Practical Takeaways Authenticity is the ultimate differentiator in an AI-saturated market. To build a credible community today: - Grant exclusive access: Invite your top 5% into a private beta group or customer advisory board. - Amplify user voices: Don’t just collect testimonials; give your advocates a platform. Let them lead your next training session or co-author your next industry report. - Embed social proof everywhere: Move your customer stories out of a case studies tab and embed them directly into the product experience and sales journey. ## Trend 5: The Shift from Reporting to Prediction Marketing without analytics is like flying blind — a financial risk SaaS can’t afford. With acquisition costs rising and competition intensifying, gut-feeling strategies are a bad idea. Success requires moving beyond traditional reporting (what happened) toward prescriptive analytics (what we should do next). According to László Attila, founder of Pixel & Prompt, “The marketers who master prediction today will own performance tomorrow.” ### The metrics that matter in 2026: - Net revenue retention (NRR): High-growth companies are hitting 120% to 130% NRR, proving they can grow through expansion without constantly hunting for new brands. - Customer acquisition cost (CAC) payback period: In a capital-efficient market, the goal is to recoup acquisition costs within 12 months. - Burn multiple: This metric measures how much you’re spending to generate each dollar of ARR. A multiple below 2.0 is the gold standard for sustainable growth. - Activation and intent signals: Tracking “Aha!” moments and high-intent behaviors (like visiting a pricing page three times) to trigger human sales intervention at the right moment. (Source: Visdum) ### The AI Advantage While still a useful tool for crunching numbers, AI has become an active part of the marketing team. According to recent B2B benchmarks, AI-related skills are the fastest-growing digital requirement for marketers, as teams move from exploratory use to advanced execution. How it’s changing the game: - Prescriptive insights: AI flags at-risk accounts for churn before they leave and recommends specific retargeting schedules based on historical success patterns. - Automated experimentation: Instead of quarterly campaigns, AI allows teams to run thousands of micro-tests in real-time, optimizing ad bids, creative assets, and email timing autonomously. - First-party data mastery: As third-party cookies vanish, your internal data (usage patterns, support chats, and CRM history) is your most valuable asset. AI analyzes these vast datasets to surface audience segments that humans could easily miss. ### Action Steps Close the capability chasm. The gap between teams using AI-driven analytics and those stuck in manual reporting is becoming exponential. To bridge it: - Shift to always-on optimization: Transition away from static reports. Implement dashboards that provide real-time visibility into your qualified pipeline and win rates. - Instrument intent signals: Connect your ad platforms to your CRM intent data. When an account shows research behavior, your budget should shift automatically to prioritize them. - Invest in data literacy: The primary barrier to AI success is the skills gap. Train your team to move toward — and embrace — strategic system orchestration. ## Trend 6: Vertical SaaS Previously, the SaaS story included horizontal giants like Slack or Salesforce — platforms designed for everyone, regardless of industry. While these tools remain important, 2026 marks a decisive shift. Vertical SaaS, software surgically designed for the nuances of a specific niche, is outpacing generalist platforms. ### Tailored Architecture vs. One-Size-Fits-All Unlike horizontal solutions requiring heavy customization or awkward workarounds, vertical SaaS can speak the “native” language of its sector. Whether healthcare, construction, or legal services, these platforms arrive tailored to the workflows, regulatory requirements (like HIPAA or SOC 2), data structures, and customer expectations of a specific need — be that performance marketing, personnel management, or workplace benefits administration. That specialization means companies need less customization out of the box, and users typically see faster onboarding, better adoption, and more actionable insights. (Source: Oubit) ### Why the Shift? Businesses are realizing that generic tools, while flexible, often lead to feature sprawl and implementation fatigue. In contrast, Vertical platforms deliver a better fit with less effort. A construction firm may choose a system that manages tasks and integrates supply chain logistics, on-site reporting, and project timelines. A healthcare clinic needs a platform that bakes HIPAA compliance and patient record management into the code, rather than adding it as a third-party plug-in. In marketing, the latest vertical tools deliver industry-specific KPIs like creative fatigue and audience-specific conversion trends that generalist analytics often miss. #### The Power Players: Vendors as Acquirers The narrative of big tech swallowing small niche players has flipped. In 2026, vertical SaaS vendors are the primary acquirers. They’re not necessarily expanding into new industries because they’re digging deeper into their own. Investors have reset their expectations, prioritizing mission-critical ownership. Instead of a construction SaaS company trying to sell to lawyers, it’s buying niche tools for adjacent problems within construction (think: compliance training or workforce planning) to create a unified, end-to-end industry suite. Strategic shifts driving this trend include moving from a single tool to a unified platform to manage the entire business cycle. By owning every mission-critical workflow, vertical vendors make switching costs so high that they effectively dominate their niche. ### Practical Takeaways For B2B marketers in the vertical space, the goal for 2026 is dominance through depth. - Market the ecosystem: Shift your messaging from what the tool does to how it manages the entire industry workflow. - Highlight regulatory ease: For sectors like finance or legal, lean heavily into how your platform automates compliance, a major pain point that horizontal tools often miss. - Prioritize M&A messaging: If your company is acquiring niche tools, focus your marketing on the unified experience to show customers they don’t need to stitch together a dozen different apps. As one Reddit commenter noted, “It seems to me that the real issue should be, ‘What’s the use case, and does it really create value for the user?’ Whether horizontal or vertical, customer adoption will depend on how useful it is in creating value. It might be productivity enhancement, higher customer retention, or revenue growth.” ## Trend 7: Trust, Privacy, and Security as Marketing Assets Security and privacy have moved beyond IT checkboxes and legal hurdles to become the primary differentiators for closing B2B deals. As enterprise buyers become more risk-averse, trust has evolved into a tangible marketing asset. It shortens sales cycles and justifies premium pricing. Brands transparent about their data stewardship win; those treating it as an afterthought are increasingly excluded from consideration. ### The Rise of Privacy-First Personalization The uncertain future of the third-party cookie has forced everyone to embrace zero-party data (data users voluntarily share) and privacy-first personalization. Customers are willing to exchange their information, but only when the value is clear, and they maintain control. #### Why it Matters for SaaS Marketing Transparency has become a core part of the product story. About 73% of organizations say the most difficult aspect of managing security is getting visibility into security risks in business-critical SaaS apps. Leaders don’t bury privacy policies anymore — now they highlight security as a frontline feature. Over 70% of consumers report trusting AI less than they did a year ago. Winning brands combat this reticence by publishing explainability docs showing exactly how their AI makes decisions and which datasets the company used for training. Another powerful counter-narrative in an era of mass data hoarding? Marketing your data minimization policy and collecting only what you need. (Source: Achieve Unite) ### Security Accelerates Sales For B2B SaaS, security certifications (SOC 2, ISO 27001, HIPAA) appear on the landing page. Marketing your security posture in 2026 is as important as marketing your UI. Other emerging trust signals this year include: - Real-time threat visibility: Some SaaS platforms now offer customers a live security dashboard that displays anomaly detection and encryption status. - The ESG and sustainability link: Today’s savvy buyers track AI spend management and the environmental impact of data processing. Sustainability is shifting from a moral claim to an operational metric, with founders tracking ESG factors to ensure long-term viability in regulated markets. - Consent-driven workflows: Instead of aggressive gated content, brands use progressive profiling. They gather data gradually through helpful interactions that build relationships, rather than create database entries. ### Action Step Don’t wait for the security audit to discuss trust — make it a central pillar of your 2026 story. - Market your certifications: Treat a new SOC 2 TYPE II or GDPR update as a major project launch. Create a trust center where prospects can easily find, verify, and share these documents with their legal teams. - Be open about AI limitations: In your product demos, clarify where it uses AI and where humans stay in the loop. Explainability is an effective way to convert AI from a mysterious black box into a strategic partner. - Audit your small signals: Inconsistent email signatures, broken links, or non-secure URLs erode trust faster than a slick ad can build it. Standardize every touchpoint to reflect a security-first (and detail-oriented) culture. ## Trend 8: Ecosystem-Led Growth (ELG) is the New Moat The cost of direct customer acquisition has become prohibitively high for many. As traditional paid channels saturate and outreach response rates plummet, SaaS leaders are pivoting from solo sales to ELG. Instead of a linear sales motion, ELG leverages a network of technology partners, agencies, and consultants to drive high-intent leads at a significantly lower CAC. Mike Nevin, managing director at Alliance Best Practice Ltd., says that, “There is no doubt in my mind that ELG should be at the forefront of every SaaS CEO looking to grow his or her company.” ### Moving to Multi-Partner Clusters The isolated partnership manager no longer exists. Today’s deals are rarely won by a single company, they’re shaped by clusters of partners — an independent software vendor (ISV) plus a system integrator (SI) and a consultant — who collaborate to solve a complex business outcome. #### Why it Matters for SaaS Marketing A strong partner ecosystem offers a sales channel and a defensible moat. Competitors can copy your features, but they can’t easily replicate a web of 50+ deep technical integrations and a bunch of loyal consultants who recommend your tool daily. Leads sourced through partner introductions close 50% faster and have higher win rates because they inherit the trust the partner has already earned. Marketing teams have begun optimizing for “alternative to + ” searches, capturing users specifically looking for tools that play nicely with their existing tech stack. (Source: Achieve Unite) ### Strategic Partner Selection More isn’t better: Your playbook should focus on high-engagement tiers. Successful brands are reallocating resources away from hundreds of passive referral links to a few dozen strategic partners who align with their ICP and business model. Partnerships have evolved from referral handoffs to unified workflows. Sales teams use platforms like Crossbeam or Reveal to map overlapping accounts in real-time and coordinate co-selling strategies. Brands are winning partner mindshare by providing tailored support across the entire lifecycle, not just at the point of sale. This assistance includes co-branded demand generation, shared Slack channels for deal support, and joint success planning. ## Key Takeaways In 2026, SaaS marketing has evolved into a high-stakes discipline where trust, precision, and ecosystem depth are the big differentiators. Lead volume isn’t the greatest measure of success; now it’s NRR and the ability to turn security, privacy, and community into tangible growth assets. By shifting from broad horizontal reach to vertical specialization and predictive, intent-based journeys, marketers drive value beyond the initial click. The ultimate winners? Those who recognize and capitalize on the value of their partner networks and customer advocates. ## Frequently Asked Questions (FAQs) ### What are the most effective digital advertising channels for SaaS in 2026? Native advertising is out. Performance advertising is in. High-intent channels like Google Search, LinkedIn Ads, and even Reddit dominate because they allow for precise, real-time optimization against revenue outcomes like CAC and NRR. In a privacy-first world, these platforms win by turning behavioral signals and first-party data into predictable, high-performing growth engines. ### How can SaaS companies reduce churn through digital marketing? SaaS companies can reduce churn by using predictive analytics to identify at-risk users and then triggering automated, personalized win-back campaigns or educational content to reinforce the product’s value. Marketing teams can use in-app messaging and SMS to guide users toward their “Aha!” moment and foster long-term stickiness through exclusive, community-led engagement. ### What are some successful examples of SaaS digital marketing campaigns? Brands like Mailchimp transform user interaction data into actionable content to help marketers build better relationships. Canva treats user generated data and design trends as content that’s easily customized, shared, and integrated. Veema (pharma) and Procore (construction) use vertical SaaS to address niche regulatory and operational pain points that horizontal tools can’t touch. ### How is AI impacting the SaaS marketing landscape? In 2026, AI has evolved from a creative assistant into an autonomous, agentic engine that executes workflows like lifecycle sequencing and real-time churn prevention. It analyzes intent signals to deliver hyper-personalized content at scale across the entire buyer journey. Its expanded role allows lean marketing teams to achieve massive leverage, turning SaaS platforms into intelligent partners able to proactively solve customer programs. ### What are the key metrics for measuring success in SaaS digital marketing? This year, the focus has pivoted toward efficiency and durability, prioritizing NRR and the CAC payback period to ensure sustainable growth. Teams are tracking activation rates and product-qualified leads, which provide a more accurate picture of how digital marketing drives actual product adoption. These metrics align marketing spend with long-term profitability, rather than just top-of-funnel volume. --- ### Solving Attribution Gaps With S2S Tracking in Finance Lead Gen URL: https://www.taboola.com/marketing-hub/s2s-tracking-for-attribution-gaps/ Last Modified: 2026-04-12 14:50:44 As finance marketers become more tech-savvy and data-driven, they’re discovering some useful strategies to help with lead gen and growth. In this highly regulated market, accurate data is the difference between a scaled success and a costly failure. The right tracking tools make a big difference in aligning with user journeys to understand how finance users move between and among sites, and how to ultimately close deals with engaged audiences. ## Bridge the Attribution Gap for Third-Party Conversions Standard pixel tracking often falls short when financial services prospects move from a landing page to a third-party brokerage, or offline deposit site. Instead, server-to-server (S2S) tracking can offer a better way to eliminate the blind spots caused by cookie blockers or cross-domain hurdles. With S2S tracking, data moves directly from the advertiser’s server to the Realize platform. This is especially relevant for the microcap investment market. Finance advertising experts Brandon Jones, solutions engineer, and Anslynn Capps, advertising sales manager, both of Realize, explain here how to approach finance and microcap investment marketing to find and engage the right audiences on the open web. Their advice is based on a real-world use case: a finance advertiser navigating the complexities of tracking investment interest beyond the initial click. In the case of this advertiser, their ultimate success metric is stock purchases, which take place on external brokerage sites that don’t support pixel placement. See what the experts advise, and try these tips to improve your own rates of return for investment performance marketing and advertising campaigns. ### Utilize Click IDs for Post-Click Visibility Standard pixels lose sight of the user once they leave the advertiser’s domain. To solve this problem for finance advertisers, the experts recommend a handshake method using the Realize Click ID. This allows for offline conversion reporting that feeds directly back into the Realize performance marketing algorithm to optimize campaigns going forward. “The pixel is great for on-page engagement, but for finance players, the real action often happens on a brokerage site you don’t own,” points out Jones. “By passing the Realize Click ID through your tracking platform (like Voluum or Bemob), you can send that data back to Realize once the purchase is confirmed. It turns invisible offline actions into actionable optimization signals.” ## Combine Pixel and S2S Tracking for a Full-Funnel View The finance advertiser in this case was seeing high bot traffic and low-quality leads from their email signups. They chose a combination of pixel and S2S tracking for a more robust qualification method. ### Optimize Toward High-Intent Engagement Signals While waiting for the deep-funnel S2S data to populate, the experts suggested that this finance advertiser use the Realize pixel to track high-intent proxy behaviors, like 60-second time-on-page or investor presentation downloads, to train the algorithm early. “Don’t just track the lead, track the intent,” advises Capps. “In finance, we see a lot of noise with email signups. We recommend layering pixel-based engagement triggers — like a 60-second timer — with your S2S deposit data. This gives a performance marketing platform like Realize the breadbox of data it needs to find investors, not just clickers.” ## Address Content Restrictions Through Technical Precision Within the regulated finance vertical, a common challenge can be navigating the strict “unverified claims” rejection policies. This particular advertiser had to ensure that landing page disclaimers and tracking were perfectly aligned to avoid any lags in data capture. ### Align Landing Page Logic With Compliance The Realize experts found that the company’s ad rejections often resulted from unverified claims. This finance advertiser needed to pair technical tracking with clear disclaimers, ensuring that the content review team could verify the legitimacy of the service being tracked. “Finance ads are heavily scrutinized,” says Jones. “If your landing page is flagged for unverified claims, it pauses your data learning. We made sure this advertiser’s boilerplate disclaimers are visible to the review team while the S2S pings are firing in the background. Stability in your tracking setup leads to stability in your campaign approval.” ## Key Takeaways Financial services advertisers have to be strategic about how they’ll track and engage potential leads to then tailor messaging and offers to them. For open web advertising, implementing S2S tracking is more than a nice-to-have option: It’s now necessary to go beyond browser limitations, bringing trusted, high-fidelity data to the performance marketing platform or other performance engine. Then, marketers can optimize for return on ad spend (ROAS) and actual stock liquidity based on real numbers. Guesswork isn’t an option for modern finance marketers, especially in microcap investing. ## Frequently Asked Questions (FAQs) ### What is the primary benefit of S2S over pixel tracking? Generally, server-to-server (S2S) tracking is more reliable than pixel tracking, since it isn’t affected by cookie expiration, ad blockers, or browser settings. Specifically in terms of performance marketing campaigns on the open web, S2S tracking offers more protection against those browser-level disruptions, including ad blockers and poor connectivity. These issues can lead to conversion data loss, which then degrades the quality of campaign data. S2S tracking can create a more stable, accurate feedback loop, which helps marketers strategize better and improves AI learnings. Performance marketing platform AI can optimize campaigns using 100% of the conversion data instead of a fragment of it, leading to better targeting and results. ### Can I use both S2S and pixel tracking together? These two tracking methods can indeed be used together, and often are used in tandem to track different funnel stages. For performance campaigns on the open web, it may make more sense to use a hybrid setup to be able to capture immediate client engagement, as well as ensuring a more resilient, server-side record of final conversions. This approach also helps performance platform AI technology use pixel tracking data for real-time traffic signals, like page views, while the S2S tracking data is more reliable for accurate lead attribution or sales that browser scripts might miss. ### Is S2S tracking difficult to set up? This depends on your performance platform, but S2S tracking is typically a one-time structural setup task that requires either server-side knowledge or a third-party tracker. When considering setting up S2S tracking for a performance campaign on the open web, you’ll have to configure a postback URL between the server and ad platform. The long-term gains usually make this setup worthwhile — you’re creating a high-fidelity data loop that’s immune to browser-based disruptions. That leads to better optimization of performance campaigns, since that reliable data includes 100% of conversion signals. --- ### 2026 Content Marketing Statistics: Key Data to Shape Your Strategy URL: https://www.taboola.com/marketing-hub/content-marketing-statistics/ Last Modified: 2026-03-31 13:03:16 Content marketing forms the cornerstone of modern brand-building. As a B2B marketer, you create high-value, relevant experiences that nurture long-term relationships. Artificial intelligence (AI) has added another layer of complexity to the marketing landscape, though, and search behaviors have shifted as consumers rely more on AI summaries at the top of search engine results pages (SERPs). As a result, click-through rates (CTRs) have declined — and so have attention spans. You might find yourself having to justify your budget more often, or show the tangible, positive impact of your work to various stakeholders. In short, you’re facing some pretty stiff headwinds in 2026. Still, data remains king. It doesn’t matter if you’re developing a new roadmap, refining last year’s strategy, or weighing the return on investment (ROI) of new channels like short-form video or branded podcasts — grounding your decisions in benchmarks remains non-negotiable. To help you optimize your strategy for 2026, I’ve compiled a list of essential stats covering effectiveness, ROI, lead generation, and the nuances of the B2B and B2C markets. These figures highlight the ongoing transition from volume-based production to AI-enhanced personalization, quality over quantity, and the ongoing role of video. Use these insights to align your goals with strategies proven to drive growth, get results, generate leads, and make the folks who control your budget happy. ## Top Content Marketing Statistics 2026 Looking for a quick overview? These statistics highlight the direction of content marketing in 2026, and how other top marketers are leveraging it to grow brand awareness and generate leads: - 93% of marketers believe their budgets will stay the same or increase in 2026; 37% said their budgets are shrinking, by generally under 10%. - In 2026, 95% of B2B marketers are using AI-powered marketing applications; 80% use AI for content creation, and 75% for media production. - 45% cite AI-powered tools as among the top three investments, followed by events/experiential marketing (33%) and owned media (32%). - 97% of marketers have a content strategy for 2026, with 61% indicating it significantly or moderately improved results and ROI. - 89% of marketers use AI-powered tools for content creation, and 53% use them to create and edit images, videos, and other visuals. - LinkedIn is the top channel for publishing thought leadership (76%). - Short-form video remains king, generating the highest ROI in 2025 (104%). Most marketers said it was the most effective channel and plan to continue investing heavily in it throughout 2026. ## B2B Content Marketing Statistics A strategic blend of high-tech efficiency and high-touch connection defines the 2026 digital landscape. As companies navigate an increasingly saturated information market and try to cut through the noise, they’ve shifted their focus from volume to measurable impact and multi-channel integration. From the dominance of AI in the creative process — and metrics collection and analysis — to the resurgence of face-to-face engagement, these statistics highlight the benchmarks and priorities shaping today’s content marketing environment. - Content marketing is a part of 92% of B2B marketers’ marketing strategies, according to the Digital Marketing Institute. - The most effective distribution channels in 2025 were in-person events (52%), webinars (51%), email that didn’t include newsletters (42%), social media (42%), corporate website blogs (41%), and email newsletters (37%). - The most popular content types, according to Leadfeeder, were case studies/customer success stories (41%), videos (39%), blogs (37%), and podcasts (31%). - According to HubSpot, blog posts were among the top five highest-ROI content formats and among the top five content formats marketers will invest in for 2026, followed by short-form, live, and long-form video and user-generated content. - 94% of marketers will use AI in content creation in 2026. - Among social media platforms, the top ones marketers will use to support short-form video in 2026 are Instagram (48%), Facebook (43%), YouTube (42%), TikTok (32%), and Twitter/X (31%). - 12% of B2B marketers say they exceeded goals, 47% felt they met most goals, and 31% reported mixed results. ## B2C Content Marketing Statistics While B2B data often dominates industry reports, the B2C sector is seeing a demand for human-centric digital experiences. Today’s B2C success hinges on the depth of engagement and the speed of trust, so these marketers prioritize quality over quantity. The following trends showcase how B2C brands are acknowledging consumers’ prioritization of video and authenticity, and bridging the gap between automated scale and personal connection. - Video content is key to attracting audiences: 96% of consumers watch an explainer video to learn about a service or product, videos convince 85% of people to make a purchase, and 84% of consumers want more videos in 2026. - Companies embracing hyperpersonalization see 40% more value. - 43% of marketers name engagement rate as the top metric to rate. - 70% of B2C marketers have incorporated content marketing into their strategy, with 83% creating short articles or posts and 90% using social media to share content. - 83% of B2C marketers agree that quality trumps quantity — even if doing so reduces posting frequency. - B2C marketers said in-person events had the best results (48%), followed by short articles/posts (47%), videos (45%), and virtual events, webinars, and online courses (41%). - The top three paid channels included in B2C marketers’ content strategy include social media advertising/promoted posts (88%), search engine marketing (SEM) and pay-per-click (PPC) (73%), and sponsorships (55%). ## Content Marketing ROI Statistics Content marketing ROI data helps you strategize where (and how) to allocate your content marketing budget to generate the biggest impact. Here’s how some of the numbers are shaking out: - The top marketing channels with the biggest ROI, according to HubSpot, are websites/blogs/search engine optimization (SEO) (27%), paid social media content (26%), organic social media content (24%), email marketing (22%), brand awareness (19%), and content marketing (17%). - Email marketing generates $10 to $36 ROI for every dollar spent. - The marketing channels driving the highest ROI are email marketing, SEO, and content marketing (blogs and video). - 90% of marketers report a positive ROI from video content, with 95% agreeing that video is a critical component of marketing strategy, and 85% planning to maintain or increase their spend on this tool. - 90% of marketers share videos on YouTube, 86% on Facebook, 79% on Instagram and LinkedIn, 54% on Twitter/X, and 35% on TikTok. ## AI in Content Marketing Statistics For content marketers in 2026, AI has transitioned from a shiny new experiment to become an integral part of operations. The following information, drawn from major industry reports including HubSpot’s 2026 State of Marketing and the Content Marketing Institute, outlines the current impact and projected trajectory of AI in this industry. - Approximately 81% to 87% of marketers use AI-powered tools for content tasks. - 51% plan to increase their spend on AI-driven content. - The top five uses for AI in content production include generating and optimizing copy (89%), generating and editing visual assets (53%), SEO (41%), social media management (38%), and email marketing (36%). - Marketers estimate that human-written content costs 4.7 times as much as AI-generated content; many companies are experimenting with or adopting a hybrid AI-generated, human-edited model. ## Content Marketing Growth Statistics Content marketing has a direct impact on growth metrics: - 30% of content teams are investing in short-form video, 13% in live streaming video, and 12% in long-form video in 2026. - 58% of marketers say content marketing generates sales and revenue. - 72% of companies credit content marketing with boosting lead generation. - Nearly 30% of marketers name brand awareness campaigns as a top investment for delivering the highest ROI. - 33% of marketers measure revenue impact, followed by lead quality/conversion rates (27%), productivity metrics (20%), customer satisfaction scores (11%), and customer engagement (9%) to calculate ROI. - According to Forbes, personalization can boost sales by 40%, with 89% of marketers agreeing that personalization is essential. ## Content Marketing Lead Generation Statistics Marketing teams know that content marketing boosts lead generation in various ways, which is why almost 90% use it. - Compared to outbound marketing, content generates more than 3x as many leads and costs 62% less. - According to 81% of marketers, content marketing helps elevate brand awareness. Websites with regularly updated blogs have over 430% more indexed pages than static websites without one. - 51% of content marketers credit new tech with improving effectiveness, but note that human involvement contributes to 74% of that improvement. - 77% of marketers rate their leads as “high” or “very high” quality, according to HubSpot. - While B2B marketers try to balance quality and quantity, 68% of businesses struggle with lead generation. - 95% of B2B marketers use AI-powered apps, with 20% saying their implementations are exploratory, 48% developing, 24% established, 5% advanced, and 3% leading. ## Content Marketing Statistics by Channel Successful content generation does not require being everywhere at once: It’s much more effective to be in the right place with the right purpose, with content crafted appropriately for the audience(s) you’re targeting. Each channel has a different advantage, whether it’s the long-term compounding authority of SEO-rich blogs or more immediate, high-impact engagement of short-form videos and email. Abandon random acts of content and embrace a more deliberate roadmap. The insights below can help you audit your content mix, identify high-ROI opportunities, and align your distribution strategy with your target audience’s specific behaviors. ### Organic Search - Organic search results account for 94% of all clicks, but GenAI (and AI overviews) have led to 60% of searches ending without a click; 30% of marketers have reported decreased traffic since consumers have begun relying on and using AI tools more frequently. - 53% of bloggers struggle to attract search engine traffic, even though blogs remain a top content channel for SEO authority and lead generation for B2B and B2C audiences. - Even though fewer than 10% of marketers currently use voice search optimization, over 20% of the world’s 1.5 billion internet users aged 16+ use voice assistants to find information; Statista predicts that by the end of 2026, 157+ million people in the U.S. will use voice assistants. - Just over 20% of marketers write for search engines, versus nearly 80% of marketers who prioritize writing for humans. - Over 92% of marketers already use or plan to optimize SEO for AI-powered and traditional search engines. ### Video Content Marketing Statistics - 91% of companies use video as a marketing tool, a 3% increase from 2025. - 45% credited video as their top-performing content in 2025. - 82% of marketers say they’ve seen a solid ROI from video marketing, 93% credit videos with increasing brand awareness, and 82% say this content has helped increase web traffic. - 85% of marketers credit videos with helping to generate leads and 83% say it has directly increased sales. ### Podcasting - In 2025, 61% of B2B marketers (41% B2C) used audio platforms like podcasts to increase the accessibility and personalization of content delivery. - Branded podcasts can increase brand favorability by 14%, and 38% of listeners purchased products based on a podcast ad. ### Email marketing - 71% of B2B marketers use email newsletters as part of their content strategy. - 31% of B2B marketers say newsletters are an effective lead-nurturing tactic. - 72% of consumers prefer email marketing over other options. - 77% of B2B marketing teams say instructive and personalized email content performs best. - The ROI for email marketing averages $42 for every $1 spent. - By 2027, the email marketing industry may reach nearly $18 billion. ## Key Takeaways: How to Make the Most of Your Content Marketing Strategy in 2026 To stay ahead in a year defined by zero-click searches and AI-driven discovery, embrace content marketing and keep these pillars in mind: - Optimize for the zero-visit reality: With AI overviews and generative search engines (GEO) answering queries right on the search engine results page (SERP), traditional CTRs are feeling some pressure. Structure content for extraction, using clear answer blocks at the beginning of each section to help AI models identify your content as a source of truth. - Prioritize humans-in-the-loop: The cost of AI-generated content is almost 5x lower than that of human-written work; however, value follows the same curve. Use AI for the heavy lifting, like summarizing, drafting, and SEO tagging. Avoid AI schlock by incorporating personal anecdotes, original (fact-checked!) research, and a distinct brand voice AI simply can’t create (or imitate). Humans still need to own the vibe. - Champion short-form video: Video remains the key for ROI, with a large percentage of marketers reporting positive returns. A caveat: 2026 favors raw and relatable over highly polished. Record a long-form session (podcast or webinar) and use AI tools to spin it into 15+ short-form clips for social media. - Embrace owned media. As algorithms on social media platforms and search engines remain unpredictable, lean into your most stable assets — email lists and proprietary data. Double down on email marketing and interactive owned assets, like calculators, portals, or gated community groups. Consider moving your audience away from rented land (social media) and into environments where you control the data and the relationship. - Pivot from reach to relevance: Mass messaging is so last year: Audiences favor micro-communities and appreciate hyper-personalization. Ignore vanity metrics and look at engagement depth and lead quality. Build closeness and community, which fosters the speed of trust necessary for modern conversions. ## Frequently Asked Questions (FAQs) ### What is the success rate of content marketing? In 2026, the success rate of content marketing is high but increasingly stratified. The definition of success has shifted from vanity metrics to measurable business impact, with over 41% of marketing teams using sales to measure success, according to HubSpot. It generates 3x more leads than outbound marketing, and has strong conversion power when videos and email marketing are part of the toolbox. ### What is the average ROI for content marketing? ROI for content marketing is usually positive, but results vary by channel. Overall, according to SQ Magazine, the average 2025 ROI for content marketing was $7.65 for each $1 spent. Email marketing’s ROI averages $42 per $1 spent. In 2025, for every $1,000 spent on short-form video, direct sales attributed $8,900, and AI-enhanced podcasts saw a 650% increase in ROI. ### What is the average budget for content marketing? The U.S. Small Business Administration suggests that businesses with under $5 million in revenue should allot 7-8% of gross revenue to marketing. Mid-sized businesses should invest 6-9% of revenue, and startups or high-growth businesses should invest 20-50% of revenue. In a very competitive industry, you may want to allocate 10-15% or more. Once you have your dollar amount, Almcorp’s recommended budget breakdown for 2026 is: - Content marketing SEO/AEO: 25-30%. - Email marketing: 15-20%. - Paid search (PPC): 10-15%. - Paid social media: 10-15%. - Video marketing: 10-12%. - Marketing tech and AI tools: 8-10%. - Influencer and partnership marketing: 5-8%. - Testing and innovation: 5%. --- ### Best Performance Advertising Platforms for Campaign Set Up in 2026 URL: https://www.taboola.com/marketing-hub/best-platforms-for-campaign-set-up/ Last Modified: 2026-05-24 08:15:40 For performance marketers, the campaign platform you choose will likely be a tool you consult every day. With the advanced features and AI power behind many of today’s campaign platforms, these tools can run a wide range of performance marketing functions to avoid wasted budget, provide deeper data insights, and continually identify high-value audiences. To choose a campaign platform wisely, know what your specific goals are, what your budget is and how flexible it is, and have a sense of audience behavior. When you’re first researching platforms, consider targeting precision, tracking capabilities, and existing customer relationship management system (CRM) integration across different ad platforms. ## 9 Best Performance Platforms for Campaign Setup, Compared Platform Why It’s Essential Core Use Cases and Features Best for (Performance Advertisers) Pricing Model (Indicative) 1. Realize Automates campaign creation with outcome‑aligned setup logic. Key performance indicator (KPI)‑driven campaign templates, real‑time activation, performance‑linked targeting and budgeting. Performance advertisers seeking automated high‑return on investment (ROI) setups. Performance-based model; campaigns billed on CPC basis, or CPM for programmatic. 2. AdEspresso by Hootsuite Simplifies campaign creation across Meta and Google Ads. Guided setup, split‑testing creation, automated suggestions. Small-to-medium businesses (SMBs) and performance advertisers seeking fast setup. Subscription (from about $49 per month). 3. Adzooma All‑in‑one campaign management and setup. Multi‑platform campaign creation tools and automation. SMBs and performance advertisers needing integrated workflows. Free and paid plans. 4. Birch (Revealbot) Rule‑based setup automation and templates. Bulk setup rules, pre‑built automation logic, multi‑platform support. Agencies and automation‑first performance teams. Subscription (varies). 5. Google Ads Editor Free bulk campaign editing and offline campaign setup. Upload/edit campaign structures in bulk, and import/export files. Search advertisers and search engine marketing (SEM) teams with heavy Google Ads usage. Free (Google tool). 6. Optmyzr Advanced setup automation with bulk campaign creation. Bulk campaign templates, automation scripts, optimization workflows. Mid‑large pay per click (PPC) teams and agencies. Subscription (from about $200 per month). 7. Semrush PPC Toolkit Keyword/competitive analysis and setup guidance. Research tools to inform campaign setup, competitor insights. Performance advertisers focused on search strategy. Subscription. 8. WordStream Advisor Automated recommendations for campaign structuring. Guided setup, cross‑platform campaign creation, keyword suggestions. SMB and mid‑market performance advertisers. Tiered subscription. 9. Zapier (with Ad Platforms) Automated setup workflows between tools. Connect creative/data sources to ad setup flows. Teams automating setup tasks via workflows. Subscription. ### 1. Realize Why it’s essential: Realize is a performance-driven advertising platform designed to simplify and automate the campaign setup process for the open web, consolidating setup workflow into a single, intuitive interface. The platform is used to automate complex tasks such as tracking implementation and creative generation, effectively removing technical barriers and manual errors that often delay campaign launches. Showcased features:  - Abby AI automates the onboarding process and campaign launch using an Onboarding Wizard that applies Realize best practices for a faster, smarter setup. - Automates the creation of event- and URL-based tracking without developer support once the initial Pixel is active. - Social Importer quickly repurposes top-performing Facebook and Instagram creatives into Realize display ads to save time on asset production. Best for: Realize is optimized for D2C brands, as well as those operating in industries like financial services, where the customer journey involves high-consideration buying behavior, providing end-to-end support from creative generation to tracking validation. Pricing model: Performance-based model; campaigns billed on CPC basis, or CPM for programmatic. Pros:  - Automation tools drastically cut down the manual labor required to get campaigns live. - Features such as Codeless Conversions allow non-technical marketers to implement accurate tracking without writing code or using GTM. - Tools like Bulk Edit and the Tracking Test Tool provide clear visibility and validation to prevent costly setup mistakes at scale. Cons:  - Some advanced AI bidding and simulation tools require a learning phase with historical data before they reach peak accuracy. - Creatives brought in through the Social Importer are not editable once uploaded, which can limit fine-tuning during the initial setup. - Certain high-value setup and targeting tools, such as Search Keyword targeting or specific bid caps, are currently in limited beta or restricted to certain markets. ### 2. AdEspresso by Hootsuite Why it’s essential: This platform is a streamlined central option for managing multi-channel campaigns across Facebook, Instagram, and Google Ads, bringing simplicity to A/B testing and automated optimization. Showcased features:  - Very fast A/B testing with ad variation creation. - Custom rules that reallocate budget to highest performers. Best for: SMBs and solopreneurs. Pricing model: Subscription-based, from about $49 per month to $259 for enterprise features. Pros:  - Ease of use. - Library of webinars and guides for performance marketers. - Centralized dashboard and fast experimentation. Cons:  - Limited platform support beyond Meta and Google. - Some advanced features of native platforms aren’t available. ### 3. Adzooma Why it’s essential: This platform acts as an efficiency layer without high cost, offering a dashboard that shows immediate opportunities to cut wasted spend. Adzooma also provides a PPC health score across different networks. Showcased features:  - Automation rule engine can manage accounts 24/7. - Suggestion engine constantly analyzes accounts and suggests remediations. - Includes SEO and web performance tools. Best for: SMBs, freelancers, and small teams managing a mix of Google, Microsoft, and Meta ads. Pricing model: Robust free tier, with paid plans starting around $70 per month for advanced automation and more data refreshes. Pros:  - Very easy to use the free version. - Fast to set up. - Brings a consistent view across Google and Microsoft Ads. Cons:  - No advanced scripting controls. - Doesn’t offer much strategic support. ### 4. Birch (Revealbot) Why it’s essential: Birch, formerly Revealbot, can help performance marketers manage budgets across platforms with rule-building engine capabilities, and can execute complex strategies across several platforms at once. Showcased features:  - Visual builder for advanced rule construction. - Server-side tracking for first-party data. - Bulk-creation tool for ad variations. - Easy external data integration. Best for: High-growth e-commerce brands and teams managing multi-channel campaigns that need to scale quickly. Pricing model: Usage-based, tied to monthly ad spend, with plans starting around $49 per month for spend up to $10,000, and a pro tier that unlocks the automation engine and custom integrations. Pros:  - Extremely flexible in automating almost any manual action. - Consistent in applying logic across platforms. - Performs 15-minute checks to help save wasted budget. Cons: - Pricing based on spend and can grow quickly. - Platform requires technical setup. - Doesn’t create any visual content itself. ### 5. Google Ads Editor Why it’s essential: Google Ads Editor is a fast desktop app that offers offline management, cutting out the latency of web browsers and providing bulk processing capabilities. Showcased features:  - Bulk editing tools work across thousands of ads at once, and offline management works quickly. - Can manage multiple Google Ads accounts from a single dashboard, very useful for agency-side performance marketers. Best for: Power users, SEM specialists, and agencies managing complex accounts. Pricing model: Free. Pros:  - Very fast, with time savings when duplicating campaign structures or updating messaging. - Safety net lets you review changes before they affect spend. - Massive scale is possible. Cons:  - Syncing latency can be tricky with multiple users on the same account. - Steep learning curve and lots of manual work. ### 6. Optmyzr Why it’s essential: This automation layering tool brings granular control to performance marketers to audit AI-driven campaigns and create custom rules without writing code. Showcased features:  - PPC investigator diagnostic tool maps why a campaign went up or down. - Rule Engine builder lets users automate complex tasks. - Campaign automator can build and manage ads based on live inventory or product feed. Best for: Intermediate to advanced PPC managers, teams managing high-spend accounts, and e-commerce brands with a wide product range. Pricing model: Tiered subscription based on ad spend, starting around $200 a month and going up as managed spend increases, with more features available. Pros:  - Opens up hidden Google Ads data. - Brings efficiency to teams. - Includes a library of automation scripts. Cons:  - Cost can escalate quickly. - Technical setup is required and learning curve can be steep at first. ### 7. Semrush PPC Toolkit Why it’s essential: Semrush helps performance advertisers map out the competitive landscape, including keywords that rivals are bidding on, how much they’re spending, and what their ad copy is. Their massive keyword database helps find gaps and opportunities. Showcased features:  - Keyword Magic Tool taps into a 25 billion-keyword database for terms that can lower CPC while maintaining high intent. - Analyzes competitor strategies over time. - PPC Keyword Tool organizes keywords into ad groups and optimizes lists. Best for: Performance marketers, e-commerce managers, and teams prioritizing competitive intelligence and keyword discovery as the foundation of their campaign setup. Pricing model: Subscription-based, with three tiers ranging from about $140 per month to almost $500 per month. Most PPC research tools are included in the two lower tiers. Pros:  - Competitor keyword visibility offers rich data. - Performs end-to-end workflows, from research to ad copy creation and tracking. Cons:  - Data estimations aren’t completely precise. - Can supply too much data depth for smaller campaigns. ### 8. WordStream Advisor Why it’s essential: WordStream provides a simplified approach for non-experts and smaller businesses vs. working directly in Google and Meta interfaces. It’s still useful for those without a lot of time and resources for PPC software and expertise. Showcased features:  - Proprietary “20-Minute Work Week” dashboard suggests changes to make. - Performance graders audit your account to find wasted spend. - Smart ads tools help repurpose existing assets. Best for: Small business owners, marketing generalists, and beginners spending between $1,000 and $10,000 per month. Pricing model: Subscriptions based on ad spend, starting around $300 per month. Pros: - Simple, easy-to-use interface. - Cross-platform sync. - Context on why the platform suggests certain optimizations. Cons:  - Costs can scale quickly. - Less granular control than other platforms or for enterprise-level accounts. ### 9. Zapier (with Ad Platforms) Why it’s essential: Zapier can serve as a connector for performance advertising by automating tasks like moving lead data, syncing audiences, and tracking offline conversions across the CRM, lead forms, and ad platforms. Showcased features:  - Real-time lead sync from Facebook or Google into the CRM or Slack for fast reaction times. - Offline conversion tracking opens up value-based bidding for high-value customers. - Dynamic audience syncing saves a lot of update time. Best for: Growth marketers, revenue teams, and performance teams that need to integrate fragmented tools. Pricing model: Free tier for single-step automations; paid plans start around $20 per month and go up to $49 or more per month for more features and speed. Pros:  - Highly connected to more than 8,000 apps. - Cuts lead response times from hours to seconds. - Easy to use for non-technical marketers. Cons:  - Task usage can add up quickly. - API changes in connected apps can cause data gaps. - Some latency is possible, depending on the plan. ## Getting Started with Campaign Setup in Performance Advertising Tools Getting things right from the start is always easier than troubleshooting later, especially when you’ve got campaigns up and running and there’s real budget on the line. Here are some tips to consider. ### Common Mistakes to Avoid When Setting Up Advertising Campaigns When you’re getting started setting up a performance advertising campaign, try to avoid these common mistakes: - Failing to set realistic or clear goals. - Improper audience targeting. - Using low-quality creative. - Not using negative keywords. - Poorly managed budget. - Inaccurate assumptions about CPA numbers. - Not closely or continually monitoring ad performance. ### Establishing a Good Starting Budget for a Performance Ad Campaign Performance marketers are likely already keeping a careful eye on all the data required to set up an ad campaign, including target spend, ROI, and ROAS. The initial budget is ideally big enough to collect useful data for algorithms to learn from — that’s usually $500 to $3,000 per month for small businesses. For faster learning, consider spending 15x-20x the desired daily CPC. Other tips to consider: - Work backward: Decide how many leads per month you need, at least for the first few months, then multiply that by your cost per lead and set that resulting number as a monthly budget. - Use percentage of revenue: Another common benchmark is to allocate 5% to 10% of revenue for marketing, though fast-growing startups may increase that to 15% to 20%. - Include a testing allowance: Earmark 15% to 20% of the total marketing budget for testing and piloting new tactics to figure out what works as quickly as possible. ### Performance Advertising Campaign Creation Checklist Setting up a first campaign involves a lot of details, even when you have help from a strong platform. Here’s what to remember before, during, and after launch. #### Before Launch - Define objectives and KPIs, with timelines and specific numbers. - Define target audience: demographics, interests, location, online behaviors, and any other available data points. - Set the budget: daily and monthly and divided up by platform. - Decide on the offer that you think will be a good incentive for the audience. - Research competitor strategies, keywords, and creative. #### Creative and Content - Write the most succinct, clear, action-oriented messaging you can for the user’s problems and needs. - Create high-quality visual assets, including images and videos. - Develop variations for testing, including headlines, images, and calls to action (CTAs). - Build a landing page tied to the ad you’re running, making sure it’s fast and mobile-friendly. #### Infrastructure Setup and Launch - Implement pixels for the right platform, like Meta or Google, and set up conversion tracking before the campaign launch. - Organize the campaign into ad groups with relevant keywords and targeting. - Decide which bidding strategy to use, like target CPA or maximum conversions. - Perform a compliance review ahead of launch to make sure your ads meet compliance requirements. #### Learn and Optimize - As data starts coming in from your first campaign, monitor the relevant KPIs, like ROI, CPA, click-through rate (CTR), and whatever else is relevant to your business goals. - Plan and carry out A/B testing regularly with ad creatives, audience targeting, and landing pages. - Adjust budget to put more money toward high-performing campaigns, focusing on gradual scale. - Update ad content to prevent fatigue. ## Key Takeaways Choosing the right platform for campaign setup is one of the most important tasks a performance advertiser does. With a wealth of options on the market, performance marketers should balance the capabilities that the business needs against the allocated budget. Some key features to explore are automation, cross-platform dashboards, data analytics, ad creation vs. ad management, and tracking. ## Frequently Asked Questions (FAQs) ### What problems do campaign setup tools solve for performance advertisers? Campaign setup tools can help performance advertisers work a lot faster and more accurately. Campaign setup platforms automate complex or repetitive tasks and can more easily manage multiple channels, audience targeting and retargeting, budget pacing, and performance analytics and reporting. It’s especially useful to have a centralized tool that manages ads across search, display, and social from one place. Performance marketing teams can save hours of time and resources using a campaign setup platform, freeing them up for strategic work. ### Are these tools only for large advertisers? There’s a wide range of campaign performance tools for advertisers of all sizes. Teams with larger budgets may have more options for tools, but even small advertisers can get started with free tiers or less expensive choices that bring a lot of useful functionality. With the addition of AI to most of these tools, smaller companies can use these powerful features to create high-quality ads, manage campaigns, and access data analytics. ### Can AI tools fully replace manual setup expertise? AI tools can be hugely helpful to performance advertising teams for automating routine tasks, searching through large datasets, and other productivity-enhancing duties. But, these tools can’t replace the manual setup expertise that a performance advertiser brings. Humans are needed for critical thinking, contextual understanding, learned experience, and assessments of complicated scenarios. Plus, managed accounts can bring the best of both worlds — powerful AI capabilities paired with advertising experts to help guide and refine strategy. --- ### Vertical Video Ads to Overcome Diminishing Returns for D2C Campaigns URL: https://www.taboola.com/marketing-hub/vertical-video-ads-d2c/ Last Modified: 2026-03-23 11:09:17 In the high-stakes world of consumer electronics, standing out is no longer a matter of having the best spec sheet — it’s more about mastering the format and the environment. As traditional social media platforms become increasingly crowded and ROAS (return on ad spend) is unreliable, smart and savvy brands are migrating toward the open web to capture high-intent audiences. Vertical video ads have emerged as the premier vehicle for capturing mobile-first customers. By offering an immersive, full-screen experience that mirrors modern mobile habits, these ads do more than just show a product — they actually grab the user’s attention. In this deep dive, we’re going to look at a real-world example, with expert advice and insight from Realize advertising sales manager Ari Del Rosario. We’re looking at the strategic shift that was required to move a disruptive U.S. electronics brand from social-only advertising to a sophisticated, full-funnel presence on the Realize platform. ## Advertising Problems Faced By High-End D2C Brands ### Ad Saturation and Blindness on Social Media On platforms like Meta or TikTok, the moment a user interacts with a projector or smart-home ad, the algorithm immediately floods their feed with direct competitors. You’ve probably experienced it yourself, and it’s better known as “category clutter.” Ad fatigue occurs when users subconsciously ignore areas of a screen where ads usually appear. By placing vertical ads on the open web, brands can appear within premium editorial content like a tech review or news article, where they’re the sole focus of the user’s attention. ### Diminishing ROAS on Traditional Channels The electronics brand in question noticed that their social media returns had plummeted from a healthy 4-5x to a stagnant 2x. This diminishing return happened due to the increasing cost of reaching the same audience as more competitors began to bid for the same space. Diversifying into the open web provided a relief valve, tapping into “blue ocean” inventory that hadn’t been oversaturated. ### The Impulse Buy Barrier for High-Ticket Items High-ticket direct-to-consumer (D2C) brands face a unique set of hurdles. Unlike a $20 impulse buy, premium electronics require a level of trust and mental real estate that a cluttered social feed rarely provides. Selling a high-priced item through a static mobile ad is a steep climb; using vertical video allows for a mobile awareness phase, using the full screen to tell a story and overcoming the skepticism that often accompanies high-priced tech by showing the product in action, rather than just a polished photo. ## Leverage Vertical Motion to Disrupt the Social Wall As noted above, the greatest challenge for hardware brands is the “social wall,” i.e., the reflexive scrolling that renders most ads essentially invisible. Realize allows you to take the very same vertical assets that were built for social stories and place them in high-authority environments. ### Capitalizing on the Mobile Hoarding Habit Modern consumers spend their lives with a computer in their pocket. Because vertical motion ads occupy the entire mobile screen, it creates a stop-and-stare effect. Technically, these ads are referred to as interstitials or full-screen overlays, and they achieve significantly higher view-through rates (VTR) because they eliminate all those peripheral distractions. “If you’re on social media and a projector ad comes up, the next seven ads you see will be different projectors — it becomes a sea of noise,” says Del Rosario. “We recommend using vertical video on the open web because it places your brand in a premium, non-competitive editorial context. You’re catching the user while they’re consuming news or niche tech content, not just mindlessly scrolling a feed.” ## The “Mobile Awareness, Desktop Conversion” Funnel For premium electronics, the path to purchase is rarely a straight line. High-ticket items require a multi-device approach: A beginner might assume that if a user doesn’t buy on their phone, the ad failed, but what often happens is that consumers discover products on mobile, then turn to the security and larger screen of a desktop to complete transactions in the $250+ range. ### Synchronizing Multi-Device Touchpoints With this is mind, our expert recommends a bifurcated strategy: - Top of funnel (mobile): Use vertical video to build brand equity and warm up the audience. - Bottom of funnel (desktop): Use aggressive retargeting — ads that follow the user to their computer — to close the sale. “We see the best results for electronics when we treat mobile and desktop differently,” confirms Del Rosario. “People carry their phones everywhere, making it the perfect place for that initial vertical video discovery. But, when it’s time to spend $250, they often move to a desktop. Our strategy is to build that interest on mobile, then use Realize’s tracking to meet them on their computer when they’re ready to buy.” ## Transitioning From Impulse to Research-Based Creative Premium tech requires technical validation and social proof. Realize’s vertical format supports advertorial style storytelling — a blend of advertising and editorial that builds the trust necessary for high-value purchases. ### Moving Beyond the Video Sales Letter (VSL) Hard-sell videos can certainly work for lower-cost items, but high-end hardware needs to highlight defensibility. This means showing why your product is better than a cheap knock-off. Highlighting local support, proprietary technology, or software licensing helps to build a moat around your brand. “For a high-end projector, you can’t just rely on an emotional impulse trigger,” says Del Rosario. “You need to highlight defensibility, like having a Google TV license or a U.S.-based tech team. We use vertical ads to lead users into a deeper story: It’s about shifting from a ‘buy now’ mindset to a ‘learn why this is better’ mindset, which ultimately leads to higher-quality customers.” ## Key Takeaways Vertical ads on the Realize platform allow consumer electronics brands to escape the walls and boundaries of social media. That’s the magic of combining the immersive nature of vertical video with the high-trust environment of the open web: when that happens, brands can reach consumers at every stage of the journey. ## Frequently Asked Questions (FAQs) ### Why should I use vertical ads instead of standard display banners? The biggest benefit is that vertical ads dominate a user’s mobile screen and take up way more visual real estate, which can lead to higher engagement than traditional banners. By taking full advantage of the natural/vertical scrolling habits of mobile users, vertical ads deliver the full-screen, immersive experience — one that better ensures your brand message actually sticks. While horizontal banners can often feel like annoying background noise, vertical video does a much better job at capturing the user’s undivided attention, especially on platforms like Apple News or other top-tier publishers. This is what provides the creative breathing room needed for more sophisticated storytelling and a clear, compelling call-to-action. It ultimately all adds up to much stronger conversion rates. ### How do I know if my vertical video is actually driving sales? The standard way to measure success is by using tracking pixels, which follow the user’s digital footprint from the initial click all the way to the final checkout. To really understand the ROI of your vertical content, though, you need to look at the full picture through conversion-based attribution. This means moving beyond basic engagement stats like view counts and focusing on view-through conversions — sales triggered by users who were influenced by your video, but chose to finish their purchase a little later in their journey. ### Is the open web effective for high-priced electronics? The open web is a powerhouse for high-ticket items because it places your brand alongside editorial content that already carries a high level of authority. When a consumer is on a premium tech or lifestyle site, they’re in a discovery and research mindset, which is the perfect time to build technical trust. Strategically, this allows you to reach high-intent audiences while they’re still weighing their options, often at a much more favorable cost than you would find in the hyper-competitive (and expensive) search auctions. --- ### Can AI Predict Ad Creative Performance? URL: https://www.taboola.com/marketing-hub/ai-predictive-capabilities-ad-creative-performance/ Last Modified: 2026-03-23 10:39:51 Every performance marketer has felt the sting of the budget burn — when you spend a week crafting what you think is a masterpiece, throw a few thousand dollars at an A/B test, and then watch as the data reveals your audience hated it. For years, marketers have treated the gap between art and science with a sort of PPC intuition — a fancy way of saying we were guessing based on experience and the overall vibe. But, in 2026, there’s a new wave of predictive capabilities in advertising that claims to have cracked the emotional code. The big question now: Can AI really tell you if your ad will be a hit before you spend a dime? ## The Shift From Art to Science in Creative Testing For decades, creative testing was pretty much the Wild West. Marketers and advertisers operated on a spray and pray mentality, launching 10 versions of an ad and hoping the algorithm caught a breeze. Today (thankfully) we’re moving away from that part art, part science paradigm, and instead, toward a world of data-backed forecasting. Predictive creative performance isn’t just about showing what happened in the past; it’s about simulating the future. By moving the goalposts from reactive to proactive, advertisers are essentially getting a weather report for their campaign before they set sail. ## Have AI Products Really Cracked the Emotional Code? It sounds a bit sci-fi, as well as a bit Black Mirror, but affective computing is very real. Modern AI doesn’t just see pixels; it understands psychological triggers. Through Natural Language Processing (NLP) and Computer Vision, these tools analyze visual hierarchy, focal points, and sentiment patterns. They know which shades of blue build trust and which headlines trigger urgency. By mapping these elements to specific emotional responses, AI can predict how a human might feel when they see your ad, allowing you to tweak those small but important details before it hits the open web. ## How Predictive Creative Modeling Works As with anything, the magic behind the curtain is actually just a massive amount of work, data, and research. Predictive models are trained on billions of historical ad impressions, correlating every tiny detail — face density, text-to-image ratio, color saturation — with actual conversion outcomes. It’s pattern recognition on a massive, global scale. When you upload an asset, the AI compares it to this library of winners and losers to provide an AI creative scoring report. It’s essentially giving your ad a grade based on how its “ancestors” performed in similar conditions. ## Predicting ROI: Forecasting Performance Before You Launch The real value here is the impact on your budget. Identifying duds during the pre-launch ad testing phase can save an advertiser up to 40% of their testing budget. Some industry benchmarks even suggest that ad performance forecasting has reached a 90%+ accuracy rate. That’s a huge boost for anyone on a tight budget. By cutting out the bottom-tier performers before they go live, you’re focusing your spend on the assets that have the highest probability of delivering a strong creative ROI optimization. ## Leading Tools in the Predictive Advertising Space If you’re looking to get started, there’s a growing list of specialists. ### AdCreative.ai Great for high-speed scoring. ### Madgicx Offers deep creative intelligence for social platforms. ### Alison.ai Provides detective-level analysis of your competitors’ visual elements. ### Pencil Focuses on the actual creative generation. Each of these tools aims to take the guesswork out of the creative process, though they often require a significant amount of your own historical data to be truly effective. ### GenAI Ad Maker From Realize This is where the power of network-level intelligence comes into play. The GenAI Ad Maker from Realize doesn’t just look at your past ads — it looks at the entire first-party publisher network it leverages. Because it’s plugged into real-time trending data from across the open web, it can refine your creative assets based on what people are actually clicking on right now. It’s a unique framework that bridges the gap between your brand’s voice and the internet’s current mood, making it a powerhouse for PPC creative automation. ## The Real-World Verdict: Do These Products Actually Work? There’s plenty of healthy skepticism in the PPC community, and rightfully so, since no machine can account for every cultural nuance. However, success stories are piling up where predictive tools have outperformed manual creative selection by 20% - 30%, as well as studies, which show that by incorporating long-standing Creative Shop best practices into its algorithms, the GenAI Ad Maker produces assets that are statistically more likely to match or exceed human-made benchmarks. So, what’s the consensus? Should you use it? The truth is, they aren’t magic wands, but they are incredibly effective filters. Bottom line, though, is you still have to put the work into it. ## The Strategic Framework for Using Predictive AI Start small by using these tools for headline optimization. Once you see a lift, move toward full multimodal scoring for images and video. The best workflow is a hybrid one: Let the AI handle the heavy lifting of sorting through the data, and let your human team focus on the big-picture strategy and brand storytelling. ## Limitations: What AI Still Can’t Predict AI is great at patterns, but it struggles with the new. If you’re trying to start a brand-new cultural trend or using a very specific, quirky brand voice, the AI might flag it as a risk because it doesn’t look like anything in its database. Human oversight is still mandatory to ensure brand safety and to catch the kind of long-tail creative that defies historical data. ## The Future of PPC: From Reactive to Predictive Looking ahead, we’re moving toward real-time emotion recognition and cultural readiness. Imagine an ad that adjusts its color palette based on the time of day or the local weather to better match the user’s mood. The goal for 2026 and beyond is an ecosystem where advertising is no longer an interruption, but a relevant, helpful suggestion based on predictive accuracy. ## Key Takeaways Predictive AI is no longer a futuristic nice-to-have, but a core optimization tool for modern performance marketers. By replacing creative gut feelings with data-backed forecasting, brands are saving significant portions of their previously wasted testing budgets. This leveraging of massive network intelligence and emotional AI allows advertisers to finally automate the complex science behind high-performing creative. ## Frequently Asked Questions (FAQs) ### Can AI really understand what makes a human feel an emotion? Obviously AI has no understanding of the human emotional experience, but it can grasp the various components. Through the study of affective computing, AI is able to analyze facial expressions, color psychology, and sentiment patterns to gauge a likely human reaction. On the open web, tools like Realize take this a step further by training on current network trends. This ensures that the emotional triggers being used aren’t just theoretically effective, but are contextually relevant to what users are actually engaging with in real time across premium publisher sites. ### Will predictive AI eliminate the need for A/B testing? It probably won’t kill off A/B testing entirely, but it definitely slims the funnel. Think of it as a pre-qualifier that identifies the most likely winners before they ever hit the auction. This significantly reduces testing costs — often by up to 40% — because your starting baseline for a live test is much higher than if you were just relying on human intuition alone. It lets you test the best of the best, rather than everything. ### Are these tools worth the investment for small budgets? Yes, as long as a 20% bump in ROAS covers the cost of the subscription. For small-to-medium advertisers, the savings from cutting out wasted spend on underperforming ads usually pays for the software itself. Furthermore, integrated platforms like Realize make these high-level predictive capabilities accessible without the need for an in-house team of data analysts, leveling the playing field for everyone. --- ### Target CPA: Do's & Don'ts for Your Performance Campaigns URL: https://www.taboola.com/marketing-hub/target-cpa-best-practices/ Last Modified: 2026-03-23 10:50:55 Target cost per acquisition (tCPA) can be one of the most powerful levers in performance marketing. But, it can also become a campaign killer if it’s turned on too soon, or configured without enough data. Across thousands of advertiser conversations, tCPA demonstrates that it works best when it’s treated as a refinement tool, not a launch strategy. Whether you’re running lead generation or direct-to-consumer (DTC) campaigns, the right approach can help you avoid the most common mistakes, protect delivery, and use tCPA the way it was designed — to improve efficiency without stalling growth. ## The “Wait” Rule: Data Before Bidding The single biggest reason tCPA campaigns fail is timing. When advertisers enable tCPA before the algorithm has enough signal data, they effectively blindfold the system, then demand efficiency. tCPA needs historical conversion patterns to understand who converts, where those conversions come from, and what price points are realistic. Without that foundation, the platform can’t bid confidently in the auction, so it may slow spend dramatically. This problem is especially pronounced in lead generation and DTC, where conversion signals are more nuanced and often delayed. ### DO: Use Maximize Conversions for the First 14–30 Days Every new campaign should begin with Maximize Conversions. This bidding strategy gives the algorithm freedom to explore inventory, test audiences, and identify real converters without artificial constraints. During this phase, the goal is learning, not efficiency. Maximize Conversions allows the system to: - Discover which placements consistently drive conversions. - Identify user behaviors that signal intent. - Build a stable conversion baseline for future optimization. “Launching a campaign with tCPA handcuffs the algorithm,” notes Brandon Jones, Realize solutions engineer. “It prevents the system from learning who to target, often causing a campaign to spend only half of its intended budget. Turn it off immediately for new launches.” Don’t think of the learning phase as wasted spend. Instead, consider it an investment in future efficiency. ### DON’T: Switch Before Reaching the 50-Conversion Floor Generally speaking, a campaign isn’t mature enough for tCPA until it’s reached sufficient conversion volume within a rolling 30-day window. Without enough historical data, the algorithm lacks the signal density required to confidently predict conversion probability at a fixed bid target. As a baseline: - Standard verticals should reach at least 50 conversions in 30 days. - Complex or highly regulated verticals, such as healthcare, generally require 100+ conversions in 30 days. “We typically look for campaigns to hit triple-digit conversions over a 30-day period before applying tCPA,” says Realize senior account manager Stephen Hollinshead. “Doing it sooner acts as a direct inhibitor to the learning process.” The key isn’t just volume, but recency. Rolling 30-day data ensures the algorithm is optimizing against current audience behavior, rather than outdated performance patterns. ## The “Step-Down” Method for Price Setting Even once the timing is right, tCPA can still falter if the goal is disconnected from reality. Efficiency targets must reflect what the auction is currently willing to support. ### DO: Anchor Your Initial Target to Actual Performance When transitioning from Maximize Conversions, look at recent performance — typically the last seven days — and set your initial tCPA at or slightly above that average CPA. If the campaign has been converting at $50, anchoring at $50 or $55 keeps you competitive while the system adjusts. “Set your tCPA where the actual average is,” advises Realize digital media manager Torogiá Stanton. “If you set it too low compared to actual performance, you minimize your reach, which actually causes the CPA to rise because the algorithm loses its ability to find opportunities.” In other words, an aggressive target doesn’t force efficiency: Instead, it limits auction participation. Starting at actual performance levels protects volume while creating room for gradual optimization. ### DON’T: Lower the Target by More than 20% at Once Once tCPA is active, restraint matters. Abrupt reductions force the algorithm to recalibrate under drastically tighter constraints, often resulting in volatile delivery or underspending. That’s why Realize account manager Renata Nugmanov recommends a phased approach. “Don’t jump to your goal price immediately,” says Nugmanov. “Start at the actual performance level and work it down by 10% increments every few days, to maintain a healthy auction presence.” This approach is simple: - Reduce the target by 10-20%. - Wait 2-3 days between adjustments. - Only make changes when the campaign is consistently spending its full daily budget. If spend drops below expectations after an adjustment, that’s a signal to stabilize before tightening further. ## Scaling and Market Strategy It’s important to separate efficiency management from your expansion strategy. tCPA may control cost, but it also introduces constraints. ### DO: Use the “Ladder Strategy” for Testing In competitive markets, a single tCPA target can leave opportunity on the table. Instead of forcing one campaign to balance efficiency and scale, some advertisers create multiple identical campaigns with different CPA goals to determine where performance and volume intersect. “Building a ‘ladder’ infrastructure of campaigns with different target increments — like $7 or $9 — allows you to see which one captures the best inventory without overspending,” explains Nugmanov. This method creates controlled experimentation within the auction. One tier may win premium placements, while another captures efficient mid-funnel conversions. Together, they reveal where the market is responding most favorably. ### DON’T: Expect tCPA to Scale as Fast as Maximize Conversions tCPA sets a ceiling. That’s its strength, as well as its limitation. When rapid scale is the priority, strict cost enforcement often slows expansion. For that reason, many performance marketers use Maximize Conversions during aggressive testing or seasonal pushes, then introduce tCPA once consistent, evergreen performance is established. “Target CPA is not ideal for scaling because it acts as a limitation,” says Stanton. “Until we find a very balanced, evergreen scale, I suggest using Maximize Conversions for the majority of the spend.” Efficiency can be protected later. The goal is to first capture scale. ## Budgetary Rules of Thumb Even well-configured tCPA campaigns will struggle if the daily budget restricts the algorithm’s flexibility. ### DO: Maintain a 10x Budget-to-CPA Ratio Budget determines how freely the system can compete across price tiers within the auction. If the allocation is too tight relative to the CPA target, the campaign effectively shuts itself out of competitive inventory. “The rule of thumb for tCPA is to ensure your daily budgets are 10x your daily CPA goal,” says Realize advertising account manager JeQuan Norris. “You need a strong budget to give the algorithm enough ‘at-bats’ in the auction to find cheap wins at your specified budget, otherwise it can cap out early and won’t remain competitive throughout the day.” Practically speaking, a $45 tCPA requires at least a $450 daily budget. That gives the algorithm room to test bids, absorb auction variability, and maintain consistent conversion velocity. ## Key Takeaways Target CPA works best as a refinement tool rather than a launch strategy. Start with Maximize Conversions to build reliable signal data, wait for statistically significant conversion volume, and set initial targets based on actual performance instead of goals. Lower targets gradually, ensure budgets are sufficient to support learning, and use Maximize Conversions for scale while reserving tCPA for cost control. When applied with the right timing and structure, tCPA improves efficiency without limiting growth. However, if it’s applied too early or too aggressively, it restricts delivery. ## Frequently Asked Questions (FAQs) ### What is the recommended conversion threshold before implementing target CPA? Most platforms require a solid base of historical conversion data before tCPA can work effectively. A common benchmark is at least 50 conversions within a rolling 30-day period, which signals that a campaign has enough activity for predictive optimization. In more complex or regulated verticals, that number often needs to be closer to 100 conversions to account for variability. ### What specific optimization strategies are used to balance lead quality with campaign scale? Balancing quality and scale requires structure, not tight restrictions. Campaigns typically launch with broader targeting to allow the algorithm to identify high-intent patterns. Creative and messaging play a major filtering role, using advertorial content, quizzes, or educational landing pages to prequalify users before they complete a lead form. On the optimization side, segmenting by placement, device, or geography allows top-performing segments to scale independently, while audience modeling and retargeting strategies refine intent without sacrificing reach. Many advertisers first scale under a volume-focused bidding strategy and later introduce tCPA to stabilize lead costs once sufficient conversion data exists. ### How does daily budget relate to the target CPA goal? Daily budget plays a central role in whether tCPA succeeds or fails. A common rule of thumb is to set the budget at least 10-15x higher than the target CPA. That range allows the algorithm to compete against price tiers in the auction, gather consistent feedback, and maintain healthy conversion velocity. When budgets are too close to the CPA goal, campaigns frequently stall because the system can’t test enough inventory to optimize effectively. Once performance stabilizes at or below the target, budgets should increase by around 20-30% every few days to avoid disrupting learning. --- ### Best Performance Platforms for Campaign Reporting in 2026 URL: https://www.taboola.com/marketing-hub/best-performance-platforms-for-campaign-reporting/ Last Modified: 2026-06-30 08:38:05 Reporting on campaign performance is one of the most important jobs a marketer does. It cuts out guesswork or tired tactics that might not be working anymore, and the data captured in campaign reporting can show actionable insights on who your audience is, what they’re looking for, how they’re connecting (or not) with your brand, as well as lots of other vital details. With so many platform options and more data available than ever, you have a wealth of options to choose the campaign reporting platform that works for you and your business’ goals. Here’s a quick view of what’s available for performance marketers, with more details below. ## 11 Best Performance Platforms for Campaign Reporting, Compared Platform Why It’s Essential Core Use Cases and Features Best for Pricing Model  1. Realize Performance‑driven reporting tied to outcomes. Integrated performance dashboards, key performance indicator (KPI) tracking tied to cost per click (CPC)/cost per acquisition (CPA) outcomes, real‑time measurement. Performance advertisers focused on outcome‑aligned reporting. Performance-based model; campaigns billed on CPC basis, or CPM for programmatic. 2. Cometly Pay-per-click (PPC)‑oriented performance reporting with insights. Automated dashboards, attribution, trend insights. PPC specialists and performance teams. Subscription (tiered). 3. Agency Analytics Automated client‑ready performance reporting, white labeling. 50+ integrations, custom dashboards, scheduled exports. Agencies delivering branded client reports. Subscription (tiered by features). 4. Databox Real‑time performance dashboards with alerts. Mobile and web dashboards, KPI tracking, customizable visuals. Teams and executives needing quick performance summaries. Subscription/ tiered plans. 5. Google Analytics (GA4) Central analytics hub linking web activity with ad performance. Traffic and conversion tracking, attribution models, custom reporting. Advertisers needing unified web and ad performance metrics. Free option; paid tier (GA360) 6. Klipfolio Custom KPI dashboards from diverse ad sources. Real‑time updates, visual dashboards. Data teams needing flexible reporting views. Subscription plans. 7. Looker Studio Flexible cross‑platform reporting and dashboards. Live dashboards from ad and analytics sources, shareable reports. Data‑centric advertisers and agencies. Free; connector costs. 8. Power BI Enterprise‑grade business intelligence (BI) with deep analytics. Advanced visualization, data modeling, secure dashboards. Enterprises with complex data needs. Microsoft subscription. 9. Supermetrics Centralizes advertising data for reporting and business intelligence. Pulls data from 100+ ad networks into sheets or dashboards. Teams building consolidated reporting stacks. Tiered subscription (from mid‑range). 10. Swydo Multichannel marketing reporting automation. KPI tracking, scheduled PDF/online reports, branding. Small-to-medium businesses (SMBs) and agencies with multi‑platform campaigns. Subscription (tiered). 11. TapClicks Unified analytics and reporting with automation. Integrates thousands of data sources, automated dashboards. Large agencies and enterprises. Custom/ subscription. ### 1. Realize Why it’s essential: Realize is an AI-powered performance platform that provides transparent, real-time reporting designed to bridge the gap between creative execution and bottom-line business outcomes. It’s used by performance marketers to gain a holistic view of their open-web campaigns, moving beyond vanity metrics to focus on deterministic data such as verified conversions, cost-per-acquisition (CPA), and return on ad spend (ROAS). Realize users can monitor how different creative variations, audience segments, and publisher placements are contributing to your funnel. The platform is used to visualize complex data through intuitive dashboards and automated reports, allowing teams to quickly identify top-performing assets and make data-backed decisions to scale or pivot their strategies instantly. Showcased features:  - Pacing Health Score provides a real-time visual indicator within the dashboard to show exactly how your budget is being utilized against your goals. - Performance Simulator uses historical reporting data to forecast future campaign outcomes, allowing you to model the impact of budget changes before implementation. - Delivers customized performance data directly to your inbox on a daily or weekly basis in CSV format for easy integration with external tools. - Offers robust reporting on both web-based and server-to-server events to capture deep-funnel conversions that occur offline or in CRMs. - Provides transparent reporting on which sites or creatives were automatically blocked by the algorithm to ensure you see exactly how your budget was protected. - Surfaces AI-driven insights directly alongside reporting metrics to suggest immediate actions for improving campaign efficiency. Best for: Realize is best for data-driven performance teams in sectors like e-commerce, finance, and travel that require granular visibility into the efficiency of their media spend. It’s particularly helpful for brands that need to report on complex, multi-step conversion funnels and want an automated way to track which specific content environments are driving the highest quality leads. Pricing model: Performance-based; campaigns billed on CPC basis, or CPM for programmatic. Pros:  - Shows exactly which publisher sites are delivering your conversions, ensuring full visibility into your media placements. - Integrates reporting with proactive recommendations, so you spend less time analyzing data and more time optimizing. - Capably tracks and reports on long consideration cycles through advanced server-to-server integration. Cons: - Automated reports are currently restricted to CSV files, which may require manual work for those preferring PDF or Slide formats. - Organizations that rely on real-time data streaming into their own custom BI tools may find the standard export options limiting. - Some advanced reporting visualizations, like those in the Performance Simulator, require a minimum volume of conversion data to populate accurately. ### 2. Cometly Why it’s essential: Cometly can serve as a corrective function for ad platforms, fixing attribution gaps that often happen when several platforms claim the same attribution or encounter browser blocking. It uses server-side tracking to capture conversion data, and feeds more accurate data back to an ad platform to improve targeting for new customers. Showcased features:  - AI-powered dashboard provides active recommendations for ad sets. - Shows multi-touch attribution and sends enriched conversion data back to Meta and Google to reduce CPA. Best for: Direct-to-consumer (D2C) brands spending a lot on social ads; teams that advertise on multiple channels and need to prove return on investment (ROI); and marketers who don’t trust conversion numbers from Meta or Google. Pricing model: Pricing scales based on monthly ad spend volume, with monthly cost starting around $199 for smaller spenders. Pros:  - Very useful for tracking conversions, particularly in a post-iOS 14 market. - Provides actions to take for ad performance improvement. - Fast, and offers a unified view. Cons:  - Hard to find specific pricing numbers, which is challenging for smaller businesses. - Focuses heavily on paid ads. ### 3. Agency Analytics Why it’s essential: Agency Analytics is an all-in-one client hub, bringing search engine optimization (SEO), PPC, and social media tools into one interface, with a focus on showing customer value through portals and automated reporting. Showcased features:  - Lets users add their own branding. - Includes built-in SEO tools, plus automated KPI tracking and alerting. - Ideal for marketers who need to present to clients or stakeholders frequently, as the results are very polished. Best for: Agencies or teams managing clients that want to standardize and scale reporting processes, and full-service marketers that want all results in one dashboard. Pricing model: Pricing based on how many client campaigns are running; it starts around $59 per month and goes up to $349 per month, with more features at each tier. Discounting is available. Pros:  - Easy to set up and build reports. - Built-in SEO tools. - Strong support, and the option to give clients their own dashboard login. Cons:  - Costs can quickly grow because of the per-client model. - Some features are locked for the lower-cost plans. - Easy to use, but there aren’t many customization options. ### 4. Databox Why it’s essential: Databox offers holistic performance tracking, pulling data from platforms and devices for a 360-degree company view. Plus, its mobile app is well-developed, and the platform focuses on trackable key performance indicators (KPIs). Showcased features:  - Set targets for any metric. - Tracks progress in real time. - Lets users anonymously compare performance metrics against competitors. - Allows users to develop custom KPIs by combining data from different sources. Best for: Performance marketers who have to align with revenue and sales pipeline data, leaders who want a high-level business view quickly, and any team that wants to see marketing and sales metrics in the same interactive portal. Pricing model: Free plan includes three data sources and daily data refreshes, while tiered subscriptions start at $47 per month and go up to $799 per month. The cost scales according to how many data sources you connect. Pros:  - Easy alerting, plus scorecard and report creation. - Modern, intuitive interface. - Top-notch mobile experience. - Allows comparison against competitors and provides visualizations of actual vs. target data. Cons: - Dashboard design is somewhat rigid. - Users have to pay more to get hourly or 15-minute data syncing; otherwise it’s just once a day. ### 5. Google Analytics (GA4) Why it’s essential: GA4 is an Industry standard for understanding how users find and use your site or app. Showcased features: - Machine learning (ML) and data modeling fill gaps created by cookie sunsetting in a secure way. - Predicts future behavior in GA4, and tracks common interactions automatically without additional code or setup. Best for: Google ecosystem users, companies with both a website and mobile app. Pricing model: Most consumers can use Google Analytics for free, with some limits on data retention and number of custom metrics. The enterprise tier, GA360, is usage-based, starting around $50,000 per year. Pros: - Widely used, well-established tool. - Can offer a lot of detail, particularly for Google Ads users. - The free version offers lots of capabilities. - Cross-device tracking and near-real-time metrics are very useful. Cons:  - Steep learning curve, with expertise needed to get the full functionality of GA4, though there are free courses and certifications offered through Google. ### 6. Klipfolio Why it’s essential: Klipfolio offers many customization options, allowing users to define a metric once and then use it across dashboards. Plus, it uses real-time visualization to show marketing performance minute by minute. Showcased features:  - Low-code Editor transforms raw data into custom visualizations. - Allows users to store historical performance data to view trends over years. - Makes it easy to join data from various platforms, and layer two sources to see correlations. Best for: Technical performance marketers who want more control over what they build, teams without big budgets who want sophisticated data modeling, and teams with unique KPIs. Pricing model: Free version is good for up to two users, then paid tiers range from $90 to $800 per month, with improved refresh rates, data limits, and white-labeling available the more you pay. Pros:  - Lots of flexibility in building calculations. - Good price for the features. - Permanent data storage to protect from API changes or expirations. Cons: - Steep learning curve for the formula language. - Can be more time-consuming, since there’s a lot of customization available. ### 7. Looker Studio (formerly Google Data Studio) Why it’s essential: This platform turns raw data from disparate sources into visual dashboards and makes it available to non-technical users via live sharing. Showcased features:  - More than 800 data connectors to Google and other platforms available. - Users can join up to five sources in one chart. - Google users will find reports easy to share. - Simple creation of automated, real-time updated client reports. Best for: With its low price point, Looker Studio is accessible for small businesses. Pricing model: Free version includes core features, while Looker Studio Pro is $9 per user a month. Partner connectors may require a separate monthly subscription. Pros:  - Very easy to build dashboards without expertise. - Web-based. - Includes hundreds of free templates. Cons:  - Dashboards can be slow with large datasets. - Reports for the free tier are tied to individuals. - Support in the free version is only on community forums. ### 8. Power BI Why it’s essential: Power BI is a general-purpose business intelligence platform, so it can process huge amounts of data quickly, and analyze complex data beyond what a dashboard can do. It includes an Extract, Transform, Load (ETL) tool to clean and merge data before using it. Showcased features: - Support for natural language questions, with AI-generated charts in response. - Specialized charts that use ML to show which variables are driving conversions. Best for: Those using Microsoft tools already; marketers who are technical or have data analysis expertise; and big enterprises combining marketing and internal data frequently. Pricing model: Power BI Desktop is free, with no limit on the number of reports, but no sharing capabilities. Power BI Pro is $10 per user per month and the Premium option is $20 per user per month, with more functionality at each tier. Pros:  - Lots of features for the cost. - Sophisticated customization options. - Easy integration with Excel and other Microsoft tools. Cons:  - Steep learning curve to get the most out of Power BI. - No native connectors for ad platforms. ### 9. Supermetrics Why it’s essential: Supermetrics is a data pipeline automation leader, eliminating the need for manual imports and pulling data in from sources like LinkedIn or Meta Ads for de-siloed data insights and reporting. Showcased features:  - Offers 150+ native connectors to ad platforms, and many more. - Provides options on where to send data. - Enables custom calculations and automated refreshes. Best for: This is a solid tool for performance marketers running ads across five or more platforms, and teams that need to scale report creation without hiring a data engineer. Pricing model: Tiered pricing goes from $29 to $399 per month, or the per-destination model charges based on where you’re sending data. Pros:  - Massive integration library. - No code required - Great templates. - Reliable, stable connectors vs. less costly competitors. Cons:  - Costs can increase quickly. - No visualization interface, so you’ll need another tool to view data. ### 10. Swydo Why it’s essential: Swydo is great for efficiency, since it automates workflows, going beyond dashboards to actually get the data, build the report, and email a PDF. There are also proactive monitoring features. Showcased features: - You can merge up to five ad platforms’ data for particular data points, like total ad spend. - The platform constantly monitors API connections. Best for: Great for PPC specialists managing large ad budgets, and good for teams that want to scale beyond a free product. Plus, Swydo has an unlimited users policy. Pricing model: Volume-based model, so users pay by how much data they use, and everyone can access the same features. Base plan starts at $69 per month. Pros:  - No charges for extra users or dashboards. - Pricing is predictable and transparent. - Support is strong. - The health check feature ensures that no connected data goes down. Cons:  - Less flexible for layout customization. - Smaller integration library of about 35 native connectors. - Users have to manually remove old data sources, or you’ll be billed accordingly. ### 11. TapClicks Why it’s essential: This platform is built for massive scale, with thousands of users and data streams, and is used by large and multi-location companies. Beyond reporting, TapClicks includes order entry and workflow management, and can incorporate offline data like Connected TV (CTV) or out-of-home (OOH). Showcased features:  - Smart Connector tool offers thousands of data sources to pull from. - Visualization engine supports data grouping for visual reports. - Can send automated alerts for over- or underspending. - Compares with competitors to see what they’re publishing and spending. Best for: Any company managing a massive budget with lots of clients, such as national brands with local franchises and media companies with omnichannel results. This tool can serve as a global command center. Pricing model: Premium pricing is normally based on the number of users, the number of client records, and which module (reporting, orders, workflow, etc.) is needed. Budget for high hundreds or low thousands of dollars per month for this platform. Pros:  - End-to-end platform can handle the whole workflow from sales to reporting. - Brings data depth from proprietary or niche sources. - Creates clear summaries, even from complex data. Cons: - Setup is complex and more technical than other options. - User interface (UI) can be overwhelming for a smaller team. - Support may be more complicated and require engineering input. ## Understanding Campaign Reporting for Performance Advertisers ### What Is Campaign Reporting for Performance Marketing? For performance marketing teams, campaign reporting refers to the work of tracking, analyzing, and visualizing data to see how effective a campaign has been. Marketers have to gauge performance against specific KPIs that align to business goals. Campaign reporting can show areas of success as well as opportunities to improve. Continuous reporting and iterating are essential for a performance marketing team to get the most out of their budget and engage audiences. Performance marketers ultimately have to show how their efforts are contributing to revenue, so campaign reporting is an always-on function. A report might compare click-through rates (CTR) across paid search channels to show which is performing best, so a marketer can reallocate budget accordingly. Every company’s campaign reports will reflect which channels matter most to them. ### What Should a Performance Marketing Report Include? A performance marketing report should include analysis along with data so that it isn’t just surface-level. At minimum, the report should include: - Executive summary: A high-level overview of results with broad insights or recommendations. - Metrics or KPIs: Explain which KPIs are in use and why. - Channel data: show performance across specific channels to see highs and lows, with visualizations. - Trends analysis: Discuss why numbers changed or stayed stagnant to show trends over time, also with visualizations. - Next steps: Recommend optimizations, budget adjustments, or new strategies or tactics to explore. ### Key Metrics for Performance Advertising Campaigns Performance advertising metrics typically fall into these categories: - Revenue metrics: ROAS, ROI. - Cost metrics: CPA, CPC, CPM. - Conversion metrics: Conversion rate, total leads, landing page conversion rate. - Engagement and traffic: CTR, impressions, page views, scroll depth, and other session data. ### Automating Performance Advertising Reports Performance advertising reports may have been simpler before walled gardens and the open web added more complexity to the assessments. Automating reports can save performance marketers a lot of time when they’re gathering campaign data from multiple sources. Automation can help eliminate manual spreadsheet exports and reconcile data quickly, and set up integrations and connectors to avoid any manual data entry or lookups. A typical approach to automating reporting might look like this: - Define the KPIs/metrics. - Integrate data sources into your preferred tool. - Standardize terms and naming across all platforms. - Schedule delivery or live dashboard update times. - Use AI alerts, if available, to see any performance anomalies in real time. ## Key Takeaways Campaign reporting should be complete, accurate, and frequently updated or published by performance platforms. Performance marketers can use these reports as a collaboration tool, as well as a gauge of how well their ad campaigns are working for their business. Some strong performance platforms are available to implement campaign reporting best practices, so advertisers should carefully consider their needs, as well as budget and future possibilities, before adopting one. ## Frequently Asked Questions (FAQs) ### What distinguishes a reporting platform from analytics tools? A reporting platform shows historic data from a particular time period, with a particular set of parameters. It uses visualization, whether that’s a chart, graph, heatmap, or other option, to show what happened. Analytics tools, meanwhile, explore data to explain why something happened, such as a dip in page views or conversion rates. Reporting platforms are geared toward informing users through pre-built reports or dashboards. Analytics tools are typically more complex and can interpret large data sets to improve performance. ### Should I use an all‑in‑one reporting tool or connect multiple tools (e.g., Supermetrics + Power BI)? It depends on your business’ size, goals, budget, staff, and number of data sources. An all-in-one tool may work well for a smaller or startup business that needs a quick solution, and doesn’t have technical staff or many data sources. Consider connecting multiple tools if you need complex data combinations, big datasets, and in-house analytical capabilities. ### How do these platforms help with attribution measurement? Performance marketers need to understand which tactics and channels are best for lead generation and revenue, so accurate attribution is essential. Attribution platforms aggregate data across channels to show the full customer journey. These platforms might use server-side tracking, pixels, AI, and multi-touch modeling to show which touchpoints are working. They also aggregate data from multiple sources for a more complete picture. --- ### Best AI Performance Platforms for Ad Creative Management URL: https://www.taboola.com/marketing-hub/best-ai-performance-platforms-for-ad-creative-management/ Last Modified: 2026-03-23 14:48:24 In performance marketing, your creative isn’t a pretty addition tacked on at the end; it’s the system that decides whether your targeting, bidding, and landing page work get to matter. The best teams don’t just ship ads, they produce variations quickly, learn what’s actually driving outcomes, and scale winners without turning the process into a never-ending scramble for new ideas. That’s where AI creative performance platforms earn their keep. Done right, they help you move faster and smarter by turning ad creative into an operational advantage, instead of a recurring bottleneck. ## 10 Best Performance Advertising Platforms for Ad Creative Management Compared Platform Why It’s Essential Core Use Cases and Features Best for  Pricing Model (Indicative) 1. Realize AI-driven creative performance prediction and optimization. Generate creative insights, test ad variations, integrate with campaigns for predictive return on investment (ROI). Marketers needing performance-backed creative insights across channels. Performance-based model; campaigns billed on CPC basis, or cost-per-mille (CPM) for programmatic. 2. AdCreative.ai AI generation and scoring of ad creatives across formats. Generate visuals and copy; creative performance scoring; competitor insights. Rapid ad creative generation with predictive performance scoring. About $39–$599+ per month (tiered credits and brands). 3. Canva Ads (AI) Accessible AI design with brand consistency. Bulk create ad variations; Magic Studio features; brand kits. Teams needing speedy ad drafts with design control Subscription (Pro plan required for full features). 4. Connected‑Stories AI workflows for personalization and creative orchestration. Campaign brief → personalized content strategy; GenAI chat interface. Brands needing tailored content at scale. Custom pricing. 5. Gethookd AI ad research and user-generated content (UGC)‑style creative generation. Competitor analysis; UGC video generation; script creation. Teams focused on social ads and UGC creatives. About $47/mo (indicative). 6. Madgicx Creative intelligence and automated ad management. AI insights across creatives; budget shifts toward best creations. Performance marketers running multi‑platform campaigns. Tiered/custom pricing. 7. Pencil Predictive creative generation with performance data. AI video/image generation; predictive scoring; ad variation suggestions. Marketers focused on high‑quality, data‑driven ad iterations. Free trial; starts at about $14 per month; custom enterprise. 8. Segwise Creative analytics and AI variation generation. AI tag analytics tied to performance; generates new variations based on winning elements. Performance marketing teams running large tests. Custom pricing (enterprise‑oriented). 9. Solara AI AI content plus automated campaign posting and optimization. Authentically styled ads; automated posting and performance tracking. Businesses wanting full campaign automation. Custom pricing. 10. Vibemyad AI workflow automation and performance insights. Ad research database; workflow automation; performance audits. End‑to‑end creative strategy and production teams. Custom pricing. ### 1. Realize Why it’s essential: Realize is a comprehensive performance platform powered by AI that enables advertisers to manage, optimize, and scale ad campaigns across a massive network of premium publishers. It serves as a central hub for bridging the gap between social media assets and the open web, using predictive algorithms to match content with high-intent audiences in real-time. Use Realize to transform static assets into dynamic native, video, or display formats that blend seamlessly with publisher content. It’s primarily used to automate the creative production cycle — from generating AI-driven variations to importing top-performing social posts — while using real-time engagement signals to ensure budget is directed toward the highest-performing supply. Showcased features: - Social Importer: Automatically repurposes existing Facebook and Instagram creatives into high-performing display ads for use on the open web. - GenAI Motion Ads: Transforms static images into short, looping motion-based creatives that drive higher engagement and conversion rates in native environments. - Abby: An AI performance expert that provides real-time fixes for creative rejections and automates the troubleshooting of non-serving campaigns. - SpendGuard: An optimization algorithm that automatically blocks underperforming creatives and sites to minimize wasted spend in real-time. Best for: Realize is ideal for performance marketers, such as those in D2C or high-consideration industries, who need to scale creative production without expanding their design teams. It’s particularly helpful for brands moving away from social walled gardens that want to leverage existing social assets on premium news and tech sites to reach new, incremental audiences. Pricing model: Performance-based model; campaigns billed on CPC basis, or cost-per-mille (CPM) for programmatic. Pros: - Seamless Cross-Channel Scaling: Easily repurposes existing social media assets for the open web with minimal manual effort. - Automated Creative Production: Generates high-quality, platform-optimized ad variations using built-in generative AI tools. - Enhanced Performance AI: Leverages predictive algorithms to automatically optimize bidding and creative delivery for a lower CPA. Cons: - Advanced Tracking Integration: For brands tracking deep-funnel CRM or offline actions, the platform utilizes a robust Server-to-Server (S2S) setup, which offers higher data precision but involves more technical coordination than standard pixel deployment. - Asset Integrity Focus: To ensure creative quality remains consistent with the original source, assets brought in via the Social Importer are treated as ready-to-go, meaning any fine-tuning should be finalized within your design suite before the final upload. - Strategic Market Rollouts: To maintain high data accuracy, specialized tools like Search Keyword and Mail Domain Targeting are currently prioritized for specific high-intent markets as the platform continues its global expansion. ### 2. AdCreative.ai Why it’s essential: Most accounts don’t have a targeting problem, they have a creative volume problem. AdCreative.ai is built to solve for this core performance reality. The platform helps advertisers generate large batches of static ad creatives and copy quickly, then uses AI scoring to prioritize the concepts most likely to perform. In the context of ad creative management, you would use AdCreative.ai to accelerate your creative testing cadence. It’s especially useful when you need to keep multiple campaigns fresh with new angles, layouts, and hooks, without waiting on a full design cycle. Showcased features: - AI Creative Generation: Produces static ad creatives across common formats and sizes to support rapid testing. - AI Copy Generation: Generates headlines and primary text variations aligned to different offers and audiences. - Predictive Creative Scoring: Ranks creatives with performance-oriented signals so teams can shortlist what to test first. - Brand and Asset Management: Organizes brands, creative libraries, and generation workflows to support repeatable production. - Competitor/Market inspiration: Surfaces patterns and examples to guide creative direction and iteration. Best for: AdCreative.ai is ideal for lean performance teams, affiliates, and direct-to-consumer (DTC) marketers who need high output and fast iteration, especially when creative fatigue is the main limiter on scale. It’s also a practical fit for agencies managing multiple clients who need to deliver a steady flow of fresh creatives across offers and geos. Pricing model: Pricing is offered in a tiered monthly or annual subscription model based on usage (credits for downloads), starting around $39/month for small, single-user plans up to $599+/month for agencies. Plans offer varying limits on users, brands, and creative generation/downloads, with annual plans offering up to 50% savings. Pros: - High-volume output: Produces many variations quickly to keep tests running continuously. - Built-in prioritization: Scoring helps teams avoid testing everything and focus on the most promising options first. - Operational efficiency: Reduces dependence on full-time design cycles for early-stage testing and exploration. Cons: - Template risk: Without strong direction, outputs can look generic across campaigns. - Scoring isn’t proof: Predictive signals can guide prioritization, but controlled testing is still required to validate lift. - Static-first emphasis: Best suited to rapid static iteration. Motion and video workflows may require supplemental tools. ### 3. Canva Ads (AI) Why it’s essential: It’s easy to generate ideas, but teams often struggle to produce enough on-brand variants across sizes and formats without slowing down. Canva makes creative production scalable for non-designers while still protecting brand consistency through templates and brand kits. In the context of ad creative management, Canva turns creative concepts into launch-ready assets, quickly resizing, formatting, and producing a high volume of variations that still feel cohesive. Showcased features: - Magic Studio AI Tools: Accelerates creation, editing, and variation production. - Brand Kits: Maintains brand consistency across fonts, colors, logos, and templates. - Bulk Create: Produces many variations quickly for different audiences, offers, or geos. - Template-based Production: Standardizes output so teams can scale without reinventing design each time. - Collaboration and Approvals: Supports team workflows for review and iteration. Best for: Canva is ideal for teams that need fast creative drafts and scaled production — especially lean performance teams and agencies that want speed without sacrificing brand consistency. Pricing model: Subscription-based; Pro plan typically required for full AI and workflow feature access. Pros: - Fast on-brand scaling: Makes it easy to produce many variations without design bottlenecks. - Accessible to non-designers: Lowers the barrier to creating and iterating quickly. - Great for production ops: Strong for resizing, formatting, and standardizing ad output. Cons: - Analytics-light: Not purpose-built for creative performance intelligence, so you will have to pair it with testing and reporting tools. - Can encourage “more, not better” workflow: Volume is easy, but strategic differentiation still requires direction. - Motion/advanced formats vary: More complex motion/video may require complementary tools. ### 4. Connected-Stories Why it’s essential: Connected-Stories is built for teams that need personalization and orchestration, not just asset generation. As performance programs evolve, “one creative” rarely fits every audience, placement, or moment in the customer journey. Connected-Stories approaches creative as a system that can be tailored and scaled. It translates campaign strategy into personalized content paths that help brands generate, adapt, and coordinate creative experiences across segments at scale. Showcased features: - Brief-to-Strategy Workflow: Transforms campaign inputs into structured content directions and personalized creative plans. - GenAI Creative Orchestration: Uses AI-driven workflows to manage variations and tailor content across audiences. - Chat-based Creative interface: Enables teams to iterate through prompts and guided workflows for faster production. - Personalization at Scale: Supports multi-audience creative programs where segmentation is central to performance. - Operational Coordination: Helps align creative production with distribution and optimization requirements. Best for: Connected-Stories is ideal for brands with multiple audiences, products, or lifecycle stages that require personalized messaging, when orchestration and governance matter as much as generation. Pricing model: Custom pricing; typically enterprise-oriented based on scope and workflow complexity. Pros: - Personalization native: Strong fit for multi-segment creative strategies. - System thinking: Helps connect strategy, production, and variation management into one workflow. - Scalable governance: Useful when consistency and coordination matter across many outputs. Cons: - More than you need for simple use cases: Can be overkill if your primary need is fast asset generation. - Best with mature creative ops: Works best when teams have defined segmentation and messaging frameworks. - Custom implementation: Expect onboarding and configuration to maximize value. ### 5. Gethookd Why it’s essential: Gethookd is designed for a world where UGC-style creative and competitive pattern recognition drive performance. In many categories, winning ads don’t look like they’re made for TV, they look like they were filmed on a phone by someone in the driver’s seat of their car: they’re convincing, fast, and built around hooks that match how people actually consume content. Teams get the most use out of Gethookd by researching what’s working in your market to generate UGC-style scripts and variations, and keep your creative pipeline stocked with new hooks that feel native to social environments. Showcased features: - Competitor and Market Research: Helps teams identify patterns and angles from existing ad ecosystems. - UGC Video Generation: Supports creation of UGC-style video assets designed for social performance. - Script and Hook Creation: Generates scripts, hooks, and variations for rapid production cycles. - Angle Exploration: Helps teams expand creative coverage across value props, objections, and use cases. - Iteration Workflow: Encourages fast “build → test → refresh” cycles in UGC-heavy categories. Best for: Gethookd is ideal for teams running social-heavy programs that depend on UGC-style ads and need to scale hooks, scripts, and variations quickly. Pricing model: Subscription-based; indicative monthly pricing with tiered plans. Pros: - UGC velocity: Speeds up concepting and scripting — the hardest bottlenecks in UGC production. - Research-led creative: Anchors creative direction in market patterns instead of guesswork. - Hook coverage: Makes it easier to test multiple angles quickly across the funnel. Cons: - Brand QA required: UGC-style output needs oversight to avoid off-brand tone or credibility issues. - Risk of derivative creative: Market-based inspiration can drift too close to “copying” without strong strategy. - Not a full performance platform: Best paired with strong measurement and testing discipline. ### 6. Madgicx Why it’s essential: Madgicx is designed for performance teams that want creative insights connected to optimization decisions. When you’re managing many campaigns, you need a workflow that helps allocate attention and budget toward what’s working and away from what’s wasting spend. Madgicx lets teams monitor creative performance, identify actionable signals, and support automated or semi-automated optimization workflows that help scale winners and manage fatigue. Showcased features: - Creative Performance insights: Helps analyze which creatives and themes drive results. - Optimization Workflows: Supports shifting strategy toward top-performing assets and campaigns. - Multi-campaign Management: Built for teams operating at scale across many active initiatives. - Automation Features: Designed to reduce manual optimization and accelerate decision cycles. - Performance-oriented Tooling: Emphasizes outcomes like return on ad spend (ROAS)/cost per action (CPA) rather than just asset production. Best for: Madgicx is ideal for performance marketers managing high campaign volume who want an optimization-oriented system that helps keep creative performance aligned with budget efficiency. Pricing model: Madgicx employs a dynamic spend-based pricing structure where your monthly software fee scales directly with your ad budget, starting at approximately $44 per month for accounts spending under $1,000. As your advertising volume increases, the subscription cost rises through defined tiers, reaching upwards of $500 monthly for high-spend accounts before transitioning to custom Enterprise quotes. Pros: - Optimization-driven: Built for teams that want creative insights to feed into action. - Scales with campaign volume: Useful when manual optimization becomes too slow. - Efficiency focus: Helps reduce wasted spend by highlighting what to prioritize. Cons: - Cost can scale: As usage/spend grows, pricing may increase — model ROI carefully. - Depends on clean measurement: Creative optimization is only as good as the underlying attribution signals. - Not a pure creation tool: Best paired with strong creative production resources or generation tools. ### 7. Pencil Why it’s essential: Pencil is designed around frictionless iteration. This feature is key to your ability to manage campaigns: Instead of treating creative generation as a one-time event, Pencil supports an ongoing cycle of producing image and video variations informed by performance signals. Pencil generates learnings and turns them into new iterations faster to keep creative fresh, expand angle coverage, and reduce the lag between “what we learned” and “what we ship next.” Showcased features: - AI Image and Video Generation: Produces new creatives and variations designed for performance marketing use cases. - Variation Suggestions: Generates alternative hooks, visuals, and formats to explore adjacent winning concepts. - Performance-informed Iteration: Supports workflows that connect results to next-round creative development. - Creative Production Workflow: Helps teams standardize creative development from inputs to variations and finally to deployment-ready assets. - Brand Guardrails: Maintains consistency through structured inputs and templates. Best for: Pencil is ideal for performance marketers who already run structured creative testing and want a platform purpose-built for iteration — especially teams that need to scale video and motion alongside static assets. Pricing model: Pencil operates on a tiered subscription model driven by “Generation Credits” and seat access, starting at $14 per month for individuals and scaling to $119+ for commercial teams. The model is split into Self-Serve (credit-card based) and enterprise (contract-based) tiers. Most plans include a 7-day free trial to test the ad generation engine. Pros: - Iteration engine: Strong fit for “always-on” creative testing programs. - Video-friendly workflow: Useful when your roadmap includes scaling motion and short-form assets. - Speed-to-next-test: Helps compress the loop between insight and new creative deployment. Cons: - Needs clear inputs: Without clear angles, offers, and positioning, AI iteration can amplify confusion instead of clarity. - Adoption matters: Workflow value is highest when teams commit to using the platform consistently. - Not a standalone measurement strategy: You still need clean attribution and controlled testing to quantify creative lift. ### 8. Segwise Why it’s essential: Segwise focuses on the part of creative management that breaks first at scale: knowing what’s actually working. When you’re running dozens or hundreds of assets across platforms, it becomes hard to distinguish true winners from noise or to detect fatigue before performance drops. To effectively manage your creatives, Segwise turns scattered creative results into actionable patterns. It helps teams connect creative elements to outcomes, making it easier to replicate what works and systematically generate better variations. Showcased features: - Automated Creative Tagging: Classifies creative elements (hooks, formats, themes) so performance can be analyzed at the component level. - Creative Performance Analytics: Connects asset-level and element-level signals to performance outcomes across tests. - Fatigue and Lifecycle Signals: Helps identify when creatives are losing effectiveness and need refreshes. - Insight-to-Variation Workflow: Guides creation of new variations based on winning patterns rather than guesswork. - Cross-Account/Portfolio Views: Useful for teams managing many campaigns or brands. Best for: Segwise is ideal for performance marketing teams running large creative test volumes who need clarity, consistency, and repeatability when creative reporting has become too manual or too slow to drive decisions. Pricing model: Segwise operates on a custom enterprise pricing model based on scale, integrations, and reporting scope. Pros: - Pattern discovery: Makes it easier to understand why something worked and what to build next. - Faster creative decisioning: Reduces analysis time and helps avoid “testing blind.” - Scales with volume: Becomes more valuable as creative volume increases and manual reporting becomes unreliable. Cons: - Requires hygiene: Naming conventions and asset discipline impact the quality of insights. - Implementation lift: Enterprise analytics tools typically require setup and integration effort. - Analytics-first: Best paired with a strong production workflow to turn insights into new creatives quickly. ### 9. Solara AI Why it’s essential: Solara AI sits in the automation-first category, aiming to reduce the manual workload of campaign execution and creative deployment. For smaller teams, that’s often the real constraint: not the lack of ideas, but the lack of time to consistently produce, launch, optimize, and report. Solara AI allows teams to streamline content creation and automate parts of campaign posting and optimization to keep your pipeline moving even when bandwidth is tight. Showcased features: - AI Content Creation: Generates ad-like content designed to match brand tone and style. - Automated Posting and Execution: Helps reduce manual steps in launching and managing campaigns. - Performance Tracking: Provides feedback loops to inform optimization decisions. - Workflow Automation: Supports teams that want a more hands-off operating model. - End-to-End Assistance: Positioned to combine creation with execution rather than separating the two. Best for: Solara AI is ideal for resource-constrained teams that want more automation across campaign management and creative execution — especially those that value simplicity and reduced manual work. Pricing model: Typically custom pricing; may vary based on supported channels and automation scope. Pros: - Bandwidth relief: Reduces operational load for small teams. - Execution-oriented: Focuses on getting campaigns live and managed, not just generating assets. - Streamlined workflow: Can simplify multi-step processes into a single toolset. Cons: - Integration depth varies: Confirm ad platform integrations and reporting depth early. - Less control by design: Automation can trade off with granular creative and optimization control. - Validation required: “All-in-one” tools require careful evaluation to ensure they match your needs. ### 10. Vibemyad Why it’s essential: In performance marketing, the difference between average and elite teams is often how quickly they can move from research to production, from production to testing, and from testing to audited learning. Vibemyad is built for teams that want to operationalize creative strategy. To manage your creative workflow, you can use Vibemyad as a workflow layer that connects research, creative planning, production, and performance audits into a continuously updating system. Showcased features: - Ad Research Database: Helps teams study creative patterns, competitive positioning, and emerging angles. - Workflow Automation: Supports creative operations — organizing briefs, concepts, iterations, and outputs. - Performance Audits: Provides structured ways to diagnose what’s working, what’s failing, and why. - End-to-End Creative Planning: Helps connect strategy with production so outputs align to test goals. - Team Collaboration Tools: Useful for aligning stakeholders around creative direction and iteration. Best for: Vibemyad is ideal for creative strategy and production teams, agencies, and performance organizations that want a repeatable process to research, build, and audit creative — especially when multiple stakeholders are involved. Pricing model: Often custom pricing; may include tiered plans depending on features and scale. Pros: - Process-driven: Helps turn creative work into a repeatable operating system. - Stronger alignment: Makes it easier for teams to agree on what to build and why. - Audit mindset: Encourages learning loops that prevent repeating the same mistakes. Cons: - Requires adoption: Workflow value depends on consistent team usage. - Not just a generator: Teams looking only for instant creative output may not use the platform fully. - Best with defined goals: Strongest results happen when audits feed into planned test roadmaps. ## More About Performance Ad Creatives ### What Is an AI Ad Creative Platform? An AI ad creative platform is any system that uses AI to help you create, evaluate, or improve ad assets. Some tools are generation-first, to make more creatives faster, and others are intelligence-first to help you understand what’s driving performance. The most valuable platforms increasingly combine both: they help you produce variations and then learn from results so the next set is better than the last. ### How Does AI Improve Ad Creative Performance? AI improves creative performance by compressing time. It shortens the distance between an insight and a new test. Instead of waiting for a designer, rewriting briefs, or debating what to try next, AI can quickly produce structured variations and help teams explore more angles while maintaining consistency. The performance advantage combines velocity and discipline: You can test more intelligently, and sooner, and stay ahead of creative fatigue. ### What Are the Key Features to Look for in AI Ad Creative Platforms? The strongest platforms tend to share a few practical traits. First, they make it easy to generate variations across formats and sizes without breaking your workflow. Second, they provide enough governance of templates, brand controls, and approvals so that scaling doesn’t destroy consistency. Finally, they provide analytics that reveal which elements are driving performance, signals that indicate fatigue, and integrations that make it easy to deploy and measure without manual overhead. Ultimately, you’re looking for a platform that improves the full creative loop by improving your ability to manage the entire process. ### How Are AI Ad Creative Platforms’ Pricing Structured? Pricing typically maps to what the platform sells. Generation tools usually charge subscriptions tied to credits or usage volume. Intelligence and workflow platforms are often enterprise-priced because they integrate into broader reporting systems. Media platforms with creative capabilities frequently use performance-based pricing because creative is part of a larger distribution and optimization system. The simplest way to evaluate pricing is to ask: Does this tool reduce production cost, reduce wasted spend, increase conversion efficiency, or all three? If it doesn’t move at least one of those levers, it’s hard to justify at scale. ### AI Creative Generation vs. Manual A/B Testing AI doesn’t replace testing; rather, it changes what you can test and how much you get out of the results. The best approach is to use AI to generate high-quality hypotheses and structured variations, then let controlled tests validate what works. Predictive scoring can help you prioritize, but real-world performance is shaped by your audience, offer, landing page, attribution model, seasonality, and platform dynamics. Manual A/B testing is still a proof layer, however. AI is the acceleration layer that keeps the proof pipeline full. ### Case Studies of AI Improving Ad Campaign ROI A recent large-scale performance advertising case study led by Columbia University found that AI can scale creative experimentation without creating a performance penalty, if you use it to produce ads that still follow human-centered best practices. In one of the largest live analyses of GenAI display advertising to date, researchers from Columbia, Harvard, Technical University Munich (TUM), and Carnegie Mellon partnered with Taboola’s Creative Shop and used Realize performance data to compare AI-generated and human-made ads in real market conditions. Across the full dataset, AI-generated ads delivered comparable click-through rates (CTR) to human-made ads, and under strict controls the performance was statistically equivalent. The key takeaway is that generative AI helps teams to expand their creative surface area, then lets performance signals sort winners from losers without worrying that you’re trading quality for scale. The research also found that ads perceived as AI-generated underperform, regardless of whether they were actually made by AI or by humans. The takeaway here is straightforward: To increase your ROAS, don’t just use AI for the sake of using AI, use AI to generate authentic-feeling creative that doesn’t trigger AI skepticism. What helps creative feel human? The study found that using large, clear human faces was the single most influential factor in making ads feel human-made and in driving higher engagement. On the other hand, visual cues like overly stylized or highly polished imagery, heavy color saturation, and strong symmetry — whether AI-generated or human-made — decreased confidence in the ads. ## Key Takeaways AI creative platforms are the infrastructure for performance teams that want to scale. The right tool depends on your bottleneck: If you need more output, generation tools can help you produce variations without expanding headcount. If you need clearer learning, creative intelligence platforms help you see why something worked and what to do next. And, if you want creative and optimization tightly connected, performance platforms that treat creative as part of the delivery system can reduce the gap between insight and impact. The winning pattern across all of them is consistent: build a creative loop you can run every week, not a one-off burst you hope carries the quarter. ## Frequently Asked Questions (FAQs) ### Which AI creative platform is best for performance testing? If by “performance testing” you mean understanding which creative elements actually drive outcomes, you’ll typically want a platform that’s analytics- and insights-forward. If you mean performance testing as “launch variations and optimize them at scale,” platforms that combine creative tooling with distribution and optimization can make testing feel less like a project and more like an always-on engine. ### What are the best AI creative platforms for e-commerce advertising? E-commerce teams usually win with the combination of a tool that makes it easy to generate and format large volumes of product-led variations, and a system that keeps learning tight as performance shifts week to week. Different platforms tout different advantages. The industry leaders are AdCreative.ai for high-volume static performance, Realize for closing the performance gap where increasing spend on search or social leads to diminishing returns due to audience saturation, and Pencil for brand-safe prediction. ### Are these platforms suitable for both static and motion ads? Increasingly, yes. Some tools excel at rapid static iteration, while others emphasize video and UGC-style outputs. Platforms that can turn static into motion and automate variants tend to make motion more accessible for performance teams, though these tools run the risk of flagging to the viewer that they are AI-generated. ### Do AI creative tools integrate with ad platforms? Many do, but the depth varies. Some integrate directly through imports and publishing workflows. Others operate as production layers where you export assets and then deploy them manually. If integrations matter to you, validate the exact platforms supported and confirm whether performance data flows back cleanly enough to support real iteration, rather than spreadsheet archaeology. --- ### Retargeting vs. Lookalike Targeting URL: https://www.taboola.com/marketing-hub/retargeting-vs-lookalike-targeting/ Last Modified: 2026-03-23 09:30:19 Historically, performance marketing has never required as much strategy and execution as it needs today. With tight marketing budgets, rising acquisition costs, time-strapped teams, and a scattered user journey, search and owned platforms alone may not drive sustainable growth. Since understanding customers is key to getting in front of them and enticing them to engage, two advertising strategies rise to the top: retargeting and lookalike targeting. While both can help solve modern-day marketing challenges, and are both often lumped together in overarching plans and budgets, they each solve a different problem at different points in the customer journey. Understanding what each tactic really means, when each should be deployed, and the considerations and outcomes of each, will help create a campaign that executes for maximum return. ## Retargeting Retargeting is a lower-funnel marketing tactic that uses behavioral intent to serve more effective ads to nudge users into action. To be most effective, this type of advertising needs to appeal to the mindset the user has when engaging online, with a goal of conversion. ### Description Retargeting allows for people who are searching online, but have not yet taken action, to be served digital ads about the brand, product, goods, or service. The idea is to give a friendly reminder to those who have shown interest but haven't yet converted, nudging them toward the intended activity. ### How It Works User actions (such as browsing history, clicks, engagement, or abandonments) are tracked across devices and platforms via cookies, pixels, or tags installed on your website. If a user abandons items in their cart, e.g., email retargeting would see an email sent to the user as a reminder, specifically encouraging them to return to their cart. ### Benefits Because retargeting audiences are already familiar with your brand, and are lower in the marketing funnel, retargeting tactics produce more engagement and conversion. As such, the cost per click is lower than with many other tactics, since these ads are being served to a warmer audience. Repeat brand awareness is another retargeting benefit, as the brand, product, or service is served as a reminder. Retargeting ads can even offer personalization, based on information collected from the user or their experience, which can help to create a brand connection. ### Considerations There is a limit to both the audiences available for retargeting — i.e., those who have previously interacted with you — and how many times you can get in front of them before diminished ROAS occurs. If the campaign isn’t well optimized, ad funds may be wasted on people who have already converted. Retargeting can also be subject to privacy concerns over use of personal data, and requires user consent for personalization through GDPR, CCPA, and other similar regulations. ### Use Cases Cart abandonment advertising is a strong use case for retargeting. This can show up as an ad featuring a product a user browsed online, or an email reminding them they have items in their cart. These users have gone far in the purchase process, and enticing them to complete a transaction is direct revenue for the business. ## Lookalike Targeting ### Description Lookalike targeting finds new potential audiences who share attributes or characteristics with your existing customers. This information comes from your first-party customer data. ### How It Works When you feed current customer data into ad platforms, they use this information to analyze other users and find the ones that match the profile of your customers. This gives you a warm audience to work with, since they match your existing audience through multiple dimensions and common attributes. ### Benefits Because this is a new audience, there is a larger pool of potential customers to target, as opposed to retargeting campaigns, which focus on a current audience. This tactic is relatively easy to get off the ground, using data you already have from your current customers. Some models can even adjust and refine targeting and messaging based on campaign learnings. Lookalike audiences are a higher-funnel tactic, meaning that fresh leads are being added to the marketing funnel for more sustainable funnel growth. ### Considerations Since this targeting is directed at people who have not yet shown engagement or interest in your brand, it likely will take longer to warm them up, and more touchpoints to see significant results. ### Use Cases As lookalike targeting is putting you in front of a net-new audience, it’s a good tactic to expand awareness when you need to fill prospects higher in the marketing funnel. If cost per acquisition is an important metric, lookalike targeting is a solid solution that can increase conversions in a cost-effective way. With a strong database of current customer data, you can create nuanced lookalike campaigns based on product, behavior, time, location, and more to boost the likelihood of positive results. ## How Does Retargeting Compare to Lookalike Targeting? Feature Retargeting Lookalike Targeting Privacy Compliance Relies on first-party data, server-side tracking Relies on both first- and third-party data, consider geographic privacy laws and regulations, needs consent Campaign Goal Conversions, increase intent Awareness, reaching new prospects Setup Complexity Moderate, with more technical and strategic planning and implementation Low, based on quality source audience data Audience Scalability Fixed to current audience as the maximum Unlimited, based on quality of data and budget Immediate Performance High, since this audience is warm and lower in the funnel Moderate, since this audience needs to be created and nurtured Long-Term Brand Lift Shorter timeframe focused on conversions rather than brand lift Longer timeframe, but expands brand influence and reach Cost Efficiency Strong at first, but becomes a diminished return on ad spend from messaging fatigue Good cost per acquisition compared to other tactics, can shift if audience isn’t responding Automated AI Integrations Can display dynamic ads based on personal user behaviors and actions Refine audience and messaging based on response and information learned from existing customers and new leads Brand Safety/Suitability Main consideration is placement next to or within other platforms that don’t align to brand or product Improves with quality of seed audience data A/B Testing Testing creative and messaging and adapting based on response, and adjusting segmenting Test lookalike audience sizes, adjust segmenting, test creative ## How to Decide When to Choose Retargeting Retargeting works well when you have a lot of stalled lower-funnel audience near conversion, or significant monthly traffic (for a larger retargeting audience potential). If you need a quick win, or have abandoned products you’re looking to move, retargeting could be a great tactic to drive completion. ## How to Decide When to Choose Lookalike Targeting If you have more time and need to bring more people into your marketing and sales pipeline, lookalike is a smart tactic to expand reach and awareness. Devoting time to testing audiences, segments, and creative will result in more success. ## Key Takeaways Both retargeting and lookalike targeting have a place in the marketing and sales plan. With retargeting focusing on moving your current audience closer to conversion, and lookalike targeting focused on bringing awareness to new audiences, a mix of both addresses short- and long-term needs. Budget and current audience data will both factor into results for each campaign. ## Frequently Asked Questions (FAQs) ### What is the primary difference in audience intent between retargeting and lookalike targeting? Regargeting targets users who have previously shown intent to purchase. This was demonstrated through actions including product views, content engagement, and cart abandonment. Because they have already interacted with your brand, they are more likely to convert on first impression. Lookalike targeting is aimed at users who have never interacted with your brand, but share characteristics with your existing customers. While they haven’t shown intent, their behavior patterns suggest they are more likely to be interested than other audiences. Still, because they have not yet interacted with your brand, they may need further nurturing before a conversion. ### How does lookalike targeting help scale my performance campaigns beyond retargeting? Retargeting has a limited reach based on your existing audience and, once that audience is saturated, offers diminished return on retargeting spend. Lookalike targeting, however, allows for scaling through different methods such as increasing ad spend, or adjusting or expanding targeting attributes. With lookalike targeting, new prospects enter the marketing funnel, and can later become retargeting audiences. ### If my site has significant traffic but a high cart abandonment rate, should I invest my remaining budget in lookalike targeting to find new prospects, or in retargeting to bring back those who left? Lookalike targeting can be a great way to grow upper-funnel awareness for the future, while retargeting can provide quick wins on those audiences already expressing an interest. Each has a place in the marketing and sales funnel. You should also consider a thorough appraisal of your checkout page to see what’s causing the high cart abandonment rate. --- ### Mail Domain Targeting vs. CRM Lookalike Audiences URL: https://www.taboola.com/marketing-hub/mail-domain-targeting-vs-crm-lookalike-audiences/ Last Modified: 2026-05-24 09:26:53 Both mail domain targeting and CRM lookalike audiences can be useful for performance marketers looking to reach their campaign goals with increased numbers of users. Mail domain targeting is an email marketing strategy designed to create multiple domains for outbound email campaigns, while lookalike audiences are a feature within CRM systems and advertising platforms that use AI functionality to create new segments, based on existing customer data. Here’s more on each of these options, how they work, and when to use them. ## Mail Domain Targeting Mail domain targeting can be useful for marketers and advertisers expanding their company’s reach. Let’s get into the details. ### What Is Mail Domain Targeting? Mail domain targeting is a way to use email marketing techniques for cold outreach to prospects. It’s useful for businesses facing spam risks and that want to experiment with new audiences or increase email volume quickly. Mail domain targeting creates alternative but still on-brand domains, such as store.company.com versus company.com to send outbound emails. This is ideal if a business wants to protect the primary domain from spam filters when scaling up email volumes, or experiment with new formats or audiences in their email marketing campaigns. It protects the core brand while allowing for new approaches to email. ### How Mail Domain Targeting Works There are several ways to approach mail domain targeting, depending on your existing tools and the associated goals. - Domain segmentation: With this method, you create a segmented list within an email provider based on industry domain, such as .gov or .edu, or a specific company’s domain. - Geographic segmentation: This method segments users based on the country domain, like .uk or .in, to then target emails according to regional preferences. - Behavioral or intent targeting: Incorporate data from user interactions, like past purchases or web visits, to target the specific domains or high-intent users likely to convert. - Sub-domain segmentation: This refers to using subdomains within a company’s primary root email domain to send targeted emails to specific audiences. This can help improve sender reputation and tailor messaging to keep it from being too broad. ### Benefits Mail domain targeting can be a very useful way to boost existing and new or desired audiences through email, particularly for B2B marketers. You may, e.g., want to explore whether those working in higher education are a good target audience for your company’s product, and thus create a targeted email domain for .edu recipients. Mail domain targeting also allows marketers to create focused, relevant messaging for different audiences to increase reach and conversions. Additionally, this method is useful when you’re scaling up an email marketing program or making other changes in strategy, and want to ensure the protection of your domain name and associated brand identity. ### Considerations Keep in mind that mail domain targeting requires the same attention to regulatory compliance as any other email marketing program. In addition, mail domain targeting can be onerous to set up and carry out manually if your marketing technology platforms don’t include the feature. It’s also limited to outbound emails, so you’ll likely want to use it as part of a broader performance marketing strategy that includes open web, social and search, and other motions. For larger companies or those with multiple, unique audiences, mail domain targeting helps differentiate messaging and offers. Consider your in-house and platform capabilities, the variety of users and audiences, and number and breadth of products before setting up mail domain targeting. ### Use Cases Generally, mail domain targeting is useful for tailored B2B outreach, improved email deliverability, and segmentation by characteristic, like behavior or geography. Here are a few common use cases: - ABM: For account-based marketing (ABM) strategies, you might target specific high-value corporate domains in order to find and personalize outreach to the relevant decision-makers. - Competitor conquesting: This option identifies and targets users who have interacted with competitors’ domains. - Improving cold email deliverability: Rotating email domains can help avoid spam filters and ensure the email reaches inboxes. - Specialized or targeted offers: Mail domain targeting allows you to customize and personalize content based on industry, company size, or location, then send appropriate offers. - Sub-domain email strategy: Creating separate domains for separate email types, such as marketing versus transactional, can help manage the sender’s reputation and assure delivery. - Domain activity: This is useful for segmenting by domain-specific activity, e.g., if an email provider’s users are becoming inactive, or if a certain amount of time has passed since users have engaged with the business. - Engagement segmentation: For lead nurturing, you may create domains to target users who have engaged with content in order to follow up with tailored offers. - Retargeting: This type of domain targeting is useful to re-engage users who have abandoned their cart, so you can send personalized reminders and offers. ## CRM Lookalike Audiences Using your CRM system or advertising platform to create lookalike audiences can help expand your campaign reach, by finding new audiences modeled on proven customers, or by finding similar users to those you already have. This lookalike audience method also helps capture incremental reach and improve relevance as you’re driving toward campaign success. ### What Are CRM Lookalike Audiences? A lookalike audience is essentially a customer database made up of a group of people who have similar characteristics to your ideal customer, or one of your ideal customer types, depending on the breadth and variety of your business. Modern ad platforms or CRM systems incorporate AI to analyze audience data and create the lookalike audiences, saving users a lot of time. Lookalike audiences might include users with similarities like purchasing behaviors, preferences, and characteristics like location and other demographic details. Then, marketers can direct tailored, personalized messaging and offers to each audience accordingly. Lookalike audiences can start with a small percentage, reflecting how closely you want the new lookalike audience to match the source audience. ### Types of Lookalike Audiences There are a few types of lookalike audiences to consider. - Conversion-based lookalikes: These audiences are created based on actions that users have taken — filling out a form, visiting a campaign landing page, or subscribing to a newsletter. - High-value modeling: These audiences use modeling to group and target users that match high-lifetime-value customer profiles. - Segmented lists: Similar to mail domain targeting, this type of audience is based on specific customer segments, such as those in a particular country, demographic, or in a particular product category. - Seed optimization: This refers to continual refining of the source (or “seed”) customer data list, which is essential for AI tools to accurately predict new lookalike audiences. - Combined or stacked lookalikes: These audiences are combined out of several different lookalike audiences, such as people who have attended an event combined with people who work at financial services firms. This helps expand reach while maintaining relevance. - Engagement-based lookalikes: These audiences incorporate users who are similar to those who have interacted deeply with a brand, whether that’s filling out a form, visiting the website, or watching videos. ### How CRM Lookalike Audiences Work CRM lookalike audiences create new groups to target based on existing users. These audiences can be incredibly helpful in expanding reach and meeting campaign goals by driving new, high-intent visitors to your site or other destination. There are multiple ways to create and use lookalike audiences, including site visitor information through tracking pixels on your website, like those that Realize offers. Without pixels, you can use information like email list signup or completed purchase. Or, upload data directly from your CRM email lists and then create a lookalike audience. Once you’ve created a CRM lookalike audience, you can apply it to a new campaign, or to an existing campaign that aligns with this new audience group. Make sure you optimize the ad creative to match, then monitor metrics like conversion numbers or return on ad spend (ROAS) and refine accordingly. Or, let your ad platform automate this work for you! ### Benefits CRM lookalike audiences can be an easy way to find and target new, high-intent audiences for your brand. It can be a low-lift activity, particularly if you have a platform to do it for you, and newer AI capabilities can delve into data more quickly and at a broader scale than human teams. ### Considerations When you’re using lookalike audiences, make sure to keep these tips in mind: Data is paramount: Marketing outreach with lookalike audiences depends heavily on up-to-date, continually refreshed contact data for customers and prospects. With AI capabilities, this is even more essential so that predictions are as accurate as possible. Use high-intent lists: Lookalike audiences should include users who are most similar to those users who have recently converted, by whatever standard your business uses. You can also add users who have engaged with your business on outside media channels. Get to know percentage modeling: Lookalike audiences work best when you start small and grow. So, try a 1-10% number first so that you get highly similar matches to those valuable prospects or customers. Then, expand the size and reach from there. ### Use Cases CRM lookalike audiences leverage first-party data to target new prospects who mirror high-value customers, driving higher conversion rates and reducing ad spend. Key use cases include targeting lookalikes of recent converters for acquisition, reaching high-lifetime-value (LTV) customers to boost revenue, and finding new users for specific products, like sustainable goods. These are some common use cases for lookalike audiences: - User acquisition at lower cost and higher quality: Lookalike audiences are similar to your existing high-value users and thus more likely to convert and engage more quickly. This costs less than broader acquisition efforts. Plus, you can more easily suppress lower-value users. - Moving beyond saturation: If your targets are saturated, lookalike audiences can help find new leads that are similar to your current customer or prospect base. - Increased intent or awareness: Using existing marketing data can be super useful in building more targeted, specific new audiences, whether you’re working toward higher intent leads, cross-selling or selling new products, or increasing brand awareness. ## How Does Mail Domain Targeting Compare to CRM Lookalike Audiences? Here’s a quick overview of what to expect from mail domain targeting and lookalike audiences when you’re deciding which to choose. Mail Domain Targeting CRM Lookalike Audiences Privacy Compliance Similar considerations to any email campaign: adhere to GDPR, CCPA, CAN-SPAM, and use encryption Use first-party data; ensure explicit, informed consent for marketing use Campaign Goals Match domain targeting with campaign goals, such as increased ROI or conversion uplift; avoid generic emails Tailor audience to campaign goal: narrower for conversion or sales, and broader for brand awareness Setup Complexity Technical complexity: 3-6 weeks of domain warmup, DNS authentication, and reputation management Straightforward within CRM or ad platform Audience Scalability Requires some back-end work to move to a horizontal scaling approach Starting with 1% matching then expanding to 5-10% matching to broaden reach Immediate Performance No, requires 3-6 week gradual warmup Yes, since new audience lists can be activated immediately and with more precision Long-Term Brand Lift Yes, mail domain targeting fosters personalized content and marketers can balance frequency and reach   Yes, since high-value, first-party data continuously feeds into the ad platform to refine marketing efforts Cost Efficiency Yes, since it uses existing tools, data, and platforms and avoids overly broad, wasted targeting Yes, as it targets prospects for high-value outcomes and uses first-party data and systems Automated AI Integrations Possible, with good data hygiene and modern tech Yes, AI integrations are key to build successful lookalike audiences Brand Safety/Suitability Yes, since targeted mail domains protect the brand’s reputation Yes, as they use first-party, consent-driven data A/B Testing Yes, with two email versions across segmented email domains; try large sample sizes and time it right Yes, by adjusting audience percentages, different source data, location, and other variables ## How to Decide When to Choose Mail Domain Targeting Mail domain targeting is ideal when a marketing team wants to expand its email usage, whether for experimenting with new audiences or increasing volume. It works best for cold outreach or scaling sales efforts by email, while keeping your business’ outbound email reputation intact. It’s useful if the business is facing any spam issues, and to ensure deliverability. ## How to Decide When to Choose CRM Lookalike Audiences Choose CRM lookalike audiences when you’re working to acquire new users who are likely to convert. This is ideal when you already have a strong database of existing customers, and when your CRM or ads platform includes this capability. Lookalikes are ideal for targeted advertising, and for ambitious acquisition goals. ## Key Takeaways Mail domain targeting and CRM lookalike audiences are both important tools for modern marketers as they’re growing audiences and expanding reach. Mail domain targeting can help companies experiment and expand their email marketing programs intelligently. Lookalike audiences can take advantage of AI’s scale and depth to analyze data, and make predictions to help performance marketers and advertisers find users who are a good match for what the business is offering. For scale, improved ROAS, and better reach, these tools can open up new opportunities. ## Frequently Asked Questions (FAQs) ### Which targeting method should I use to reach people who are already using my competitor? If you know a group of people are already using your competitor, this is a great chance to use competitor conquesting. Try creating a small lookalike audience based on the information you’ve captured in a CRM platform. This can target those people who have used a competitor app or visited competitors’ websites, then tailor messaging, creative, and offers that highlight the value of your product or service. You can try a 1-3% lookalike audience to target the top percentage of people who match this competitor profile, then expand if needed later. ### I have a list of my top 500 VIPs; should I use mail domain targeting or CRM lookalike audiences to use it? Use CRM lookalike audiences to target this list of top 500 VIPs. This is high-intent, first-party data that represents your best customers, so you’ll be able to use the lookalike audience to find people who share these VIP characteristics. Try a small initial seed list to start finding new valuable users. AI and ML tools do better with groups of more than 1,000 users, but 500 is adequate for their capabilities to help you scale. Using mail domain targeting for this use case may be too broad, with lower-value users than your VIP list. ### Should I use mail domain targeting or CRM lookalike audiences to target people subscribed to specific industry publications? For this use case, try mail domain targeting for outreach to subscribers of specific industry publications. For B2B audiences, mail domain targeting can be very useful for reaching professionals in a particular industry, since you can tailor messages carefully. Of course, if you’re looking for people like those subscribers, rather than actual known subscribers, try lookalike audiences. --- ### Manual vs. Automated Campaign Optimization: Which is Best? URL: https://www.taboola.com/marketing-hub/manual-vs-automated-campaign-optimization/ Last Modified: 2026-06-30 08:17:50 More than ever, performance advertisers are turning to the open web, as rising costs and signal loss make traditional search and social channels harder to scale profitably. But, while native environments, publisher networks, and recommendation platforms offer new reach and diversification, they also add complexity. Unlike walled gardens, the open web forces advertisers to navigate fragmented audiences, varied placements, and rapidly shifting performance signals. This raises an important question: Should campaign optimization rely primarily on human judgment, or machine intelligence? The debate isn’t new, but it’s become an urgent one as AI in advertising becomes more sophisticated. This guide explores the trade-offs between manual precision and algorithmic power, and what you need to consider to build open web campaigns that scale efficiently without sacrificing ROAS optimization. ## Precision vs. Power: The Core Trade-Off in Optimization At its core, campaign optimization is about balancing precision and processing power. Manual optimization gives advertisers detailed control and allows for human judgment, whereas automated bidding systems offer speed, scale, and the ability to recognize patterns that humans cannot match. Neither approach is better in every situation, since each solves a different challenge in performance advertising. Manual campaign management works best when you’re prioritizing strategic thinking over large amounts of data. Automated optimization performs best when campaigns generate enough data and fast execution and campaign management efficiency are the priority. The real challenge for modern advertisers is knowing when each approach creates value and when it increases risk. ### The Argument for Manual Control Manual optimization still matters because advertising is not just mathematical. Context, creative quality, and business goals still require human interpretation. When you first launch a campaign, you often don’t have much historical data available. Without enough signals, algorithms struggle to learn effectively. In these low-data situations, manual bidding and placement decisions help protect return on ad spend (ROAS). Experienced marketers can often recognize signs of high-quality traffic long before conversion data becomes statistically reliable. Manual control is especially useful for niche audiences. Open web campaigns frequently reach specialized users across smaller publishers where conversion volume may be low, but highly valuable. Human operators can notice qualitative signals, such as editorial alignment, audience intent, or brand safety, that algorithms may initially miss. Manual optimization also helps during periods of change. Product launches, seasonal promotions, or creative testing phases often require careful pacing instead of aggressive campaign scaling. Humans can prioritize learning and long-term strategy rather than reacting only to short-term performance signals. In simple terms, manual optimization works like surgical precision, protecting campaigns when uncertainty is high and data is limited. ### The Case for Algorithmic Speed Automation becomes powerful once campaigns generate enough data. Machine learning systems can analyze large numbers of real-time variables at once, including device type, location patterns, time-of-day performance, publisher context, engagement signals, and historical conversion data. No human can adjust thousands of bids across placements in real time. Automated systems can. This advantage comes from response speed. On the open web, performance conditions change constantly. Traffic quality shifts, inventory changes, and user behavior evolves throughout the day. Algorithms respond instantly, moving budgets toward new opportunities before manual managers can even spot the trend. Automation also improves efficiency. As campaigns grow, manual management becomes harder to sustain. Reviewing reports, adjusting bids, and reallocating budgets across many publishers creates operational bottlenecks. Automated systems remove these limits, allowing campaigns to scale without increasing workload at the same rate. The result is not just faster optimization, but continuous optimization. Machines work without fatigue, making thousands of small adjustments that add up to meaningful performance gains over time. ## When to Choose Manual: The Performance Advertiser’s Checklist Even with advances in AI, there are clear situations where manual optimization is the smarter choice. New campaign launches are a key example. Early campaigns lack conversion history, which makes algorithms vulnerable to incorrect assumptions. Manual oversight ensures early traffic sources align with campaign goals before automation scales inefficient signals. Manual optimization is also important when conversion data is limited or delayed. Some campaigns rely on longer attribution windows or offline conversions, which reduces the real-time feedback algorithms depend on. Human judgment helps fill this gap. Highly niche audiences are another case where manual management excels. When targeting specialized or high-value segments, advertisers may prioritize quality over volume. Automated systems focused solely on maximizing conversions may chase scale over relevance unless carefully guided. Manual control also improves transparency. Many automated platforms operate as “black boxes,” making it unclear why budgets shift or placements change. Manual optimization allows advertisers to clearly understand performance drivers and maintain confidence in campaign direction and direct response marketing. In short, manual optimization works best when strategy matters more than speed. ## Leveraging Automation to Scale Beyond Search and Social As campaigns mature, automation becomes essential for unlocking the full potential of open web advertising. Unlike search advertising, where user intent is clear, open web environments rely on probability models. Algorithms analyze behavioral patterns across large datasets to identify users who are likely to convert, even when intent isn’t obvious. This allows automated systems to discover high-performing placements and audience segments that would be extremely difficult to find manually. Data volume plays a critical role, too. The more conversion signals a campaign produces, the more accurately machine learning models can predict results. With enough data, optimization shifts from reacting to performance toward predicting future outcomes. Automation also removes repetitive operational tasks. Instead of manually adjusting bids across hundreds of placements, advertisers can focus on higher-level strategy, such as improving creative messaging, testing offers, and strengthening conversion funnels. This shift changes the performance marketer’s role. Rather than acting as a tactical operator, the advertiser becomes a strategic leader guiding automated systems toward business goals. Automation does not remove human involvement; it redirects it toward higher-impact decisions. ## The Hybrid Framework: A Strategic Path to Maximum ROAS Today, the most effective optimization strategy is not manual or automated alone, but a hybrid. A hybrid framework combines human strategy with machine execution. Advertisers set campaign goals, creative direction, audience parameters, and performance limits manually, while automated bidding handles real-time optimization within those boundaries. This “human-in-the-loop” approach solves many limitations of using either method alone. Humans provide context that algorithms lack, including brand positioning, messaging nuance, and long-term business objectives. Machines provide computational power through rapid bid adjustments, large-scale testing, and continuous learning. The hybrid model also helps address the Black Box Problem. Advertisers maintain visibility into strategic decisions while automation manages complex execution. This preserves control while benefiting from efficiency gains. This approach is especially valuable on the open web, where inventory, formats, and audiences vary widely. Hybrid optimization allows strategic direction from humans while machine learning in advertising adapts dynamically to performance changes. In practice, optimization often follows a progression: manual control at launch, assisted automation during growth, and full algorithmic scaling once sufficient conversion data is available. The result is stronger ROAS and more sustainable campaign growth. ## Key Takeaways The debate between manual and automated optimization is often framed as a competition, but modern performance advertising does not work that way. Optimization exists on a spectrum, not as a simple choice. Manual optimization provides the strategic foundation. It helps campaigns launch effectively, protects performance when data is limited, and incorporates human insight into creative and audience decisions. Automated optimization provides execution power, enabling rapid responses, large-scale testing, and operational efficiency that manual processes cannot match. For open web campaigns, success comes from combining both approaches. Human expertise sets direction, while machine learning accelerates execution. Advertisers who balance these strengths gain both control and scale, and achieve the ultimate goal of performance marketing: predictable, scalable growth. ## Frequently Asked Questions (FAQs) ### How much data do I need before switching from manual to automated optimization? Most automated bidding systems require approximately 30–50 conversions a month per campaign to learn effectively. Below this threshold, algorithmic optimization may struggle due to insufficient data density, making manual optimization more reliable. Tools that provide a “manual-assist” approach can help advertisers transition gradually by feeding algorithms higher-quality initial signals. ### Will automation overspend my budget to find conversions? During the learning phase, automated systems often explore aggressively to gather data, which can temporarily increase cost per acquisition (CPA). However, implementing performance benchmarks, budget caps, and conversion tracking safeguards helps prevent inefficient spending. When properly configured, automation shifts from exploration to efficiency once enough performance data accumulates. ### Is manual optimization still relevant in the age of AI? Absolutely. Manual optimization remains essential for creative strategy, interpreting brand nuance, and making strategic pivots that algorithms cannot contextualize. The strongest results come from human-in-the-loop AI, where automation handles mathematical optimization while humans guide strategic direction and long-term growth decisions. --- ### Best Targeting Methods for Performance Campaigns in 2026 URL: https://www.taboola.com/marketing-hub/best-targeting-methods/ Last Modified: 2026-03-19 12:43:17 Performance advertising in 2026 is a balancing act. You simultaneously need more scale and more efficiency, but signal loss, consent, and platform-level privacy changes keep tightening the screws. Even as the third-party cookie situation continues to evolve, and Chrome’s plans continue to shift based on Privacy Sandbox changes and regulatory scrutiny, the direction is consistent: enduring performance comes from targeting methods that don’t depend on brittle identifiers. Using smart ways to activate first-party data when you have it is crucial. ## What Targeting Strategies Can Performance Advertisers Leverage in 2026? Here’s a practical map of the major targeting methods performance advertisers use today, organized by the data they rely on, what they’re best at, and where they fit in the funnel. Targeting Method Data Source Primary Goal Funnel Stage Contextual Real-time page content Privacy-safe relevance Awareness and consideration Topic AI-classified themes Trending interest capture Interest and awareness Search Keyword High-intent queries Precision conversion capture Consideration and action Broad Performance AI signals Maximum reach and scale Awareness and discovery Behavioral Taboola first-party data Relevance via interest signals Mid to bottom Predictive Modeled future intent Anticipate high-value actions Mid to bottom Lookalike CRM/seed data High-quality prospecting Awareness and growth Retargeting Brand engagement data Conversion recovery Bottom (action) Mail Domain Domain-level engagement Target competitor audiences Consideration and action Below, I’ll dive deeper into how each targeting strategy works, the benefits and considerations for each, and explore some use cases. ### Contextual Targeting Contextual targeting places ads based on what a user is reading right now, not who the user is. #### How It Works Platforms analyze real-time page content (keywords, entities, sentiment, metadata, semantic meaning) and match ads to pages that align with your product and message. #### Benefits/Considerations - Privacy-resilient, since it doesn’t require user identifiers or third-party cookies. - Mindset alignment means you’re buying attention when the topic is already top-of-mind. - You’ll want strong creative-message fit; broad creative can underperform if it doesn’t “belong” in the context. #### Use Cases - New product launches where you want relevant reach fast. - Regulated categories where privacy-safe relevance matters. - Always-on prospecting paired with conversion-optimized bidding. ### Topic Targeting Topic targeting is contextual’s more structured cousin. Instead of matching to individual page content, you target AI-labeled themes like “personal finance,” “fitness,” and “home improvement.” #### How It Works AI classifies pages into topic clusters. You choose the themes most aligned with your audience and test into adjacent topics to expand your reach. #### Benefits/Considerations - Great for scaling awareness while staying relevant. - Easier to operationalize than granular contextual lists. - Topics can be broad; you may need exclusions or creative variants to avoid wasted spend. #### Use Cases - Seasonal pushes like tax season or holiday gifting. - Category conquesting (your brand vs. category leaders). - Upper-funnel feed into retargeting/predictive pools. ### Search Keyword Targeting Search keyword targeting captures users when they show their intent to make a purchase. For instance, a search for “best running shoes for flat feet” is a strong signal that the user intends to buy a product in that category. #### How It Works You bid on keywords or query themes, then match ad copy and landing pages tightly to the intent behind the query. #### Benefits/Considerations - High intent equals high efficiency when aligned with your landing page and offer. - Strong bottom-funnel lever for direct response. - Cost-per-click (CPC) competition can be intense; incremental scale may be limited in mature categories. #### Use Cases - Lead gen with clear “problem → solution” funnels. - E-commerce with high-intent product terms. - Promotions where urgency matters. ### Broad Targeting Broad targeting is “letting the algorithm work.” You minimize constraints to maximize reach and let performance signals determine who sees what. #### How It Works You open targeting, then rely on platform optimization to find pockets of efficiency. Your key metrics are conversion signals, engagement patterns, pacing, and creative performance. #### Benefits/Considerations - Broad targeting is usually the fastest path to scale when you have strong conversion tracking. - Great for discovering new audiences you wouldn’t hand-pick. - Broad needs guardrails — clear conversion events, clean tracking, strong creative rotation, and budget discipline. #### Use Cases - Scaling proven offers beyond saturated segments. - Geographic expansion where you lack audience knowledge. - Testing new creatives quickly across diverse inventory. ### Behavioral Targeting Behavioral targeting uses first-party interest signals to reach people likely to care based on what they do, not who they are. #### How It Works Platforms build interest segments from on-platform and publisher network engagement, then you target those segments to improve relevance. #### Benefits/Considerations - Mid-to-lower funnel relevance without needing third-party tracking. - Often stronger than broad when your category has clear interest patterns. - Segment definitions vary by platform; validate with holdouts and incremental tests. #### Use Cases - Subscription offers like news, streaming, and apps. - Considered purchases in categories like finance, education, and home services. - Always-on acquisition with stable cost-per-acquisition (CPA) goals. ### Predictive Targeting Predictive targeting models which users are likely to take a future high-value action based on conversion patterns. #### How It Works You seed the model with conversion events (pixel or server-to-server). The platform’s performance AI finds new users whose behaviors mirror converters, often in a cookie-resistant way. #### Benefits/Considerations - Beta results cited by Realize show up to 23% conversion (CVR) lift and about 13% CPA improvement in some testing contexts, though results vary. - Finds incremental users who haven’t visited your site yet, making it great for prospecting. - Predictive models are only as good as your conversion signals — tracking quality and event choice matter. #### Use Cases - Scaling lead gen without blowing up CPA. - Moving beyond retargeting ceilings. - Growing lifetime value (LTV)-positive customer cohorts. ### Lookalike Targeting Lookalike targeting expands beyond your known customers by finding new users who resemble your best audience. #### How It works You upload CRM/seed data, like a hashed email, device ID, or ZIP code, depending on the platform. The system builds a modeled audience that matches those traits at scale. #### Benefits/Considerations - High-quality prospecting method when your seed list is clean and value-weighted. - Pairs well with creative personalization and the ability to align your message with demos like “my best customers.” - Too much information can create noise instead of signal. Be sure to segment your seed rather than dumping in everything you have. #### Use Cases - Pipeline growth for B2B and services. - DTC customer acquisition beyond interest targeting. - Geographic expansion using proven customer profiles. ### Retargeting Retargeting re-engages people who already interacted with your brand, but didn’t convert. #### How It Works You build audiences from site visits, product views, cart actions, or engagement (depending on the platform), then serve tailored ads to bring them back. #### Benefits/Considerations - Retargeting is a high return on investment (ROI) “conversion recovery” lever. - Great for sequential messaging to create urgency and draw customers into your story with testimonials and offers. - In signal-loss environments, retargeting pools can shrink, so pair it with predictive and contextual/topical prospecting. #### Use Cases - Cart abandonment and browse abandonment. - Lead form-starters who didn’t submit. - Post-click nurtures for longer sales cycles. ### Mail Domain Targeting Mail domain targeting reaches users based on domain-level engagement. It’s often used to draw attention away from competitors or target specific ecosystems (e.g., enterprise domains). #### How It Works You target audiences associated with engagement patterns around specific domains. #### Benefits/Considerations - Get in front of users already researching alternatives. - Strong mid-to-lower funnel intent signal when domains map to active consideration. - Use carefully to avoid overly narrow reach; pair with compelling “switcher” messaging and proof points. #### Use Cases - Competitor conquest campaigns. - “Why us vs. them” comparison creative. - High-intent acquisition in crowded categories. ## Key Takeaways In 2026, the best targeting stacks combine privacy-resilient reach with performance scaling (broad/predictive) and efficiency levers, instead of relying on a single “magic” audience. Remember that predictive and lookalike targeting work best when your seed signals are clean and tracking meaningful conversion events with segmented CRM lists. Also consider that the ongoing “cookieless” shift isn’t just about cookies, it’s about durability: Consent, regulation, and platform changes keep evolving, so prioritize methods that stay strong even when identifiers weaken. ## Frequently Asked Questions (FAQs) ### How can I scale my prospecting campaigns without sacrificing my CPA goals? To scale prospecting without sacrificing CPA, you usually want to broaden your reach gradually, while keeping your optimization anchored to a high-quality conversion event. Pair that with modeled audiences from first-party CRM seeds to find incremental users who still resemble converters, then scale your budget in controlled steps. ### What is the best strategy for staying competitive in a “cookieless” environment? Fortunately — or unfortunately! — for advertisers, third-party cookies are not close to being fully retired in Chrome. Google changed course over the last two years, moving away from a forced deprecation toward user-choice controls after years of delays, mixed stakeholder feedback, competition, and regulatory scrutiny around Privacy Sandbox. But, advertisers aren’t waiting for Google to resolve the issue. Across the industry, the “cookieless” playbook has shifted from waiting on a single Chrome deadline to building privacy-resilient stacks with first-party data and consented identity where available, heavier contextual targeting, and modeled measurement to keep attribution and incrementality credible as signals fragment. For performance marketing on the open web, contextual and topic targeting tools don’t rely on user identifiers; instead, they use AI to analyze page content and semantic themes to place ads where your audience is already focused. Because they target the current mindset of the user rather than their past browsing history, these tools remain effective and are naturally aligned with privacy-forward standards as Chrome and the broader ecosystem continue to evolve. ### How do I effectively activate my CRM data for high-quality lead generation? In practice, activating CRM data is often harder than it sounds because the data isn’t “activation-ready”: emails may be outdated, duplicated, or missing consent flags, and your lead records may be poorly segmented, which leads to a situation where offline conversions don’t map cleanly to the platform events you’re optimizing toward. The fix is to integrate your CRM with lookalike targeting (and, where it fits your strategy, mail domain targeting for conquesting). By activating first-party data, you can build modeled segments that find new prospects similar to your best leads and then use mail/domain signals to engage users already interacting with competitor ecosystems. Start with segmented seed lists (qualified leads, highest LTV customers) to keep efficiency high from day one. --- ### Broad vs. Lookalike Targeting: When You Should Use Each URL: https://www.taboola.com/marketing-hub/broad-vs-lookalike-targeting/ Last Modified: 2026-05-31 11:47:52 As AI capabilities and quality of first-party data become more refined, targeting tactics are increasingly able to produce worthwhile results. With the rise of signal marketing and optimizing on the fly to refine outcomes, advertisers have choices when it comes to strategies, and two that rise to the top are broad targeting and lookalike targeting. Whether looking to scale awareness or grow successful segments, each targeting tactic has nuanced considerations, but each could improve results for efficient campaigns. Here’s what to know about both approaches. ## Broad Targeting ### What Is Broad Targeting? Broad targeting is an audience targeting tactic where the algorithm figures out the best audience for an ad. With minimal constraints, the system tests, iterates, and learns what works based on engagement and, ultimately, conversions. Simply put, you provide the goal for which to optimize, and the platform finds the audience. ### How It Works Broad targeting works by bringing together clear campaign objectives and high-quality creative to test from a large pool of users. It refines outcomes and keeps testing different creative on different audiences to see which approach produces optimal results. ### Benefits One benefit of broad targeting is that the largest pool of prospective audiences is tapped, maximizing ad dollars early and lowering CPC and CPM, while the algorithm learns and adjusts. Another bonus is that minimal human intervention is needed for this approach once a campaign launches, since parameters are set and a large pool of creative loaded early on. Since little identifiable information is put into the platform to begin with, broad marketing relies less on specific identifiers and demographic information, making it a more attractive option for privacy compliance. ### Considerations While broad targeting is generally an efficient campaign tactic, it takes some learning to refine the approach. This can result in inefficient spend and conversions early in the campaign, until the algorithm learns what works. In this phase, stepping back and waiting for results is important: Starting with solid creative to test and refine, then stepping back and letting the tool refine its audience, is key. ### Use Cases Broad targeting is a good tactic when you’re promoting mass-market products, such as e-commerce, or in a learning or testing phase (for example, looking to identify new, high-converting audiences, or ahead of a larger campaign push). ## Lookalike Targeting ### What is Lookalike Targeting? Lookalike targeting is the practice of building lists of potential customers similar to your existing ones, based on shared characteristics and behaviors. The thinking here is that marketing efforts will reach new users likely to convert, since they mimic your current customers. ### How It Works Establish lookalike targeting by defining an audience from first-party data (for example, customer list, CRM segment, or users/visitors), which serves as a seed list. The lookalike targeting tool will then compare this grouping against potential users likely to convert. Once identified, campaigns are run against the modeled audiences. ### Benefits Since lookalike audiences are based on people who have already converted, there is generally less guess work in figuring out the correct approach. Having been built based on an engaged audience, it often delivers better conversion rates and lower cost per acquisition than other advertising campaigns. With lookalike targeting using first-party data, it’s also a safe option as privacy concerns and regulations increase. ### Considerations Outcomes are dependent on seed information, so if that list is outdated, incomplete, or based on a small sample size, there might not be enough quality data to create a high-converting lookalike audience. With lookalike targeting using first-party data, there can be geographic and platform limitations, too, and should privacy laws change, lookalike targeting may need to be adjusted. ### Use Cases If you want to bring on new customers through a cost-effective method, lookalike targeting can be a smart way to expand reach, since it’s designed to mimic people who have already converted. ## How Does Broad Targeting Compare to Lookalike Targeting? Feature Broad Targeting Lookalike Targeting Privacy Compliance Very high privacy compliance since this tactic relies on contextual data and behavioral signals High, especially when started with compliant and consented first-party seed data Campaign Goal Best suited for discoverability and reach Acquisition- and ROAS-focused Setup Complexity Low complexity since broad targeting learns and adjusts over time More complex since a strong source audience is needed for better results Audience Scalability As scalable as budget and inventory allow Limited to seed and market size Immediate Performance It takes time, as performance improves with learnings More stable sooner, especially with high-quality seed Long-Term Brand Lift Creates additional brand awareness higher in the funnel Limited to people similar to current converting customers Cost Efficiency Generally lower, and more aligned for high-volume conversion Typically higher conversion rates, but needs a stronger seed list Automated AI Integrations Very commonly native to advertising platforms Often integrated alongside advertising platforms Brand Safety/Suitability Relies on platform controls Safer since it mimics a known audience A/B Testing Best suited for testing creative and messaging Best suited for testing audience segments ## How to Decide When to Choose Broad Targeting The goal of broad targeting is to focus on reach and awareness. It’s a wise choice when you have a consumer-facing product or are looking to reach a wide audience, want to lean into creating more awareness, or don’t have historical data or audience learnings. ## How to Decide When to Choose Lookalike Targeting The goal of lookalike targeting is to reach more of a specific type or makeup of audience. With quality data and a leaner budget, you’ll more likely reach people who are closer to converting. ## Key Takeaways Both broad targeting and lookalike targeting are first-party data- and privacy-resilient. The quality of data dictates success with lookalike targeting, while the success of broad targeting is based on high-quality and diverse ad creative from which to learn and adjust. Before beginning, take a look at what your own team can support (creative vs. data quality) and the goals for the campaigns, for realistic expectations and aligned outcomes. ## Frequently Asked Questions (FAQs) ### When launching a campaign with broad appeal (e.g., a mobile game or utility app), is it better to use a lookalike audience to guide the initial spend, or go broad and let the AI find my audience from scratch? For broad-appeal campaigns, it’s recommended to start — and stay — broad, allowing AI to find audiences. These campaigns learn as they go and become more cost-conscious with spend, figuring out what works, and when to lower CPC over time. ### I have no pixel history and a small daily spend; which method minimizes my risk? With no pixel history and small daily spend, broad targeting is a better strategy, as lookalike targeting relies on higher-quality historical data, and could incur a higher cost to reach the intended audience. Broad targeting allows you to be more wise with spend, finding competitive pockets of opportunity. ### My lookalike costs are rising; can I achieve the same ROAS with cheaper inventory? For lookalike targeting alone, costs may increase as competition intensifies for the market size and audience. To account for a better return on ad spend, a blended advertising approach that includes both broad and lookalike targeting could help lower cost, keeping both the middle funnel and upper funnel healthy. --- ### Content Tags: What Are They? Why Are They Important? URL: https://www.taboola.com/marketing-hub/content-tag/ Last Modified: 2026-03-19 07:00:13 What if your website were like a massive, disorganized library? Can you imagine finding the book you wanted without the Dewey decimal system and stickers on the spines? Locating a specific thriller, self-help book, or cookbook would be a nightmare. Those same stickers live in the digital world, too, as content tags. Understanding and maximizing content tagging can mean the difference between creating (and managing) a high-performing digital ecosystem and a cluttered graveyard of forgotten blog posts and child pages. ## What Are Content Tags, and Why Do They Matter? The basic definition: A content tag is a label. It’s a piece of metadata — information about information — that describes what a piece of content is about. But, these tags also tell your content management system (CMS) or performance advertising platform (e.g., Realize) how to categorize, surface, and track your work. Without those tags, your deliverables exist in a vacuum. With them, your content becomes part of a searchable, scalable, and intelligent network. ## What Are the Key Benefits of a Tagging Strategy? It doesn’t take long — maybe 30 seconds — to tag a post, but when you’ve got a thousand things to do, time adds up. Is it worth the bother, then? Absolutely: The return on investment (ROI) on those seconds is massive. Here’s what you gain: ### Improved Discoverability and Searchability Internal search engines rely on tags to serve up relevant results. If a user searches your site for “native advertising,” and you’ve tagged your articles correctly, they find what they need quickly. If not? They bounce. ### Better Content Organization and Governance Tags let you instantly audit your library. Want to see every case study you published in 2025? A “2025” and “case study” tag combo makes that search a one-click task rather than a manual slog. ### Flexible Content Relationships and Cross-Linking Ever see a “related articles” section at the bottom of a blog? That’s tagging in action. By tagging two pieces of content with “lead gen,” for example, you tell your site to suggest them together, a strategy that keeps users on your site for longer. ### Easier Team Collaboration and Workflow Management Tags help teams stay sane. You can tag creative assets by “winter sale” or “video ad,” empowering your design and media-buying teams to find exactly what they need without digging through endless folders. ### Ability to Support Advanced Use Cases The Realize Pixel on Realize is a prime example of high-level tagging. Adding this tag (a snippet of code) to your site lets you track conversions and build remarketing audiences. This data facilitates sophisticated automation you just can't do with naked (tagless) content. ## What Are the Common Challenges and Pitfalls in Content Tagging? The “downside” to tagging? If you don’t have a plan or a strategy, it can get messy quickly. ### Inconsistent Tagging If one person tags a post “social media,” another tags it “social-media,” and a third uses “Instagram,” cohesion falls by the wayside and data becomes fragmented. Consistency is king. ### Over-Tagging (Tag Bloat) or Under-Tagging Applying 50 tags to one post confuses the algorithm (and probably the user, too). Conversely, a dearth of tags makes the content invisible. Your sweet spot? Specific but concise tags. ### Lack of Governance or Taxonomy Related to the inconsistent tagging issue, if everyone has the freedom to create new tags on the fly, your library will balloon to thousands of redundant tags. Create a source-of-truth list and consider limiting who has the authority to add and manage the tags. ### Difficulty Scaling What works for 10 blog posts probably won’t work for 10,000. As your brand grows, manual tagging becomes a bottleneck. Certain platforms offer strategies to address this issue. ## Tagging Implementation: Metadata, Taxonomy, and CMS Features So, how does tagging implementation actually look under the hood? Like a McLaren F1, it’s all about structure, functionality, and performance. ### Internal vs. Public Tags Your readers see the topics, which are the public tags. Internal tags are for your team’s eyes only. Think “draft,” “needs review,” or “persona: marketing manager.” These internal tags keep the pipeline flowing behind the scenes. ### Tag Limits and Governance Rules Smart marketers set rules. For example, “Every post must have one Category tag and no more than five Topic tags.” This standard operating procedure (SOP) prevents tag bloat. ### Use of Taxonomies (Hierarchies) Digital marketing is the parent category, and search engine optimization (SEO) and pay-per-click (PPC) are its children. This structure helps platforms understand the context of your ads. ### Support for Multiple Content Types and Assets A good tagging system handles everything — images, videos, PDFs, and articles. You might, e.g., use third-party tags to bridge the gaps between your ads on Instagram and Google to get a unified view of your campaign’s performance. ## Tagging Strategy and Best Practices for Consistent, Scalable Tag Management - Create a tag dictionary: Document every approved tag and its definition. Anyone who wants to add a new tag must justify it first. - Automate where possible: Use tools like Google Tag Manager to deploy code-based tags without needing a developer. - Audit regularly: Each quarter, clean house by merging duplicate tags and deleting any you’re not using. - Train your team: Confirm that everyone understands the tagging SOP — why they’re tagging, not just how. For example, if a marketer understands that tagging a post “high intent” leads to better retargeting, they’ll be more likely to tag correctly. Pro tip: Use tags to group content, track performance, or power recommended widgets. Don’t use tags as a substitute for a clear URL structure or as a way to stuff SEO keywords (Google catches on to those “strategies” really quickly). ## Key Takeaways Content tags are the DNA of your digital strategy, providing structure and meaning that allow your systems to function. Embrace consistency using a defined taxonomy (your Dewey decimal system) to avoid tag chaos, and leverage platforms like Google Tag Manager to simplify implementing tracking and third-party tags, giving you more power with less coding. Also, start small! Do you need 1,000 tags today? Nope! Begin with the basics and scale as your content library grows. ## Frequently Asked Questions (FAQs) ### How do I know which tags are the “right” ones to use for a piece of content? Generally speaking, your best guide is to think about your users first. If they were looking for this specific bit of information, what would they type in the search bar? Start with broad categories (e.g., strategy) and move to specific topics (e.g., native ad design). If you’re a performance marketer on the open web, the “right” tags are the ones that align with user intent, buyer stage, and measurable outcomes, rather than just topic labels. Start with the core keyword themes driving traffic and conversions, then layer in audience attributes (e.g., industry, role, funnel stage) that map to campaign targeting and retargeting segments. The best tags are specific enough to signal intent to ad platforms, content syndication partners, and recommendation engines, but broad enough to scale distribution. If a tag doesn’t support segmentation, optimization, or reporting, it’s likely not pulling its weight. ### Should tags change over time as content or strategy evolves? Yes! But, proceed with caution. Changing a tag name can break old links (generating the dreaded 404 error) or mess up your historical data. It’s better to map old tags to new ones or use a system that permits bulk editing. In performance advertising, your strategy should evolve as your messaging, positioning, and performance data evolve. Taxonomy is a living system: As new audience segments emerge, products launch, or keyword trends shift, your tagging structure should reflect those changes. Updating tags can improve content resurfacing, retargeting pools, and contextual alignment across programmatic channels. Regular audits, meanwhile, help ensure older content remains discoverable and aligned with current campaign goals. ### Do tags influence personalization or recommendations? Absolutely. Tags power most “recommended for you” engines. If you read three articles tagged “email marketing,” the system assumes you’re interested in that topic and will serve up a fourth. On the open web, tags are foundational signals for personalization engines and content recommendation platforms. They inform contextual targeting, dynamic creative optimization, and behavioral segmentation, helping match the right content to the right user at the right moment. In performance marketing, strong tagging improves recommendation relevance, which can increase engagement rates, lower acquisition costs, and strengthen downstream conversion metrics. Without structured, strategic tags, personalization systems have less data to work with, and performance typically suffers. --- ### Predictive vs. Lookalike Targeting: Which Is Best? URL: https://www.taboola.com/marketing-hub/predictive-vs-lookalike-targeting/ Last Modified: 2026-03-08 16:30:47 For years, performance marketers have often relied on the same tried-and-tested playbook: upload a customer list, build a lookalike audience, scale until costs spike, then rinse and repeat in another campaign. That approach still works well on platforms that have endless audience signals, cheap reach, and predictable attribution, but today’s reality for many platforms looks very different. Costs are rising, privacy regulations limit available data, and many advertisers feel constrained by search and social platforms that no longer deliver the incremental growth they once did on their own. As a result, marketers are re-evaluating not just where they advertise, but how they define and reach the right audiences. Two of the most commonly compared approaches are lookalike and predictive targeting. While they may sound similar on the surface, they rely on fundamentally different inputs and produce varying outcomes depending on your goals, data maturity, and growth stage. ## Predictive Targeting ### Description Predictive targeting shifts the focus away from who users are, and toward what they’re likely to do next. Instead of building audiences based on identity matching, predictive modeling analyzes real-time behavior, contextual signals, and historical patterns to identify users who are actively demonstrating purchase or intent-driven behaviors. This approach is designed to identify high-intent users earlier in the funnel, even when little or no first-party data exists. ### How It Works Using machine learning, predictive targeting is only possible when learning models can be trained on large-scale behavioral data. These models analyze signals like content consumption, engagement patterns, browsing behavior, and on-site interactions across the web. Rather than matching users to a static profile, the system predicts which users are more likely to take a desired action based on these signals and behavioral momentum. Because it operates independently from individual user identities, predictive targeting can adapt in real time as behavior changes. ### Benefits One of the biggest advantages of predictive targeting is speed. Campaigns don’t need weeks of conversion data to stabilize: Instead, models begin optimizing as soon as intent signals emerge, allowing you to see meaningful patterns more quickly. Predictive targeting also scales more efficiently. Because it continuously refreshes its audience pool based on behavior, it avoids the saturation issues common with static lookalike audiences. Predictive targeting pairs well with high-impact creative formats and automated optimization tools that adjust bids, placements, and messaging dynamically. ### Considerations Predictive targeting requires trust in automation and machine learning. Advertisers who prefer manual audience controls may find it less familiar at first. Success also depends on high-quality creative and landing page experiences, as intent signals must be matched with relevant messaging to convert efficiently. ### Use Cases Predictive targeting works at its best for new product launches, especially when no historical conversion data exists. A new direct-to-consumer (DTC) wellness brand introducing a first product, e.g., could use predictive targeting to identify users actively engaging in relevant topic content, such as health routines or product comparisons. This allows campaigns to start generating traction straight away, rather than waiting weeks for data to accumulate. In competitive retail categories like beauty or electronics, predictive targeting helps brands break through the saturated auctions and cement their own place in consumer minds. For instance, a skincare brand facing rising costs per acquisition (CPAs) on social platforms could reach users who are researching by ingredient or for a specific solution to a skincare problem. ## Lookalike Targeting ### Description Lookalike targeting is data-driven audience expansion, building a new prospects pool based on existing customer or conversion data. Advertisers provide a “seed” audience such as purchasers, leads, or high-value prospects, and the platform finds new users who statistically resemble that group. This approach has become a staple of performance marketing because it feels intuitive — if your best customers share certain characteristics, finding more people like them should produce similar results. ### How It Works Lookalike audiences start with a first-party data set. This may include customer emails, website converters, app users, or customer relationship management (CRM) records. The platform analyzes that audience to identify common attributes, behaviors, or signals. Using those insights, it creates a larger audience pool made up of users who closely match the original group. Advertisers can often control the similarity level, choosing between tighter matches with smaller reach, or broader matches that trade precision for scale. ### Benefits The biggest strength of lookalike targeting is familiarity. It’s easy to understand, simple to set up, and integrates seamlessly with existing performance workflows. For advertisers with strong historical data records, it can quickly identify users who behave similarly to proven converters. Lookalikes work well for remarketing-adjacent strategies, loyalty programs, and scaling campaigns that already have a predictable conversion path. ### Considerations Lookalike targeting is only as strong as the data you feed it. If your seed audience is small, outdated, or skewed toward low-value users, performance can quickly suffer. It also relies heavily on user identity and platform-specific signals, which makes it more sensitive to privacy restrictions and signal loss. As campaigns scale, lookalikes can saturate quickly, leading to rising CPAs and creative fatigue without delivering incremental reach increases. ### Use Cases Lookalike targeting is especially effective for established e-commerce brands with a clear understanding of who their best customers are. For instance, a DTC apparel brand with several years of purchase data may create lookalike audiences based on repeat buyers or customers with higher average order values. By doing so, the brand can effectively reach shoppers who share similar purchasing behaviors, demographics, or interests. This is especially helpful in predictable sales periods like seasonal promotions or product restocks. For lead generation businesses in the service field, like insurance or education, lookalike audiences work well when campaigns are anchored on qualified lead data, rather than simply form fills. A regional home services provider, for example, may seed lookalikes from customers who completed a booking consultation, rather than just a quote. ## How Does Predictive Targeting Compare to Lookalike Targeting? Feature Predictive Targeting Lookalike Targeting Privacy Compliance Built around behavioral and intent signals Relies on user identity and first-party data Campaign Goal Capture active intent and drive action Scale known audience patterns Setup Complexity Minimal setup, model-driven Simple with existing data Audience Scalability Continuously refreshed and scalable Can saturate quickly Immediate Performance Faster optimization from launch Slower without stronger data Long Term Brand Lift Broader discovery and awareness impact Limited incremental reach Cost Efficiency More efficient as models adapt CPAs rise as audience saturates Automated AI Integrations End-to-end AI optimization Limited automation Brand Safety/Suitability Contextual and intent-aligned Platform-dependent A/B Testing Built-in testing across formats and signals Requires manual segmentation ## How to Decide When to Choose Predictive Targeting Predictive targeting is the better choice when growth has plateaued, costs are rising, or traditional platforms are no longer delivering increased performance. If you’re launching something new, entering a new competitive market, or trying to reach high-impact users beyond your existing audience pool, predictive models offer a faster, more scalable approach. ## How to Decide When to Choose Lookalike Targeting Lookalike targeting makes sense when you have strong, recent first-party data and want to extend what’s already working. If your customer base is stable, your funnel is predictable, and you’re optimizing within familiar platforms, lookalikes can be a great way to deliver consistent results. This type of targeting is also best when used for efficiency, rather than discovery, or when campaigns are focused on reinforcing proven acquisition paths, rather than finding entirely new ones. ## Key Takeaways Lookalike targeting and predictive targeting aren’t interchangeable tools, but instead built for different stages of growth and different performance challenges. Lookalikes extend what you already know, while predictive targeting uncovers what’s likely to work next. As privacy constraints increase and competition intensifies, advertisers who rely solely on identity-based strategies may struggle to maintain the same level of growth they once did. Incorporating predictive targeting into your overall strategy can help offer a more adaptable path forward. ## Frequently Asked Questions (FAQs) ### I am launching a new e-commerce product with zero conversion history. Which method is better for a “cold” product launch? Predictive targeting is typically the better choice because it doesn’t require historical conversion data and can identify high-intent users based on real-time behavioral signals. ### We want to maximize efficiency for a mature lead-gen campaign in a competitive auction. I have plenty of data but high CPAs; how do I lower them? Predictive targeting can help lower CPAs by continuously refreshing audiences and optimizing toward intent, rather than repeatedly serving ads to saturated lookalike groups. ### How do I find repeat purchasers rather than one-time buyers for scaling a high-LTV subscription service? Both predictive and lookalike targeting can work well, especially with large amounts of historical conversion data. Predictive targeting is often more effective at identifying users who exhibit the desired behaviors around long-term engagement and repeat purchases. --- ### Ad Rejection: Causes, Examples, and Prevention URL: https://www.taboola.com/marketing-hub/ad-rejection/ Last Modified: 2026-06-04 09:35:19 It’s the moment every advertiser dreads. You build your campaign and hit publish, but instead of impressions, you’re met with the dreaded notification that your ad didn’t pass review. Yes, it’s frustrating, but ad rejection isn’t random: It’s the result of increasingly strict platform policies designed to protect users, maintain trust, and ensure safe, high-quality advertising environments. Even the most experienced advertisers can be caught off guard by how nuanced ad reviews have become. Understanding ad rejection, and how each platform’s review system works, is the first step toward preventing costly delays and wasted effort. ## Advertising Platforms and Policy Compliance: What You Need to Know Every major advertising platform operates under its own detailed policy framework. These policies are living documents that evolve in response to regulatory pressure, user behavior, and emerging risks like misinformation or new deceptive advertising practices. At a high level, most platforms review ads across several layers: - Content compliance: Does the ad violate rules on restricted or prohibited topics? - Creative quality: Are images, videos, headlines, and copy clear, accurate, and non-deceptive? - Landing-page integrity: Does the destination page function properly and deliver what the ad promises? - User safety and trust: Could the ad mislead, confuse, or exploit audiences? - Technical compliance: Does the ad meet format, file size, tracking, and disclosure requirements? While automated systems handle initial review at scale, human reviewers often examine flagged ads or sensitive categories. This hybrid approach means your ad has to pass both automated checks and human judgment. ## Common Reasons Ads Get Rejected Ad rejection usually falls into one of three broad categories. Knowing which category applies to your ad can help you resolve the issue faster. ### Content and Messaging Violations Messaging issues are the most visible causes of rejection. They include: - Misleading or exaggerated claims: Ads that promise unrealistic results, guarantees, or outcomes without evidence often fail review. - Restricted or sensitive topics: Health, finance, and political content are heavily regulated across platforms. - Fear-based or manipulative language: Content that pressures users or implies negative outcomes if they don’t act can trigger rejection. - Inconsistent claims: Statements in the ad that don’t match what appears on the landing page are a frequent red flag. Even subtle word choices can push an otherwise compliant ad toward rejection. ### Creative, Media, and Technical Issues Sometimes rejection has nothing to do with what you’re saying and everything to do with how the ad is built. Here are some execution issues that can lead to rejection: - Low-quality images or videos: Blurry visuals, distorted text, and pixelation frequently fail quality checks. - Improper formatting: Incorrect aspect ratios, excessive text overlays, and unsupported file types can trigger automated rejection. - Broken creative elements: Missing thumbnails, silent videos without captions, and unreadable text can all cause issues. - Tracking or loading errors: Ads linked to slow-loading pages or broken tracking parameters may be flagged. These issues can be especially common when you’re scaling campaigns quickly or repurposing creative across platforms. ### Structural, Policy, and Platform-Specific Violations Structural violations often feel less intuitive but are increasingly common. They include: - Ad-to-landing page mismatch: When your offer, language, or visuals don’t align clearly, platforms may reject the ad. - Missing disclosures or legal text: This is especially critical if you’re in a regulated vertical. - Unauthorized third-party content: Promoting offers, testimonials, or brands without permission or attribution can lead to rejection. - Policy circumvention signals: Repeated resubmissions with minimal changes or attempts to “work around” policies can bring penalties. If your ad appears to follow the rules, look for potential structural issues that could be getting in the way of approval. ## Platform-Specific Common Rejection Reasons While advertising policies share broad themes, each platform applies them differently based on audience expectations, risk tolerance, and type of product. Understanding these differences helps advertisers anticipate rejection triggers before submitting creative. ### TikTok TikTok’s short-form videos create a fast-moving environment, and the platform’s ad policies prioritize both community safety and creative authenticity. Ads are evaluated not only for compliance, but also for how transparently they communicate claims to viewers. Content that might be rejected includes: - Sensational or misleading hooks. - Visual claims that aren’t clearly explained in text. - Health, financial, or self-improvement content without sufficient disclaimers. - Ads that mimic organic content too closely without transparency. Short-form video demands clarity and context, not just a strong hook. ### Meta Meta enforces some of the industry’s most detailed and granular advertising policies. Its review systems closely examine whether ads comply with laws, platform standards, and rules designed to protect users. Some common reasons for Meta ad rejections include: - Content that violates applicable laws or regulations in the advertiser’s jurisdiction. - Ads that are discriminatory against a particular group or demographic. - Messaging that uses deceptive or misleading practices to promote products, services, schemes, or offers. - Attempts to obtain money or personal information through misleading or scam-like tactics. - Content that shares or requests sensitive information about people or implies things about a user’s personal traits, circumstances, or identity. Meta’s systems are particularly sensitive to subtle language that implies personal judgment or targeting. ### Google Google focuses heavily on accuracy, transparency, and end-user experience across search, display, and video placements. Ads are assessed not only on content, but also on technical performance and destination quality. Common Google ad rejection triggers include: - Misleading claims or unverifiable promises. - Poor landing page performance, pop-ups, or redirects. - Technical noncompliance, especially for tracking and disclosures. - Circumventing systems or masking content. Google holds search and display ads to high standards for trust and relevance. ### Open-Web Ad Networks Open-web environments emphasize brand safety, editorial alignment, and user trust. Ad platforms operate review systems that assess not just compliance, but also contextual alignment. Some reasons your ad might be rejected include: - Clickbait-style thumbnails or headlines. - Ad creative that exceeds the campaign’s selected safety settings. - Landing pages with hidden issues or technical problems. - Poor-quality images or deceptive formatting. Performance-driven platforms often evaluate ads on quality signals, not just reach or volume. ## Preventing Ad Rejection: Best Practices and Pre-Submission Checklist You can wait until you’re rejected and scramble to fix things, but it’s much easier to ensure your ad makes it through the first time. A structured review process can dramatically reduce delays. Before submitting an ad, confirm the following: - Creative clarity: Images and videos are high-quality and easy to understand. - Landing-page alignment: The destination page clearly delivers on the ad’s promise. - Messaging accuracy: Claims are factual, verifiable, and supported on the landing page. - Technical readiness: Landing pages load quickly, tracking works, and no errors appear. - Disclosure completeness: Required legal or regulatory information is visible and accessible. - Platform fit: The creative matches the norms, tone, and formats of the platform. Many advanced performance platforms offer tools that can help you anticipate rejections before they happen. These tools can flag mismatches, highlight quality risks, and identify policy gaps before the platform begins its review. This proactive approach can reduce days of iteration. ## What to Do When Your Ad Is Rejected: Troubleshooting and Next Steps So, your ad didn’t pass review. It’s natural to feel frustrated, but a rejection doesn’t mean your campaign is dead, it simply means you need to make some adjustments. Start by reviewing the platform’s rejection notice carefully. Identify whether the issue is with your creative, messaging, or landing page. Then: - Make clear, substantive changes, not superficial edits. - Re-check alignment between your ad, targeting, and landing page. - Fix all technical issues before resubmitting. - Document changes in case you need support later. On some platforms, repeated rejections can lead to penalties, so it’s important to review any rejections carefully and make sure you address all possible issues. In managed environments, you’ll typically be limited in how many times you can submit the same ad. This encourages quality-first iteration rather than trial-and-error guessing. This model is designed to protect your ad accounts and boost each campaign’s long-term performance. ## Key Takeaways Ad approval isn’t just about avoiding “bad” content. You’ll need to meet each platform’s standards for trust, clarity, and user experience. Most rejections stem from small oversights: Your creative may not align with your landing page, or the format of your creative simply might not meet platform expectations. By treating compliance as a strategic advantage, you can launch faster, scale more efficiently, and protect the health of your advertising accounts over time. ## Frequently Asked Questions (FAQs) ### Can my ad be rejected even if it seems “safe” and compliant? The short answer is yes — even an ad that seems benign can be rejected. Review teams evaluate more than obvious content issues: Creative quality, landing-page functionality, visual alignment, targeting, disclosures, technical specs, and platform-specific rules can all factor into your ad’s approval. An ad may pass a basic self-check but fail due to technical or structural problems. On performance advertising platforms, subtle misalignments can result in rejection, even when the offer itself is harmless. This can include thumbnails not matching campaign safety levels, discrepancies between ad messaging and landing-page content, poor media quality, and hidden landing-page issues. ### What makes a landing page noncompliant and lead to ad rejection? A landing page can trigger rejection if it’s broken, slow, misleading, inconsistent with the ad, or missing required information. Unauthorized third-party content, exaggerated claims, and unclear disclosures are also common culprits. For performance-focused platforms, additional red flags include unrealistic promises, missing legal disclosures for regulated verticals, undisclosed third-party ads, language mismatches, and content that misrepresents what the ad is offering. ### If I fix the issues, how many times can I resubmit the ad/campaign? Most major platforms allow multiple submissions as long as each version complies with policy and doesn’t attempt to circumvent review systems. However, repeated violations can result in account restrictions or additional verification requirements. On performance-focused platforms, resubmissions are usually capped, and repeated rejections can trigger stricter scrutiny, longer review times, or even account-level risk scoring. It’s best to treat resubmission strategically, resolving all cited issues at once, documenting changes, and ensuring creative and landing page alignment before sending it back for review. --- ### Benchmarking: Your Secret Weapon for Marketing Growth URL: https://www.taboola.com/marketing-hub/benchmarking/ Last Modified: 2026-03-15 13:56:06 To run a successful marketing campaign, you need to understand where you’re starting. As the campaign launches, you’ll want to know how you’re stacking up against competitors and, as time goes on, you’ll want to measure your results against your past performance. It’s only by understanding where you were, and where you’re going, that you can achieve real growth. ## What Is Benchmarking? Benchmarking is the process marketers use to gauge how well they’re doing against competitors, and how well results are improving within their own organizations. To benchmark your organization’s advertising and marketing campaigns, you’ll need to look at the tools and processes of other companies of a similar size within your industry, or industries like yours. ## Importance of Benchmarking Benchmarking is important for understanding how quickly your company is growing and if there are things you can do to improve conversions or growth — otherwise, you’re just tossing spaghetti at a wall to see what sticks. It’s also important to analyze how well you’re doing against competitors. You may want to benchmark your practices against their tools, processes, AI adoption, content, and marketing channels. While marketers may not focus on areas outside their expertise, in general, companies also want to benchmark factors like employee salaries and benefits to stay competitive and attract the best talent. ## Benchmarking Benefits and Challenges Companies who practice benchmarking diligently and effectively will find many advantages. It gives marketers performance-based goals to strive for, whether that’s open rates, clicks, or conversions. Some of the challenges involve a lack of access to real-time, reliable data, especially if you’re looking to benchmark against competitors rather than comparing your own campaigns. Pros Cons Helps your company improve results with data-driven analytics. May be hard to choose or find companies to benchmark against. Creates measurable goals. Difficult to find reliable analytics. Identifies gaps and opportunities. Can be time-consuming. Encourages growth and innovation. What works for one company may not work for yours. Enhances industry awareness. ## Channel-Specific Benchmarking Marketers can adopt benchmarking best practices within various channels or across omni-channel campaigns. Let’s look at some specific benchmarks in various channels. ### Email Marketing Some of the benchmarks to look at in email marketing include open rate and click-through rate (CTR). It’s also worth looking at your unsubscribes, but don’t stress over them, as unsubscribes can help boost your open percentage, which is a good thing. If they don’t like what you’re offering, and they let you know by opting out, they aren’t your ideal customer. In the words of Mel Robbins, “Let them.” Overall, you want to see solid growth in your engaged subscriber list to achieve the most benefits from your campaigns. Email marketing services like Mailchimp provide benchmarks for others in your industry, which is a helpful tool for you to compare your campaigns against others in real time. ### Social Media Social media benchmarking involves looking at the specific platforms you’re on, your numbers of engaged followers, and other KPIs important to your company. These may include likes, views, clicks, comments, and shares. You can also gauge overall market sentiment toward your brand and its content, your voice share in the market compared to competitors, and if you’re engaged in social selling, actual sales that come through social platforms. ### Paid Ads Paid ads include cost-per-click (CPC) and cost per impression (CPM) models. Ads can include display ads, paid search, paid social, boosted posts, and native advertising, where ads blend into the content and platform where they appear. A key analytic in paid ads is your return on investment (ROI). How much are you spending for each new customer or each sale? With performance marketing campaigns, e.g., you only pay for conversions, which boosts your ROI. ### SEO and Content In the ever-changing world of SEO and GEO (Generative Engine Optimization), benchmarks for content optimization are shifting as fast as ChatGPT can hurl clichés and obsequious compliments at users. Some tried-and-true benchmarks for SEO include: - Google rankings. - Google local rankings. - Domain authority. - Linkbacks. - Click-through rate. - Featured snippets. As AI has become a major player in search, you also want to appear in AI summaries and in recommendations from generative AI tools like Claude and ChatGPT. The real magic, though, still happens when users get to your website. This task has grown harder to accomplish, since people are finding all they need in AI summaries without clicking through to the primary source. Benchmarking KPIs like organic traffic, bounce rate, site load time, and conversions can help you improve the ROI that comes from website visitors. ## How to Benchmark in Marketing Initiatives ### Define Your Objectives Benchmarking begins with the broad picture of defining your objectives. These aren’t KPIs: Objectives and goals describe, in plain language, what you want to accomplish. This might involve statements like “grow our social audience by 20% within six months,” “increase engagement on Facebook,” or “increase online sales by 20%.” You might also aim to grow your audience of prospects in the consideration or buying stage of the sales funnel, as these efforts often yield the highest ROI. ### Establish Your KPIs Your objectives show what you want to accomplish, while KPIs are the measurements you use to determine if you’ve achieved your goals. KPIs might include followers, shares, comments, or conversions. ### Find Tools That Offer Accurate Analytics Finally, you’ll need a way to measure your KPIs. Look for platforms that provide real-time analytics as well as competitor benchmarks. ### Optimize Your Budget Once you have benchmarks in place, optimize your budget to achieve them. ## Advertisers’ Competitive Benchmarking Competitive benchmarking involves looking at what similar companies in your industry are doing and finding gaps and opportunities in your own processes and tools. While analytics such as CTR can be useful, competitive benchmarking is more about an overall evaluation of how your competitors got where they are — and whether you’re ahead of them, or falling behind. This involves comparing not just analytics, but processes. For instance, a 2025 study from Pipedrive found that roughly one in five sales teams are using AI to analyze data to identify sales patterns (23%) and identify new leads (22%). Meanwhile, 19% of marketing teams are using AI for ad optimization and 8% are using it for programmatic advertising. These aren’t substantial numbers (yet), which means adopting AI tools for advertising can put you ahead of the curve if you act quickly. ## Examples of Benchmarking Benchmarking comes in many forms. You might start with the area where your company needs the most help, or has the most to gain, but ultimately, you’ll want to pursue benchmarking in each of these areas for the best results. ### Strategic Benchmarking Strategic benchmarking looks at where your brand is, where it’s going, and the broad-stroke path you’ll take to get there. Many businesses also look at strategic benchmarking as a comparison of your overall processes against top competitors. You’ll want to analyze their business strategy against yours and see where yours may have gaps. Don’t have the time or the marketing team for a full strategic benchmarking project? Chron recommends a SWOT (strengths, weaknesses, opportunities, threats) analysis as a quick way to identify strategic benchmarks against competitors. ### Internal Benchmarking Internal benchmarking involves comparing where you were to where you are now, based on analytics. You can look at individual benchmarks, including SEO, social, content, and advertising, or take a more holistic view of how your overall content strategy is driving the paid and organic results you want. ### Competitive Benchmarking Competitive benchmarking can be either strategic or granular. It involves looking at how you’re performing compared to competitors. Choosing the right competitors — companies of a similar size with a similar target audience and goals — is often the hardest, yet most important, aspect of competitive benchmarking. ## Aligning Benchmarking With Business Goals Analyzing statistics is great, but you have to make sure that the analytics you’re tracking align with your business goals, whether that’s increased brand visibility or sales. ### Identify KPIs When everyone in your organization knows the KPIs, they can work toward a unified direction, allocate resources effectively, and — perhaps most importantly — know when they’ve achieved success. ### Focus on the Bottom of the Funnel When it comes to advertising, focusing your target audience on those in the consideration stage can boost your ROI. Performance marketing targets those ready to make a buying decision. At that point, your KPI becomes all about conversions, purchases, or subscriptions, not visibility or brand sentiment. Those bottom-of-the-funnel decisions move the needle for your company’s success. ## How to Use Google Analytics for Benchmarking Google Analytics can be a powerful and easy-to-use tool for benchmarking. Google’s GA4 features let you compare data that’s refreshed daily, so you know you’re seeing competitive benchmarks in almost-real-time. Here are the steps to take to use GA4 for custom analytics: ### Setting up GA4 If you haven’t already set up benchmarking in GA4, you’ll need to set your company size and industry preference. This ensures Google is benchmarking your data against similar companies for more accurate, relevant results. Then, enable the “modeling contributions and business insights” in the GA4 admin panel under “account settings.” When you view the overview card, you’ll see a trendline for each metric. Select the benchmarking category and the metric: acquisition, engagement, retention, or monetization. ### Is GA4 Secure? Benchmarking data is encrypted in an aggregated format, allowing you a broad view of your competitors while your data also remains secure and anonymous. ## Reporting and Presenting Benchmarks Benchmarks are helpful tools, but only if the decision makers in your company can clearly understand what to do with the information. Here are some best practices for reporting and presenting benchmarks. Be Clear: Once you’ve gathered the information, drill down on the data, analytics, and conclusions that are most important — the factors that will drive decisions and results. Be as clear and as concise as possible. Use Visualizations: Many marketers find visualizations like graphs and charts helpful when presenting benchmarking data. Tell Stories: Once you’ve shared the facts and figures, share anecdotes and stories to bring conclusions to life. Offer Takeaways: Finally, share an action plan to achieve goals based on benchmarks, whether they’re internal or competitors. Seek out resources that will help, including using all the features in your marketing software or advertising platforms to help you achieve the ROI you want. ## Key Takeaways Benchmarking is crucial for staying ahead of competitors. Understanding what your competitors are doing on a broad level, as well as digging deep into analytics, can help you understand the best tools to deploy and to measure your results against similar companies. Benchmarking against your own results is also useful for adjusting your marketing tactics and improving your ROI. ## Frequently Asked Questions (FAQs) ### What’s a good conversion rate for B2B landing pages? Conversion rates vary widely based on the industry, size of the company, and the landing page optimization tools you use. It also varies based on the type of conversion: For instance, it’s easier to get someone to opt in for a free webinar than to make a purchase. That said, a conversion rate of 10% is commonly accepted as “good.” ### What is the average customer acquisition cost (CAC) in SaaS? The customer acquisition cost is your total sales and marketing spend divided by the number of customers. The average CAC in SaaS varies by industry and company size, and ranges from $91 in telecommunications to $14,772 in fintech. ### What is a good ROAS for e-commerce? A ROAS of 4:1, or $4 earned for every dollar spent, is considered a good benchmark for e-commerce companies, but 2:1 is often considered the average. --- ### Target CPA (tCPA): A Practical Guide for Performance Marketers URL: https://www.taboola.com/marketing-hub/target-cpa/ Last Modified: 2026-06-30 08:17:31 Are you paying for clicks and hoping they turn into something more? That question sits at the heart of target cost per action (tCPA), a bidding strategy that shifts the focus from traffic to outcomes like leads, signups, and sales. Instead of manually setting bids and reacting after the fact, tCPA lets advertisers decide what a conversion is worth and rely on automation to pursue those results at scale. For performance teams juggling growing budgets and increasingly complex campaigns, that shift can make all the difference. ## How Target CPA Works: The Mechanics of Machine Learning Automated bidding may seem mysterious from the outside, but the underlying mechanics follow a clear process. At its core, tCPA combines real-time signals, predictive models, and flexible bidding to make decisions at auction speed. ### Real-Time Auction Signals Every time an ad is eligible to appear, the system evaluates dozens of contextual indicators in milliseconds. These signals can include the device a user is on, their location, the time of day, browser type, recent content engagement, and demonstrated interests. None of these matter on their own — what matters is how they come together at a specific moment. For example, the same user might look far more likely to convert on a mobile device in the evening than on a desktop during work hours. tCPA solutions continuously learn which contextual combinations are most likely to drive conversions. ### Predictive Modeling Once signals are captured, tCPA platforms rely on historical conversion data to estimate the likelihood that a given impression will result in an action. Instead of certainty, the system is making an informed best guess based on patterns it’s seen before. By scoring impressions this way, tCPA solutions can compare thousands of opportunities quickly and consistently. This allows these systems to allocate budgets efficiently across audiences, placements, and moments, without relying on rigid rules or assumptions. ### Dynamic Bidding The final step is bid adjustment. Instead of applying a single fixed bid, the bid is varied for every auction. If an impression has a higher chance of converting, bids increase to improve the chance of winning. For lower-probability impressions, bids are reduced or avoided altogether. The goal is not to hit the target CPA with every single conversion, but instead to maintain that average across the campaign. This is why tCPA performance often looks uneven day to day. Variability is part of the system’s job, allowing it to allocate budget where it expects the best return over time. ## Which Advertisers and Channels Benefit Most? tCPA is not a universal solution, and it performs best when aligned with the right business models and acquisition goals. Here are a few situations where the strategy thrives. ### Ideal Advertiser Profiles In practice, tCPA delivers the strongest results for advertisers with the right combination of scale, discipline, and conversion focus. That includes: - Performance marketers with defined targets: Teams operating against firm cost-per-lead or cost-per-sale goals tend to get the most value from tCPA because success is already measured in conversion economics. - Teams operating at high volume: Managing thousands of combinations across creatives, placements, and audiences quickly outgrows manual controls. Automation becomes less of a convenience and more of a necessity at that scale. - Direct response affiliate marketers: Verticals like finance, e-commerce, insurance, and subscription services often have well-defined funnels and clear post-click outcomes, making them well-suited for conversion-based bidding. ### Primary Channels for tCPA While tCPA is available across multiple environments, it performs strongest in channels that can supply rich intent, behavioral signals, and scalable inventory, including: - Search advertising: Intent-rich environments are a natural fit for tCPA. Users actively looking for solutions provide strong signals, enabling algorithms to optimize aggressively based on conversion probability. - Social platforms: Social channels benefit from their depth of behavioral and demographic data. tCPA can leverage these signals to find users who resemble past converters, even if they aren’t expressing explicit intent. - The open web: Native and display environments on premium publisher sites play a critical role in scaling beyond search and social limits. Conversion-based bidding in these contexts combines broad reach with performance discipline, enabling discovery while still keeping campaigns efficient. Ultimately, tCPA works best when you have clear conversion goals and enough data to guide decision-making. Pair that with channels where people are actively searching, browsing, or discovering content, and the strategy can deliver steady performance at scale. ## Critical Considerations Before You Start While tCPA is powerful, it relies heavily on the inputs it receives. Without the right foundation, results can be inconsistent or misleading. The following factors determine how effectively the strategy can learn and perform: ### Conversion Volume Requirements Most systems require a baseline of conversion data before optimization becomes reliable. As a general rule, 30 to 50 conversions per month is the minimum needed to exit the learning phase. Higher volumes typically lead to faster and more stable performance improvements. ### Tracking Accuracy Matters Automated bidding systems optimize exclusively based on the data they receive. If conversion tracking is delayed, duplicated, or broken, the algorithm can’t distinguish between real outcomes and noise. Accurate post-back integration with your customer relationship management solution, affiliate platform, or analytics tool is not optional: It provides the fuel necessary for the system to run. ### Learning Phase Volatility New tCPA campaigns often experience noticeable fluctuations in the first several days. During this period, the algorithm is testing different combinations of audiences, placements, and bids to understand where conversions are most likely to occur. Remember, short-term instability isn’t a sign of failure, and intervening too early can prevent the system from gathering the data it needs to improve. ## How to Set a Realistic Target CPA The right target CPA sets the tone for everything that follows. A target that’s too aggressive can limit delivery, while one that’s too loose can delay valuable performance insights. The following guidelines can help you strike the right balance. - Start with recent performance as your baseline: Use your average cost per conversion from the past 30-60 days as a reference point, rather than jumping straight to an aspirational target. - Give new campaigns some breathing room: For launches or new channels, begin with a target slightly above your long-term goal so that the system has enough flexibility to gather data and establish patterns. - Account for funnel depth: Higher-function actions like purchases or subscriptions typically require higher target CPAs than top-of-funnel actions such as form fills or sign-ups. - Review performance trends before making changes: Once conversions become consistent, evaluate CPA performance over a reasonable timeframe and adjust targets based on results over time, rather than short-term swings. The most effective target CPAs evolve over time. Treat your target as a flexible input that tightens as the campaign matures, allowing tCPA to improve results without sacrificing stability. ## Best Practices for tCPA Success While most optimization happens automatically, advertiser decisions still play a significant role in long-term performance. Applying a few simple guidelines can help tCPA learn more efficiently and deliver more consistent results over time. ### Budget Headroom Is Essential A commonly cited guideline is to set daily budgets at least 10x higher than your target CPA. This gives the system enough latitude to test variations and distribute spend effectively without being constrained by budget caps. ### Avoid Early Bid Compression Launching a campaign at your exact target CPA can restrict learning. Start with a target set roughly 20% higher to allow the system to find volume. Once conversions stabilize, you can gradually reduce the target to align with your actual goals. ### Limit Structural Changes Major edits reset learning. Frequent changes to creatives, landing pages, or targeting criteria force a system to re-evaluate assumptions. Allowing at least 72 hours between major updates usually helps the system build on what it has already learned, rather than starting over. ## Recommended Tools for tCPA Management Successful tCPA strategies rely on more than bidding automation. These supporting tools can help ensure accuracy, visibility, and informed decision-making. ### Performance Platforms for the Open Web Scaling tCPA beyond search and social requires platforms that support conversion-based bidding across premium publisher environments. Solutions like Realize are built specifically for this purpose, allowing advertisers to apply tCPA strategies across trusted open web destinations like MSN, Yahoo, and major news outlets. By combining real-time contextual and engagement signals with automated bidding, Realize delivers the reach and discovery of the open web while maintaining the efficiency and cost-control performance marketers expect from conversion-driven optimization. ### Real World Success Stories Vodafone Challenge: Vodafone Turkey sought to scale new customer acquisitions for mobile tariffs and home internet, by expanding beyond traditional search and social channels. Feature/Strategy Used: The brand implemented Target CPA (a SmartBid solution) to automate bidding for mobile and desktop audiences, specifically targeting those most likely to convert. Results: The campaign delivered a 16% lower CPA than the target goal, achieving the lowest cost-per-acquisition across all of Vodafone Turkey’s local and programmatic channels. Digital Athlete Challenge: Tasked with driving sales for a high-tech kitchen appliance, the agency needed to maximize completed orders while maintaining a strict cost-per-purchase limit. Feature/Strategy Used: They utilized Maximize Conversions with Target CPA, allowing AI-based algorithms to automate bids to drive high volume within their predefined budget cap. Results: The strategy resulted in a 12% lower CPA than their target, and a 27% lower CPA than the competitor platform average, while increasing overall customer spend by 348%. Verisure Challenge: Verisure Argentina needed to increase its lead-to-booking (L2B) rate — the frequency of leads converting into scheduled home security quotes — at the most efficient cost possible. Feature/Strategy Used: The brand combined contextual targeting with Maximize Conversions with Target CPA, to automate bidding toward high-quality leads on premium publisher sites. Results: Verisure exceeded its lead-to-booking goal by 85%, successfully identifying high-intent users on the open web and outperforming internal industry benchmarks for conversion rates. ### Attribution and Tracking Software Accurate conversion reporting is foundational to tCPA performance, since automated bidding systems optimize exclusively on the data they receive. Attribution and tracking software ensure that post-click actions like leads and purchases are reliably captured and passed back to the bidding system without delay, duplication, or loss. By resolving challenges such as cross-device behavior, delayed conversions, and multi-touch customer journeys, these tools reduce attribution gaps and provide cleaner signals for optimization. This clarity helps tCPA algorithms learn from true outcomes, rather than incomplete data, resulting in more stable bids, improved efficiency, and better alignment between automated decision-making and real business results. ### Competitive Intelligence Tools Monitoring creative trends and funnel strategies within a vertical helps inform testing priorities and shorten the path to implementation. Competitive intelligence tools enhance tCPA strategies by helping advertisers optimize the inputs that automated bidding acts upon — particularly creative, messaging, and funnel structure. ## Is tCPA Right for Your Strategy? While tCPA reduces the need for manual bidding, it places greater importance on accurate data, quality creative, and allowing time for learning. Campaigns without these elements often struggle to stabilize. When those conditions are met, tCPA gives performance teams a practical path to scale without sacrificing control over cost per conversion. ## Frequently Asked Questions (FAQs) ### How long does the “Learning Phase” actually take? In most advertising environments, the learning phase lasts between five and 10 days, depending on how quickly a campaign generates conversions and whether the budget allows enough flexibility for testing. During this period, performance may fluctuate as the system evaluates different bids, audiences, and contextual signals to understand what drives conversions. When conducting performance campaigns on the open web, the conversion phase can vary slightly because open-web environments emphasize discovery in addition to intent. Campaigns may test across a wider mix of premium publisher content before patterns stabilize. ### What happens if my campaign isn’t spending its full daily budget? When a tCPA campaign underspends, it’s often a sign that the target CPA is too restrictive relative to the available conversion opportunities. Modestly increasing the target, or expanding eligibility through broader placements or additional creatives, can help unlock delivery without sacrificing efficiency. On the open web, budget pacing can also be influenced by content availability and real-time user engagement across publisher sites. Because campaigns run in dynamic editorial environments, shifts in traffic patterns or content relevance can affect spend. Optimization and inventory controls allow advertisers to adjust creative formats, placement coverage, or bid thresholds to improve delivery while keeping campaigns aligned with conversion goals. ### Can I use tCPA for a brand new campaign launch? tCPA can be used for new campaigns, but early performance may be uneven if there’s little to no historical conversion data to guide optimization. In these cases, the system has limited context and needs time and volume to learn which signals correlate with success. For performance advertisers on the open web, a common approach is to start with a higher initial CPA target, or use alternative bidding strategies to gather early conversion data across open-web placements. Once enough post-click data is collected, campaigns can transition fully into tCPA, allowing the platform’s automated bidding and contextual intelligence to optimize more effectively across premium publisher environments. --- ### Topic vs. Keyword Targeting: Which Strategy Delivers Higher ROI for Performance Advertisers? URL: https://www.taboola.com/marketing-hub/topic-targeting-vs-keyword-targeting/ Last Modified: 2026-07-27 09:04:08 As performance advertising grows and matures, many marketers are discovering the limits of relying on search and social alone. Rising costs, creative fatigue, and increasingly crowded environments are forcing advertisers to think more carefully about how and where they reach potential customers. Two of the most commonly used contextual strategies, topic targeting and keyword targeting, offer distinct ways to connect with high-intent audiences without relying solely on personal identifiers. While they’re often discussed interchangeably, these two approaches function very differently in practice. Understanding their unique benefits and disadvantages can have a direct impact on efficiency, scale, and ultimately, results for your campaigns. ## Topic Targeting ### Description Topic targeting allows advertisers to place ads within content environments that align with broader themes or subject areas, such as fitness, personal finance, travel, or home improvement. Instead of focusing on specific words, this method evaluates the overall context of a page to determine whether it matches a defined topic category. ### How It Works Topic targeting relies on contextual signals across an entire piece of content, including language patterns, semantic meaning, and historical engagement data. Advanced systems assess not just what a page says, but what it’s about in a more holistic sense. This enables ads to appear alongside relevant articles, videos, and editorial content that naturally align with what a brand offers, even if exact-match keywords aren’t present in the content itself. Because this approach analyzes intent at the broader content level, it adapts well to formats like long-form articles, visual placements, and immersive ads. It also allows creative to be matched with high-visibility placements beyond traditional text-based environments. ### Benefits One of the biggest advantages of topic targeting is scalability. Access is granted to entire content categories, rather than narrow search terms, so campaigns can reach broader, qualified audiences without sacrificing relevance. This makes it especially effective for brands seeking visibility during the consideration phase of their sales cycle, where users are actively researching or comparing options. Topic targeting is also privacy-focused. Since it doesn’t depend on individual identifiers, it aligns well with more modern privacy expectations from users, while still delivering strong performance outcomes. The broader context of the targeting also allows creative to feel more natural and less intrusive, which can reduce ad fatigue and improve engagement over time. ### Considerations As topic targeting operates at a higher level, it may not always capture users with immediate, transactional intent. Advertisers focused solely on bottom-of-funnel conversions may need to pair this approach with optimization tools that prioritize performance signals and rapid learning cycles. Another consideration is creative alignment. Success depends on messaging that resonates within a given topic environment, which may require more thoughtful variation than keyword-driven ads that mirror search queries more directly. ### Use Cases Topic targeting is particularly effective for industries where purchase decisions are influenced by research, inspiration, or lifestyle alignment rather than an immediate transactional need. For instance, home and lifestyle brands selling decor or improvement services perform well when appearing alongside content about interior design trends, renovation planning, or seasonal home projects. In health and wellness, topic targeting allows brands to engage audiences reading about fitness routines, mental wellbeing, or preventative care, even when those users aren’t actively searching for a product. This makes it well suited for supplement brands, wellness apps, or clinics promoting services tied to broader lifestyle goals. ## Keyword Targeting ### Description Keyword targeting focuses on placing ads on pages that contain specific words or phrases chosen by the advertiser. This approach mirrors the logic of traditional search engine advertising by aligning ads with the same language that users are actually searching, signaling higher levels of intent. ### How It Works With keyword targeting, advertisers define a list of terms related to their product or service. Ads are then served on pages where those terms appear in titles, headings, or body copy. Some systems allow for inclusion and exclusion lists, enabling tighter control over where ads do and do not appear. This method is particularly effective when paired with predictive optimization and real-time bidding strategies, allowing campaigns to quickly identify which keywords drive the strongest performance. ### Benefits Keyword targeting offers precise targeting through aligning ads with specific phrases. This means that consumers can be reached when they’re actively expressing intent related to a product category, problem, or solution. This often translates into faster feedback loops and quicker performance signals, which is especially valuable for teams seeking visible results in short timeframes. This approach also provides a high degree of control. Advertisers can refine keyword lists, pause underperforming terms, and test variations with relative ease. This makes it an effective strategy for iterative optimization and A/B testing. ### Considerations While keyword targeting is a powerful approach, it can be limited in scale. Narrow keyword sets may restrict reach, while broader terms can introduce inefficiencies or brand safety concerns if not carefully managed. Costs can also rise quickly in competitive categories, particularly when multiple advertisers are targeting the same high-value keywords. Additionally, keyword targeting doesn’t always account for nuance or context. A page containing a keyword may not always align with the advertiser’s intent, requiring ongoing monitoring and refinement. ### Use Cases Keyword targeting is ideal for campaigns focused on high-intent actions, like lead generation, product comparisons, or time-sensitive promotions. It’s especially useful when advertisers want granular control and the ability to optimize quickly based on performance data. Financial services companies offering tax services or lending often rely on keyword targeting to reach users researching specific solutions or comparisons at the moment of decision-making. For B2B and SaaS companies, this approach can be effective for lead generation when focused on problem-aware audiences. Aligning ads with pages discussing operational challenges, competitive platforms, or implementation questions can capture users already evaluating their options. For e-commerce brands, keyword targeting works well for competitive conquesting and lower-funnels campaigns. By targeting keywords related to competing products, reviews, or “best alternative” content, advertisers can intercept shoppers who are ready to convert. ## How Do Topic Targeting and Keyword Targeting Compare? Feature Topic Targeting Keyword Targeting Privacy Compliance Fully contextual, no reliance on user identity Contextual, dependent on specific language signals Campaign Goal Consideration-driven scale and efficient reach High-intent actions and direct response Setup Complexity Faster to launch with broader parameters Requires keyword research and ongoing refinement Audience Scalability High, across entire content categories Moderate, limited by keyword volume Immediate Performance Strong with optimization and learning Often faster initial signals but can taper, depending on competition and budget Long-Term Brand Lift Supports sustained visibility and recall More transactional in nature, so less likely to have long-term sustainability Cost Efficiency Helps mitigate rising acquisition costs Can become costly in competitive spaces Automated AI Integrations Benefits from predictive intent modeling Relies more on rule-based matching Brand Safety/Suitability Controlled through topic and sentiment analysis Managed through inclusions and exclusions A/B Testing Effective for creative and format testing Effective for keyword and message testing ## How to Decide When to Choose Topic Targeting Topic targeting is a strong choice when advertisers want to move beyond crowded search and social environments without sacrificing performance. It’s a particularly effective approach for brands experiencing creative fatigue, rising costs, or limited scale on traditional channels. By focusing on user intent at the content level, this approach allows campaigns to reach engaged audiences in moments of discovery and evaluation, while leveraging automation to optimize towards action. ## How to Decide When to Choose Keyword Targeting Keyword targeting is best when advertisers with clearly defined intent signals need more rapid results. When campaigns are built around specific problems, solutions, or competitive alternatives, keyword targeting can deliver efficient results quickly. It works well for offers operating within a tight scope, niche products, or instances where granular control outweighs the need for broader reach. ## Key Takeaways Topic targeting and keyword targeting aren’t opposite and opposing strategies. Instead, they should be thought of as complementary tools within a modern performance marketing toolkit. Topic targeting excels at scalable, privacy-focused reach that captures users during meaningful moments of consideration, while keyword targeting delivers precision and speed for intent-driven campaigns. The most effective advertisers understand how to deploy each method strategically, often combining them with automation, flexible creative formats, and predictive optimization to drive measurable results across the performance funnel. ## Frequently Asked Questions (FAQs) ### We are a travel brand needing to avoid appearing next to negative news like airline strikes; how do we ensure safer placement? Using topic-level controls, combined with sentiment and suitability filters, will allow you to serve ads that only appear within positive or neutral content, reducing the risk of appearing alongside unfavorable news. ### We are a niche lead-gen firm for personal finance software; how do we specifically reach users who are currently using a competitor’s platform? Keyword targeting focused on competitor comparisons and solution-oriented language can help capture users who are actively evaluating alternatives, especially when paired with real-time optimization. ### We are a health brand promoting seasonal flu clinics; how do we capture mass awareness during a short outbreak window? Topic targeting enables rapid scale across relevant health and wellness content, allowing brands to reach large audiences quickly, while still maintaining contextual relevance during time-sensitive campaigns. --- ### Broad Targeting vs. Retargeting: How to Choose the Right Strategy for Your Campaign Goals URL: https://www.taboola.com/marketing-hub/broad-targeting-vs-retargeting/ Last Modified: 2026-05-31 11:47:36 Digital advertisers have more targeting options at their disposal than ever before, but more choice doesn’t always make decisions easier. For example, two of the most widely used approaches, broad targeting and retargeting, sit at opposite ends of the audience targeting range. One prioritizes scale and discovery, while the other focuses on precision and efficiency. Both approaches can be highly effective when used in the right context. The challenge is to understand how they work, what they optimize for, and how you can align each method with your specific campaign goals. In this guide, I’ll break down broad targeting vs. retargeting, compare their strengths and limitations, and help you decide when and how to use each strategy effectively. ## Broad Targeting Broad targeting is an advertising strategy that prioritizes reach and scale over defining a narrow audience. Instead of relying on detailed user-level data or past interactions, broad targeting enables platforms to serve ads to a broad audience that matches general parameters such as geography, language, device type, or content environment. Instead of asking advertisers to pre-define exactly who should see an ad, broad targeting relies heavily on platform-level optimization and machine learning to identify users most likely to engage or convert. ### How It Works With broad targeting, advertisers set minimal constraints at the campaign level and allow the platform’s optimization engine to do the heavy lifting. On performance advertising platforms like Realize, e.g., this means leveraging AI-driven signals across content consumption, contextual relevance, and real-time engagement to continuously refine delivery. Let’s look at an example: A health and beauty brand might launch a campaign using broad targeting with two basic constraints, like geography (e.g., Canada) and device type (e.g., mobile). Instead of defining audiences upfront, Realize analyzes how users interact with health, beauty, and lifestyle content across its network by tracking signals such as article topics read, scroll depth, time on page, and recent engagement with related offers. As campaigns run, the system adapts automatically, shifting delivery based on performance signals to improve efficiency over time without requiring constant manual adjustments. ### Benefits Broad targeting offers significant advantages for advertisers focused on growth and discovery. By removing tight audience constraints, campaigns can scale quickly and adapt to changing market conditions. This approach is particularly effective for reaching new users who may not yet be familiar with a brand or product. Broad targeting also reduces dependency on third-party cookies or historical user data, making it well suited for privacy-first environments. As signal loss increases across the industry, broad targeting provides a future-proof way to continue reaching relevant audiences without relying on granular tracking. ### Considerations While broad targeting excels at scale, it may take longer to reach peak efficiency compared to more focused approaches. Performance often improves as the algorithm gathers data, which means advertisers need to allow sufficient learning time and budget. Creative quality also plays a larger role. Because ads are shown to a wider audience, messaging must be clear, compelling, and broadly relevant to resonate with users at different stages of awareness. ### Use Cases Broad targeting is ideal for upper-funnel and mid-funnel campaigns focused on awareness, consideration, or demand generation. It’s commonly used for brand launches, new product introductions, and expansion into new markets. It also performs well for advertisers with limited first-party data, or those looking to move beyond reliance on retargeting pools that may be shrinking or saturated. ## Retargeting Retargeting focuses on reaching users who have already interacted with a brand, such as visiting a website, viewing a product, or engaging with previous ads. By targeting users with known intent signals, retargeting aims to drive conversions more efficiently. This approach assumes prior awareness and leverages familiarity to move users further down the funnel. ### How It Works Retargeting campaigns use audience lists built from first-party data, such as website visits, app usage, or CRM records. These audiences are then segmented based on behavior, e.g., users who abandoned a cart versus those who viewed a product category. Following the same example used earlier, a health and beauty brand might run a retargeting campaign to sell skincare products, using first-party data from its website. One audience could include users who viewed a specific serum product page but didn’t complete checkout, while another could include past customers who purchased a moisturizer within the last 60 days. Tailored ads would be delivered to each segment, highlighting dermatologist-backed benefits or limited-time offers for cart abandoners, and promoting complementary products or refill reminders to existing customers, helping the brand re-engage high-intent users and drive conversions more efficiently. Ads are delivered specifically to both groups, often with tailored messaging designed to overcome friction, reinforce value, or prompt action. Because the audience is smaller and more defined, retargeting typically delivers faster performance signals. ### Benefits One of the key advantages of retargeting is efficiency. Because ads are served to users who have already shown interest, conversion rates tend to be higher and cost per acquisition lower, especially in the short term. Retargeting also allows for highly personalized messaging: Advertisers can align their ad creative with specific behaviors, making ads feel more relevant and timely. ### Considerations A drawback of retargeting is its limited reach due to audience size. Once a retargeting pool reaches saturation, performance can plateau or decline due to ad fatigue and diminishing returns. There are also increasing privacy and compliance considerations: Changes to browser policies, platform restrictions, and user consent requirements can reduce the size and reliability of retargeting audiences over time. ### Use Cases Retargeting is most effective for lower-funnel objectives such as driving purchases, registrations, or renewals. It’s commonly used in e-commerce, subscription services, and lead-generation campaigns where prior engagement is a strong predictor of conversion. It also works well for short-term promotions, abandoned cart recovery, and upsell or cross-sell initiatives. ## How Does Broad Targeting Compare to Retargeting? While both strategies aim to improve campaign performance, broad targeting and retargeting differ significantly in how they approach audience selection, optimization, and scale. Below is a high-level comparison across key attributes. Feature Broad Targeting Retargeting Privacy Compliance Best for awareness, discovery, and long-term growth Depends on first-party data and user consent, which can limit reach Campaign Goal Best for awareness, discovery, and long-term growth Optimized for conversion and immediate action Setup Complexity Simple setup with minimal audience configuration Requires audience creation, segmentation, and ongoing management Audience Scalability Highly scalable with access to large audiences Limited by the size of existing engagement pools Immediate Performance May take longer to optimize Often delivers faster initial results Long-Term Brand Lift Strong potential for sustained brand impact Limited impact beyond existing users Cost Efficiency Improves over time as algorithms learn Efficient early, but can decline with saturation Automated AI Integrations Highly compatible with AI-driven optimization Benefits from automation, but remains data-dependent Brand Safety/Suitability Can leverage contextual controls for suitability Strong relevance but limited control over broader exposure A/B Testing Useful for testing creatives and messaging at scale Ideal for testing offer and conversion messaging ## How to Decide When to Choose Broad Targeting Broad targeting is the right choice when scale, adaptability, and long-term performance are priorities. If your campaign relies on reaching new users or expanding beyond known audiences, broad targeting allows platforms like Realize to continuously explore and optimize across a wide inventory of high-quality content environments. This approach is especially effective when advertisers want to reduce manual audience management and instead rely on AI-powered optimization to guide delivery. For brands operating in privacy-conscious markets or dealing with limited first-party data, broad targeting provides a resilient foundation for sustainable growth. ## How to Decide When to Choose Retargeting Retargeting is best suited for campaigns with clear lower-funnel objectives and a reliable source of first-party data. If your goal is to convert existing interest into action, whether a purchase, sign-up, or renewal, retargeting provides efficiency and precision. It works particularly well when user intent is high and decision cycles are short. However, advertisers should monitor frequency, audience size, and performance trends closely to avoid diminishing returns. Retargeting is most effective when treated as part of a broader strategy, rather than the sole driver of growth. ## Key Takeaways Broad targeting and retargeting are not competing strategies: They are complementary tools designed for different stages of the customer journey. Solutions like Realize can support both approaches, by using AI-driven optimization to scale broad targeting while still enabling precise retargeting when first-party data is available. The most effective advertising strategies often combine both approaches, using broad targeting to build awareness and retargeting to capture intent. By aligning each method with the right goals, budgets, and creative strategies, advertisers can drive sustainable performance across the funnel. ## Frequently Asked Questions (FAQs) ### How does retargeting improve conversion efficiency for e-commerce campaigns? Retargeting improves conversion efficiency by focusing spend on users who have already shown interest, such as product viewers or cart abandoners. This familiarity reduces friction, shortens decision cycles, and often leads to higher conversion rates when compared to prospecting campaigns. ### Can combining broad targeting and retargeting improve campaign ROI? Yes. Broad targeting helps expand the top of the funnel and introduce new users, while retargeting captures value from those interactions. When used together, advertisers can maintain scale while improving overall return on investment. ### Which method, broad targeting or retargeting, is more effective for high-frequency seasonal campaigns? Retargeting often performs well for short, high-frequency seasonal campaigns due to its immediate efficiency. However, broad targeting can play a critical role in replenishing audiences and driving awareness ahead of peak periods, making a combined approach the most effective option. --- ### How to Use Contextual Targeting to Scale Consumer Electronics Campaigns URL: https://www.taboola.com/marketing-hub/contextual-targeting-to-scale-consumer-electronics-campaigns/ Last Modified: 2026-03-09 20:15:05 Consumer electronics advertising presents some specific challenges for advertisers: many social media ads mislead consumers, and there’s an abundance of dropshippers, which don’t always fulfill their promises. Plus, consumer electronics generally market direct-to-consumer (D2C), so brands have to capture attention quickly from busy users. Then, there’s the sheer scale of the global consumer electronics market — reaching $1.03 trillion this year — which can easily overwhelm advertisers. Premium consumer electronics brands must build trust and find serious buyers, both of which can be an uphill battle. That’s especially true as search and social channels become oversaturated. Plus, many segments within electronics, like smart projectors, require a research-based process before purchase. That said, there are still areas to focus on for advertising success, depending on the particular product, its users, and the typical buying journey. Here, you’ll find campaign-saving tips from Realize advertising expert Ari Del Rosario, who has worked closely with consumer electronics clients. He has developed strategies for using contextual targeting, based on his work with a Realize client, a U.S.-based hardware innovator. This smart projector company needed to diversify its digital strategy to reach audiences across the open web, where high-intent decisions are actually made. Read the advice below and learn how to make contextual targeting work for your business, to scale and expand reach. ## 1. Transition From Social Saturation to Contextual Relevance This U.S.-based consumer electronics company, which makes smart projectors, needed to reach the right user in the right mindset. They had seen significant return on ad spend (ROAS) decay on Meta (dropping from 5x to 2x) as the platform became saturated with competitor ads. The recommendation: Move from audience-only targeting to contextual targeting. In other words, instead of targeting a person because they show an interest in projectors on social media (where they are immediately shown seven competitor ads), the business should target the environment where that user is actively consuming tech and home-lifestyle content. ### Focus on Winning the Mindset Over the Profile Contextual targeting places your brand alongside editorial content that mirrors the user's current intent. If a user is reading a 4K home theater guide on a premium site like Yahoo or NBC, they are in a research mindset, which is far more valuable than a passive scroller. “The challenge with social platforms is that once a user looks at your ad, the algorithm floods them with every competitor in the market,” says Del Rosario. “With contextual targeting on Realize, we find users when they’re in information-gathering mode. By appearing on premium publisher sites related to tech and home entertainment, we establish brand authority before the user even reaches a search bar.” ## 2. Map Contextual Signals to a Multi-Device Funnel The smart projector client noted that their product — a $250+ electronics item — is rarely an impulse buy. Users often discover the item on a mobile device, but prefer to pull the trigger and make the actual purchase on a desktop. The recommendation: Choose a dual-campaign structure. This means that they should use contextual targeting on mobile to drive awareness and interest, then utilize those signals for high-intent contextual retargeting on desktop to close the sale. ### Navigating the Research-Based Purchase High-ticket items require multiple buying journey touchpoints. By targeting specific contextual categories across the open web, Realize clients create a surround sound effect that follows the user from mobile discovery to desktop conversion. “People don't usually buy a $250 projector impulsively on their phone while walking around,” notes Del Rosario. “They do their research. That’s why we recommend a funnel that starts with contextual awareness on mobile — hitting those computer-in-pocket moments — and then retargeting contextually on desktop. This aligns the ad placement with the device the consumer feels most comfortable using for a major transaction.” ## 3. Boost Contextual Performance With Native Motion Assets To stand out against generic overseas dropshippers, the projector brand needed to emphasize its premium U.S.-based engineering and Google TV integration. The recommendation: Enhance contextual placements with native motion ads, such as GIF-like assets, to increase click-through rate (CTR) without disrupting the editorial experience of the site. ### Use Visual Storytelling in an Editorial Environment One of Realize’s key features is the use of assets that look and feel native to the targeted premium publisher sites. This helps brands build qualitative brand equity that social ads just cannot provide. “For a tech brand, standing out is about differentiation,” says Del Rosario. “We recommend using native motion — that is, turning static product shots into dynamic GIFs — within relevant contextual environments. This catches the eye of a reader on Apple News or ESPN without feeling like a hard sell banner ad, which is crucial for building a premium brand image.” ## Key Takeaways Diluted search traffic, saturated social feeds, and ad-fatigued users all mean that consumer electronics companies have to think differently about scalable growth. The old methods are no longer enough to find and engage high-intent audiences. Brands using contextual targeting in the Realize platform are able to meet customers in high-intent environments, where they’re already browsing, researching, and generally seeking relevant product information. When you align with premium editorial content for reach, you reach the right buyers at the right moment, when they’re ready to listen. ## Frequently Asked Questions (FAQs) ### What is contextual targeting and how does it differ from traditional audience targeting? Broadly, contextual targeting looks at what a prospect is reading or consuming right now. This is in contrast to traditional targeting, which looks at demographics and other details about a prospect, such as age, interests, and past history. When it comes to performance campaigns on the open web, contextual targeting prioritizes immediate user intent and environmental relevance, helping you reach high-intent customers in cookieless and privacy-safe places where their current state of mind matches your offer. For example, a performance marketing platform might analyze millions of pages across an exclusive publisher network (think MSN and Yahoo) to match a brand’s ads to the specific topic of an article. For a smart projector company, their ad would appear next to tech reviews, home theater guides, or best-of gadget lists, capturing intent in real time. ### Can I use my existing social media videos for contextual campaigns? Existing video assets can almost always be adapted for native web ad placements. When you’re considering performance campaigns on the open web, try adapting the creative to match the tone and informational intent of the surrounding publisher content where the video will play. Instead of disrupting a prospect’s experience, repurposed social videos should use value-driven or editorial-style storytelling to align with the research-oriented mindset of open web users. ### How do you track a long research-based funnel through contextual targeting? Tracking a research-based funnel, like one for consumer electronics, usually involves cookies and pixels to follow a user from first click to final purchase. For a long research-based funnel on the open web, a multi-touch attribution model is a more detailed way to capture micro-conversions, like whitepaper downloads or time on page. This multi-touch model measures how specific contextual environments influence different stages of the buyer’s journey, and lets you map the transition from early-stage research to more high-intent actions. This ensures that you can then further train the performance marketing platform’s AI to optimize for the entire conversion path versus the final click. --- ### Topic Targeting vs. Contextual Targeting: Which Is Right for Your Campaign? URL: https://www.taboola.com/marketing-hub/topic-targeting-vs-contextual-targeting/ Last Modified: 2026-03-08 09:34:11 The third-party cookie hasn’t completely crumbled, but it has gone stale. Firefox and Safari have banned them, but in response to backlash from the advertising industry and regulatory pressure, Google has moved toward a user-choice model in Chrome. As privacy regulations tighten and third-party data fades, marketers are pivoting away from — or augmenting — reliance on external trackers toward relevance based on: - First-party data. - Contextual targeting. - Zero-party data. This shift to a privacy-first approach requires establishing or rebuilding direct, trusted relationships with consumers to maintain the effectiveness of your ads. Essentially, you’ve got two options: using a laser pointer (contextual targeting) or casting a wide net (topic targeting). Each allows you to reach people without a tracking pixel following them, but knowing the strengths and weaknesses of each — and which targeting to use when — can significantly impact your campaign’s success. ## Contextual Targeting ### Description This targeting channels immediacy, placing your ads based on the specific content a user is currently viewing. Contextual targeting doesn’t care what a user did yesterday; it cares that they’re reading an article on organic gardening now. The strategy aligns your ads closely with the page’s topics, enhancing relevance and engagement. ### How It Works When someone visits a page, contextual targeting analyzes the text, keywords, and the content’s overall sentiment. Then it displays ads matching the article’s context. If your user is reading reviews of the latest smartphones, they might see ads for phone accessories or plans. Performance advertising platforms like Realize enhance contextual targeting by using artificial intelligence (AI) trained on proprietary data. They use semantic analysis to distinguish between a casual travel guide and a high-intent equipment review, leveraging code-on-code integrations with thousands of premium publishers (Realize, e.g., counts publishers like NBC News and Yahoo among its partners). The platforms analyze content metadata, user intent, and image sentiment in real time, so ads for a specific product or service appear when readers are more likely to click, consider, and buy. ### Benefits Contextual targeting’s benefits include: - Focusing on hyper-relevance: Ads relate directly to what the audience is reading, increasing the likelihood of engagement. - Prioritizing privacy: Contextual targeting is 100% cookieless. It doesn’t rely on user history because you’re tracking topics, not people. - Protecting brand safety: By analyzing sentiment on individual pages, contextual targeting is less likely to pop your ad next to negative reviews. - Driving immediate impact: Contextual targeting can generate higher conversion rates because it catches users already primed to check out and potentially purchase your product. ### Considerations The cost of this granularity is scale. This strategy works best for niche products, and if your specific target doesn’t have a high volume of new articles going live today, your daily spend might struggle to keep pace. Another consideration is that low content quality may result in lower user engagement with the ads. ### Use Cases Contextual targeting involves finding current conversations and sliding your messaging into them. Take the 2026 Winter Olympic Games in Milan-Cortina — the perfect opportunity for contextual targeting to have a moment. Sports are, of course, a rollercoaster. One minute, a nation is celebrating a surprise gold in alpine skiing; the next, a country’s fans are heartbroken over a disqualification. Brands use AI-driven contextual tools to scan the vibe of news articles and social feeds in real time. Let’s say, for example, that a Nike or Under Armour-sponsored underdog athlete wins a medal. That brand’s ads could populate instantly next to the news stories covering their unexpected (and awesome) victory. This contextual targeting captures high-arousal emotions. When a reader feels inspired, or experiences national pride, they’re more likely to recall an ad matching their mood compared to a random, retargeted ad for socks they glanced at a week ago. One of the Olympics’ fun, unique qualities is bringing niche sports like curling, luge, or skeleton into the mainstream for two weeks. Instead of targeting sports fans (too broad and expensive), brands target hyper-specific keywords and themes like “downhill speed techniques.” A luxury watch or car brand might place ads exclusively within content discussing bobsledding’s technical mechanics or the physics of figure skating. These brands bypass the gold medal price tag by targeting content rather than the prime-time broadcast. Smaller brands can reach a highly focused, engaged audience without competing for the $10 million in ads at the women’s figure skating finals. For campaigns with high-ticket items and longer research cycles, contextual targeting offers a powerful way to reach consumers in an information-gathering mindset. A prime example is how consumer electronics brands use contextual targeting to scale, moving beyond the saturation of social media to place ads alongside relevant tech reviews and home-lifestyle content. By appearing in these high-intent environments, brands can establish authority and guide users through a complex purchase journey more effectively than traditional audience-based methods. ## Topic Targeting Topic targeting takes a broader, category-level approach, placing ads on pages categorized under specific themes. It identifies the general umbrella a website or app falls under, like health or autos and vehicles. ### How It Works Platforms categorize millions of publishers into buckets. Selecting a topic like personal finance or technology expands your ad’s reach, and topic targeting can appear across any page within those domains. These ads focus less on a specific paragraph someone’s reading and more on the neighborhood they’re visiting. Performance ad platforms identify a page’s underlying theme by analyzing the semantic relationship between the content and the site’s historical data. With this neighborhood approach, you can place your ads across a broad range of relevant sites, ensuring your audience feels your brand presence wherever they hang out. ### Benefits Topic targeting’s benefits include: - Scaling: The broader reach of topic targeting makes it ideal for top-of-funnel campaigns, helping brands quickly reach millions of eyes. - Offering flexibility and control: You can choose specific topics, combine them with other targeting methods (like keywords), or exclude irrelevant topics to maintain brand safety. - Enhancing brand awareness: Topic targeting strengthens brand visibility and recognition among target audiences when you place your ads in contextually relevant environments. ### Considerations What you gain in broad reach with topic targeting, you lose in precision. Your ad may appear on a page broadly about autos, but the user might be reading about the history of 1950s cars, rather than researching SUVs to buy today. ### Use Cases Topic targeting channels the vibe. Contextual targeting is buying an ad on a specific article, like “Best Wax for Downhill Skis.” Topic targeting tells the system: Show my ad to anyone reading about winter sports across the entire open web. This strategy is fantastic for reaching audiences interested in the sport and the lifestyle surrounding the event. You could select broad interest categories like luxury travel, refined craftsmanship, or healthy living, and your ads would appear on thousands of sites related to those themes, even if they don’t mention the Olympics by name. To continue with the Olympics examples, when Italy hosted the 2026 Games, a brand like Armani or a high-end coffee company could have targeted “Italian culture and gastronomy.” These ads build an emotional connection: If someone’s reading about Milanese architecture or how to make the perfect espresso, seeing a sleek ad with an Olympic tie-in feels like a natural extension of their current interests, rather than a jarring interruption. Think, also, about brands interested in reaching the tech-obsessives of a specific category — those who care about the how as much as the who. You could pick a subtopic, like physics and engineering, or textile innovation. A tech company or material sciences firm could target “advanced engineering” during the Games. Their ads would appear on articles discussing the aerodynamics of bobsled designs, the fracas surrounding the modified ski jump suits, or the tech behind speed skating suits. Placing your brand in the middle of smart content positions your product as a high-performance tool. Sure, you’re selling to casual sports fans, but you’re also selling to a college physics professor who geeks out over technical specs. ## How Do Contextual Targeting and Topic Targeting Compare? Feature Contextual Targeting Topic Targeting Privacy Compliance Highest (no user data/history needed) High (may rely on user-level behavior or site categorization) Campaign Goal Immediate engagement, high-intent leads/direct sales, or conversion Brand awareness/traffic volume Setup Complexity Higher (requires keyword/sentiment lists) Lower (select from pre-set categories) Audience Scalability Moderate (limited to niche audiences) High, though variable depending on broader interest Immediate Performance High click-through rate (CTR) due to relevance; can be more cost-effective for niche ads Moderate CTR; broader reach; higher volume may increase ad spend Long-Term Brand Lift Possible, but likely not immediate Effective for lasting brand awareness Cost Efficiency Higher cost per acquisition (CPA), but higher conversion Lower cost per mille (CPM), but lower intent Automated AI Integrations Realize allows you to create custom contextual segments from 70,000+ granular topics Realize has built-in topic targeting from 9,000+ publisher sites Brand Safety/Suitability Page-level precision: high; ads appearing with relevant content Moderate; requires careful selection and exclusion lists A/B Testing Effective for testing specific ads Useful for testing broader themes ## How to Decide When to Choose Contextual Targeting Choose the surgical approach of contextual targeting when your product solves a very specific problem. If you’re a B2B firm selling cybersecurity insurance, you want someone reading about a recent data breach, not someone perusing general tech articles. Choose this method if your goal is: - Immediate engagement and conversion, particularly in a high-interest context. - Complying with strict privacy regulations. - Addressing creative fatigue on social media and needing a fresh look. ## How to Decide When to Choose Topic Targeting Opt for topic targeting when you want to run a broad awareness campaign and increase your share of voice. This method works well if you want to reach a large audience quickly and your brand fits into a general theme. If you’re a bank or airline, you should be everywhere your target demographic hangs out. Topic targeting ensures that users interested in investing consistently see your brand across news sites, blogs, and financial apps. Choose this method if your goal is: - Broadening brand awareness by increasing visibility and associating your brand with specific industries or lifestyle topics. - Scaling campaigns when you have a flexible budget and want to ramp up delivery quickly. - Complying with increasingly strict privacy regulations by reaching users based on what they’re consuming, rather than by using their personal data. ## Key Takeaways Contextual targeting is ideal for precise, niche campaigns because it offers high relevance and privacy compliance. Topic targeting, meanwhile, works better for broad awareness campaigns because it maximizes reach and impressions. Both methods are superior to traditional behavioral targeting with third-party data in a privacy-regulated world. ## Frequently Asked Questions (FAQs) ### How do content volume and site diversity impact the choice between topic and contextual targeting? Content volume and diversity will affect your targeting choice. Contextual targeting thrives on high-quality, relevant content, but if you’re in a niche with low content volume, this method may not generate enough impressions to justify the budget. Topic targeting benefits from a wider range of sites and themes, so if your audience tends to visit diverse sites, this method may be more effective. ### Can combining topic and contextual targeting improve cross-device campaigns? Absolutely. You can use topic targeting to find likely candidates across mobile apps and the open web, and then use contextual targeting to serve the closing ad when they’re using their desktop to research your specific product category. By engaging users across various devices with both methods, you can maintain a cohesive brand message. ### I’m a brand needing to avoid specific negative news (brand safety). How do I ensure my ads don’t appear on a travel page with content about a plane crash? Enter brand suitability! Topic targeting might place you on any travel site, but contextual targeting uses negative keyword lists and AI sentiment analysis to evaluate content. If the AI identifies an article about a tragedy, not a vacation, it will automatically block your ad from appearing. --- ### Contextual Advertising: Ad Placements That Make Sense, No Cookies Required URL: https://www.taboola.com/marketing-hub/contextual-advertising/ Last Modified: 2026-03-08 09:01:52 As stricter internet regulations require a shift away from cookies to track user behavior online, contextual advertising continues to increase in popularity. Between 2022 and 2030, contextual advertising spending is predicted to grow at a rate of 13.8% annually. Let’s take a look at what contextual advertising is, how it can benefit marketers, and the best ways to execute a contextual advertising campaign. ## What Is Contextual Advertising? Contextual advertising is a type of programmatic advertising that uses the surrounding content, rather than user behavior, to determine ad placements. Contextual targeting is the process of finding the most relevant spots to place ads, while contextual advertising refers to the actual ad creative. ## How Does Contextual Advertising Work? Contextual advertising uses algorithms and automation to find the most relevant places for ads, based on content cues such as keywords, metadata, and the overall context of videos, images, social media feeds, or written editorial (or a combination of these types of content). By serving up highly relevant ads when a consumer is in the right frame of mind — i.e., consuming content related to your ads — marketers can boost conversion rates with lower ad spend. ## When to Use Contextual Advertising Contextual advertising works best when users are in the lower decision stages of the sales funnel — consideration and action — although you can deploy it any time you want to expand your reach and find new audiences. When you focus on full-funnel campaigns, you’ll find contextual advertising extremely effective at shortening the funnel. Reach users at the right time with the right content and they might make a buying decision in the moment, even if they hadn’t been considering your brand previously. You can use contextual advertising when you want to: - Avoid cookies to track behavior. - Reach people while they’re thinking about your type of product or service. - Boost conversions. - Attract new people. ## Advantages of Contextual Advertising ### Protect User Privacy Contextual advertising doesn’t rely on collecting demographic data from consumers, which means you can reach people on platforms that block cookies and other ad identifiers. It also works on non-addressable platforms like Firefox or Safari, per Amazon’s advertising library. Contextual advertising doesn’t care who your audience is, only that they are doing something related to your brand at that moment. ### Reach New Audiences Who Are Ready to Buy Contextual advertising reaches people when they’re considering companies like yours, by placing your ads amidst relevant content. Even if they’ve never heard of your specific brand previously, contextual advertising connects you with highly relevant, highly engaged consumers. ### Attract Buyers Interested in Brands Like Yours By targeting categories complementary to your brand, you can expand your reach and grow your audience. For instance, a travel accessory company, travel booking engine, or even a sporting goods company that sells snowboards might place an ad next to an article titled, “The 10 Best Suitcases for Your 2026 Winter Getaway.” ### Reduce Ad Spend and Achieve Greater Results You can boost your return on ad spend (RoAS) with contextual advertising. Rather than paying for the number of eyes on your ad, you only pay for results. This alone makes contextual advertising enticing to marketers. ## Considerations When Marketing Your Product with Contextual Advertising Artificial intelligence (AI)-driven contextual advertising can help take the guesswork out of your campaigns. But, success starts with a clear plan, relies on effective creative, and ends in measuring your results to refine the campaign. ### Set Goals Set clearly defined, measurable goals. How will you define success? Purchases? Clicks? Downloads? Opt-ins? How will you measure your results? It’s important to establish benchmarks from past campaigns, or at least have a reasonable idea of what a successful campaign will look like when it’s done. ### Choose Your Targeting Parameters AI makes it easier than ever to find the right audience for your ads, but you’ll still need to decide on the targeting parameters, e.g., keywords, categories, themes, user intent, and sentiment. ### Design Appropriate and Relevant Creative Here’s where it gets fun — and where your actions move the needle. Design relevant creative that engages your audience. Knowing your audience based on the content they consume might help. For instance, a luggage company placing an ad next to a listicle of best winter resorts would change up the creative based on whether the ad is appearing on a website for travel and leisure, budget travel, or even men’s health. ### Measure Results Keeping your KPIs from step one in mind, track your results. Avoid KPIs that aren’t tied to revenue or business growth, e.g., social media likes that don’t convert to action, and so won’t show the real value of your campaign. Conversions, including purchases or mailing list subscriptions, make a real impact. Other KPIs might include ROAS, revenue per email subscriber, and cost per acquisition. ### Optimize the Campaign If your creative and ad placements are working to deliver an acceptable ROAS, it’s time to double down. If they aren’t, consider switching up the creative and using A/B testing to find the right content. If your ads still aren’t working, consider changing your placement parameters to reach a different audience. For instance, you might shift from social media ads to programmatic advertising on top digital publisher sites. ## Types of Contextual Advertising As communication theorist Marshall McLuhan famously said, “The medium is the message.” For marketers engaged in contextual advertising today, this means that the format of your creative is as important as the content. Some audiences respond well to text, others to video, and others to dynamic graphics. Here are some formats of contextual advertising to consider. ### Text-based Ads Often used in paid search, AI-powered chatbots, and Google Ads, text-based ads work well with a simple, straightforward message. While generative AI platforms like ChatGPT, Gemini, and Perplexity don’t currently accept paid advertising, that day could come, as well. Brands that focus on generative engine optimization (GEO) garner mentions from top chat platforms. It’s only a matter of time before companies can purchase those mentions, bypassing the slower, organic route to visibility. ### In-game Ads Marketers seeking to target an audience that lives primarily on mobile devices might use in-game advertising. For instance, a skateboarding game might share ads for real skateboards or skate gear. A digital card game developer might advertise on other card games to draw players away from the competition. ### Video Ads Video ads appear on YouTube streams, social media, and at the bottom of websites where users frequently consume video-based content. ### Native Advertising Native advertising is designed to blend in with the editorial content on a site, even though it must be clearly marked as an ad. ## How to Run a Contextual Advertising Campaign Running a contextual advertising campaign depends on the platform you choose. They vary slightly in their capabilities and effectiveness, but the basics remain similar across platforms, including Google, Amazon, Realize, and others. ### Choose Your Platform Evaluate costs, ease-of-use, and capabilities for a variety of contextual advertising platforms. Some factors to consider include the number of topics, the options to advertise on different media sites or in apps, and the use of AI to discern the context of the content and target the right audiences. ### Choose Segments or Topics to Target Once you’ve signed up, choose the topics you want your audience to be engaged in or the type of articles you want them to be reading when they see your ads. ### Create and Upload Your Content Create content tailored to the audience. For instance, Gen Z dramatically prefers video content: While 69% of baby boomers said they prefer written content via email, 98% of Gen Z like video. Similarly, 91% of millennials and 78% of Gen X also prefer video content. Across the board, 89% of consumers engage with video more than other types of content, and video is the medium most likely to trigger a purchase in the US, per the study linked above. ### Measure Results Use the onboard tools to measure your results. Keep your KPIs, including conversion rates and ROAS, in mind, as you analyze the success. Advertisers and marketing agencies vary on their recommendations for how long to let an ad run: It could take days to dial in on the right audience for a social media campaign, for example. Advertising adjacent to the editorial content of major publishers could be faster, since you’re not relying exclusively on demographic data. The Nextdoor Business blog recommends three to six months for a campaign to fulfill long-term objectives, but you’ll want to consider seasonality, holidays, and even worldwide events that might affect the sales cycle. The more you spend on a campaign upfront, the faster you’ll start to see if it’s working or not. ### Optimize Your Campaign As you find the top-performing demographics, categories, or topics, optimize your campaign for those categories. That means putting more money into the most effective types of creative targeted at the most engaged audiences, and letting it roll until you see the start of diminished returns. That’s when it’s time to switch it up again, because you’ve saturated the market. ## Create Effective Landing Pages for Your Contextual Advertisements Ideally, your contextual advertising campaign will lead to high click-through rates and, ultimately, purchases or conversions. But, just because you’ve gotten your audience to click on your ad, it doesn’t mean your job as a marketer is over. Campaign success hinges on creating effective landing pages. Several elements must work together in your design and landing page copy: copy must address specific pain points of consumers, as people buy solutions, not products, and graphics must be relatable and scroll-stopping. Use A/B testing by changing just one element of your creative and running simultaneous ads to find the best combination of creative to boost your conversion rate. ### Consistency Between the Ad and Landing Page The transition from the ad that attracted a potential customer and the landing page that will ultimately convert them should not be jarring. It should have a similar style, colors, fonts, and messaging. ### Strong Headline Provide a clear, concise headline that shows people they’ve been directed to the right place immediately. ### Clean Design Any photos should be engaging, not distracting. Use authentic-looking photos of faces, or hands holding products, which helps to engender trust in your audience. ### Compelling Call-to-Action Finish with a compelling and clear call-to-action. Keep it simple while creating a sense of urgency with CTAs like: - “Claim your free now!” - “Buy now for free shipping” - “Save 20% today only” - “Join today for free” ## How to Measure the Effectiveness of Contextual Campaigns ### Return on Ad Spend How much money did you make compared to how much you spent? Return on ad spend (ROAS) is a powerful metric that shows exactly how much revenue your campaign generated. To calculate ROAS, divide the revenue generated by the cost of the ads. Unlike vanity metrics like impressions or click-through rate (which are still important in certain settings), ROAS tells you how well your campaign is doing in dollars. ### Cost per Acquisition Your cost per acquisition (CPA) is also a tangible and useful figure. It calculates how much it costs to acquire a new customer, regardless of how much that customer spends. This can be useful if the goal is a newsletter opt-in, rather than a purchase. CPA also helps you gauge how effective the campaign is in raising the visibility of your brand. For example, it might be more valuable to gain 500 customers each spending $1, than one customer spending $500, although the ROAS would be the same in both cases, especially if you’re building loyal customers with a high lifetime value (LTV). ### Click-through Rate Click-through rate, or CTR, shows that your ad creative is resonating with your audience. While clicks may not lead to revenue, they’re a solid start. ### Conversions Conversions show how many people took the desired action, whether that’s making a purchase, filling out a form, or opting in for a newsletter. While your conversion rate may not mean as much without looking at the ad spend behind it, you can use it as one metric amongst many to evaluate the success of an ad campaign. ### Impressions The number of ad impressions reveals how many people are seeing your ad, whether or not they act on it. If the number of impressions is too low, try expanding your topics or categories to serve the ad in more places. Think, too, about content that might be adjacent to your offerings. For instance, if your brand sells non-stick baking pans, you might place ads on a recipe website that shares healthy desserts. ### Brand Awareness You can use pre- and post-campaign surveys to gauge how well your ad campaign is increasing your brand’s visibility. You can also determine brand sentiment, or what consumers think of your brand, through polls. ## How Does AI Work with Contextual Advertising? We use AI in many areas of our lives today, from helping us create shopping lists or plan a trip to creating contextual advertising campaigns. AI algorithms can work to find the best content to place ads in real time by analyzing: - Content beyond keywords. - Meta data. - Themes. - User intent. - Contextually relevant moments in videos or placements in articles. ## Key Takeaways Contextual advertising targets audiences engaged in related content at a time when they are ready to take action. Rather than serving ads to people within a specific demographic, ad placements depend on the content surrounding the ad. By reaching bottom-of-the-funnel (BOFU) audience members, contextual advertising often leads to a higher ROAS and lower CPA. ## Frequently Asked Questions (FAQs) ### Contextual advertising vs. behavioral advertising: What’s the difference? Contextual advertising places ads based on the content of a specific web page, video, or app. Behavioral advertising relies on tracking user behavior to show ads to people who have visited specific websites or taken certain actions online. ### What is the future of contextual advertising? The future of contextual advertising will rely heavily on AI for more accurate targeting, gauging the full context of the editorial rather than relying solely on keywords or metadata. By using large language models (LLMs) to determine conversion intent (CI) for consumers engaged in various content, contextual advertising can become even more effective. ### Is contextual advertising better than interest-based targeting? Contextual advertising and interest-based targeting can both be effective marketing tactics. Contextual advertising shines when marketers can’t use cookies to collect consumer data. It also tends to reach users at a point when they are ready to make a buying decision, i.e., they’re already highly engaged in content related to brands like yours. Interest-based targeting may reach the right people at the wrong time. ### Does contextual targeting perform better post-cookie? Contextual targeting is an effective marketing solution as we move toward a cookieless internet. In the absence of user data, contextual advertising serves up highly relevant ads at the right time to people who are interested in what your brand has to offer. ### What are the best AI tools for contextual advertising? Performance advertising platform Realize is designed to boost ROAS, conversions, and CTRs by using sophisticated AI technology to determine context beyond keywords and find highly engaged consumers with an intent to buy. Google AdSense is another popular contextual advertising tool and incorporates AI features for optimization. Amazon also uses AI algorithms to deliver contextual ads within its platform. ### How do you choose keywords or categories for contextual targeting? Depending on your advertising platform, you can use its AI to help you find the best keywords and categories for your contextual targeting campaign. --- ### How to Safely Scale Performance Budgets With Automation URL: https://www.taboola.com/marketing-hub/scale-performance-budget-automation/ Last Modified: 2026-03-16 14:15:48 Managing performance budgets at scale is a constant balancing act. Growth requires higher spend, but each budget increase exposes your campaigns to additional risk. Unchecked bidding, runaway ad groups, or underperforming campaigns will drain your budget before you can even determine your return on investment. The most successful performance marketing teams know that sustainable scaling doesn’t mean monitoring your campaigns 24/7: It comes from automation, predictive intelligence, and built-in protections that keep your spend in check. Realize experts Lilly Valente, Alex Christensen, Torogia Stanton, and Emma Zimmerman share their use case insights with us. ## 3 Strategies to Effectively Manage Large-Scale Budgets Without Compromising Performance Gains ### Strategy 1: Delegate Optimization to Value-Focused AI Manual bidding breaks down the moment campaigns begin to scale. When you’re managing dozens or even hundreds of campaigns, real-time decision-making becomes challenging. Bids need to adjust dynamically based on context, competition, and intent, and those are conditions that constantly shift in today’s auction environments. That’s where performance-focused artificial intelligence (AI) can help. #### Why Manual Bidding Creates Risk at Scale Manual bidding introduces several challenges that are amplified when you try to expand your budget: - Delayed responses: Chances are, you aren’t sitting in front of your computer 24/7, monitoring activity. That makes it tough to respond to changes in real time. - Inconsistent performance: Fixed bids don’t adjust as user behavior changes, so performance can swing widely from one day to the next. It becomes almost impossible to maintain steady results when every adjustment relies on manual intervention. - Greater exposure to loss: As budgets grow, even small inefficiencies add up quickly. A single overbid or underperforming placement can drain thousands from your budget before you even realize it’s happening. To protect your budget and maintain ad performance, your decisions should be guided by real-time intelligence rather than set-and-forget inputs. #### Focus on Value, Not Just Clicks Fully automated bidding strategies use real-time signals to prioritize conversions over click volume. These models are continuously learning and adjusting, ensuring that spend flows to the highest-value opportunities. Value-focused bidding takes the guesswork out of cost decisions. When the algorithm is evaluating thousands of signals per auction, campaigns can reach new levels of consistency and scale without the volatility that comes from manual oversight. “The algorithm learns from the data it accumulates over time, so for at least the first seven to 10 days of your campaign — ideally 14 — it’s absolutely crucial not to start making changes. This is the period when the algorithm is figuring out which users are most likely to convert, based on the data you’re feeding it.” - Lilly Valente, Advertising Sales Manager, Realize While automation takes care of the heavy lifting, you remain in control because you’re the one putting guardrails in place. Two commonly used controls are: - Target cost per acquisition (tCPA): Guide the system toward conversions with an expected cost threshold, ensuring predictability as your budget increases over time. - Enhanced cost per click (CPC): This semi-automated approach adjusts manual bids based on conversion likelihood. These options blend automation with strategic oversight, letting you scale confidently while still managing risk. ### Strategy 2: Implement Automated Budget Shields and Protections Higher budgets demand a higher level of protection. The more money you invest in a campaign, the more vulnerable it becomes to waste, whether through underperforming sites, duplicate impressions, or ads that run well past their point of effectiveness. Automated protection layers act as a 24/7 safety net, ensuring budgets remain focused on supply and audiences that consistently convert. Here’s how always-on safeguards keep budgets focused on efficient supply and high-intent audiences. #### Reduce Wasted Spend Through Real-Time Budget Controls Instead of waiting for weekly reporting cycles to identify issues, use automated performance triggers to intervene the moment waste is detected. These protocols can: - Pause or limit spend on underperforming traffic sources.Redirects spend toward stronger supply. - Ensures budgets stay aligned with conversion efficiency. “You should look at SpendGuard as a safety net: If a site is underperforming, it automatically cuts spend, so you’re not throwing good money after bad.” - — Alex Christensen, Advertising Sales Manager, Realize #### Use Logic-Based Rules to Maintain Always-On Optimization Customized logic allows you to set specific "if-then" triggers based on live performance. This ensures your campaigns are being optimized even when your team is offline. Effective rules include: - Pause campaigns or creatives when CPA exceeds your threshold. - Shift budgets toward high-performing campaigns. - Reduce bids on segments that stop converting. - Increase spend only when your return on ad spend meets goals. “Using these rules is a lifesaver. You can set a threshold where, for example, if CPC or CPL climb to a certain number, it cuts spend across that part of the campaign, site, or channel. It’s all automatic — you don’t even have to look at it.” - — Alex Christensen, Advertising Sales Manager, Realize #### Prevent Ad Fatigue With Frequency Caps Ad fatigue is one of the most common — and preventable — drivers of wasted spend at scale. Frequency controls reduce the number of times an ad serves to an individual user. This helps you maintain your ad’s reach without overserving your audience. Frequency controls: - Reduce ineffective impressions. - Improve user experience. - Protect budgets from diminishing returns. - Extend the lifespan of creative assets. At scale, using frequency control is essential to protect your budget and get the most out of each campaign. “Applying automated frequency limits allows you to get your ads in front of customers without overserving them. These controls reduce ineffective impressions and protect budgets from diminishing returns.” - — Torogia Stanton, Advertising Sales Manager, Realize ### Strategy 3: De-Risking Future Spend With Predictive Insights Strong performance is only part of scaling effectively — you also need to understand what tomorrow’s performance will look like. Before increasing your budgets or altering your CPA goals, you’ll need visibility into how each change will affect conversions, cost, and efficiency. Predictive intelligence eliminates guesswork and provides a safe path to scale in some key ways. #### Establishing a Systematic Performance Feedback Loop Maintaining a disciplined approach to performance oversight ensures consistent visibility into how increased spend affects your ecosystem. By establishing a regular, methodical review of key metrics, you can identify trends over time (daily, weekly, or monthly) and isolate the specific data points needed to guide smarter optimization decisions. By implementing a rigorous cadence of data analysis, you can: - Anticipate Market Shifts: Spot emerging performance trends or volatility early, allowing for course correction before they impact the bottom line. - Isolate High-Impact Variables: Filter out noise to focus exclusively on the most relevant KPIs that indicate whether a campaign is healthy enough to support additional scale. - Audit Strategic Impact: Maintain a reliable record of cause and effect, understanding exactly how specific budget adjustments influence overall campaign health and efficiency. “To scale successfully, you have to move beyond manual data pulls and embrace a systematic review of your performance metrics. By establishing a consistent data cadence — whether daily or weekly — you gain the visibility needed to identify high-impact trends before they affect your bottom line. This structured approach allows you to filter out the noise and focus exclusively on the KPIs that matter, ensuring your optimization decisions are always guided by a reliable audit trail of how budget adjustments are influencing overall campaign health.” — Alex Christensen, Advertising Sales Manager, Realize #### Centralized Management Prevents Operational Errors Scaling often means monitoring dozens, or even hundreds, of campaigns. Using unified management workflows makes it possible to update settings across large sets of campaigns instantly, including: - Budgets. - Schedules. - Geo-targeting. - Site targeting. “These workflows let you edit most or all of your ads at once, which saves a ton of time. We have a UTM macro that will put custom ID names in the UTM field for every ad. If you do this at the campaign level, it will add it to every single ad automatically as a safeguard against potential data loss.” - — Emma Zimmerman, Solutions Engineer, Realize ## Key Takeaways Safely managing budgets at scale requires more than manual oversight and reactive optimization: You need a comprehensive system that includes automation, predictive intelligence, and control that protects spend while still powering growth. Successful performance teams rely on value-driven AI to optimize bids, while automated safeguards block inefficient placements and prevent unnecessary exposure as budgets grow. Predictive tools strengthen decision-making by helping you see expected outcomes before you make changes to your budget or targets. When combined with bulk management capabilities, you can scale confidently and efficiently, with sustainable results. ## Frequently Asked Questions (FAQs) ### 1. How can I stop budgets from being wasted on underperforming placements automatically? Automation is the best way to prevent your ad spend from flowing to sites or pages that fail to deliver meaningful results. These solutions continually evaluate how each placement is performing and automatically limit or block your budget from going to sources that aren’t performing. ### 2. How can I test budget increases or CPA changes before applying them live to see the estimated impact? Large budget changes always carry uncertainty, which is where a forecasting tool can make a big difference. This technology uses past data to simulate how your campaign will perform if you make tweaks to your budget levels or targets. This can give you a clearer picture of the impact to your conversions and efficiency before you change anything in the live environment. ### 3. How can I make large-scale changes (e.g., budget adjustments across hundreds of campaigns) without risking manual errors? When you’re managing a large campaign volume, manual updates can quickly become error-prone and time-consuming. Bulk editing tools solve this by letting you apply changes to budgets, schedules, geo-targeting, and site targeting across multiple campaigns at once — all within one streamlined workflow. --- ### Travel Marketing Trends 2026: How to Make Your Spend Go Further URL: https://www.taboola.com/marketing-hub/travel-marketing-trends/ Last Modified: 2026-03-02 08:53:19 Travel behavior continues to shift as technology, costs, and expectations evolve. From immersive experiences to AI-powered personalization, the travel landscape in 2026 demands more precise targeting, faster testing, and tighter alignment between campaigns and trip-planning behavior. Here are five key trends shaping travel marketing in 2026, to help your brand stay competitive and connected to today's travelers. ## Trend 1: The Continued Rise of Experiential Travel and Personalization Travelers in 2026 are not just booking trips, they’re building experiences around identity, interests, and values. Many travelers now view wellness retreats, literary-themed getaways, and cultural stays as investments in personal growth rather than escapes. More thoughtful travel is pushing travel marketers to move past generic packages toward curated, interest-based experiences. You can see the shift toward alternative accommodations and slower travel. Farm stays, cabins, and other rural rentals appeal to travelers who want to unplug and reconnect with nature. Major cultural and sporting events are also driving demand for marquee trips centered around a single experience, with travelers willing to pay for upgraded accommodations, premium seating, and tailored add-ons. Source: Kayak The real opportunity is to match people with personalized trips instead of interchangeable itineraries. Segment audiences by interests, intent, and values, then use first-party data to highlight the best combination of stays, activities, and experiences for each group in your ads and landing pages. You will see more conversions and repeat bookings when your creative reflects what travelers care about. Key stats to know: - 91% of travelers say they seek getaways focused on reading, relaxation, and quality time with loved ones. - 84% of travelers are interested in staying on or near a working farm at least once, and 57% of travelers say they are likely to attend a uniquely regional sporting experience while on a trip, per the same report. - About one in four travelers describe their ideal getaway as a mix of soft adventure and relaxation, and another 23% say they are adventure- and nature-seekers. ## Trend 2: The Dominance of Mobile and Seamless Digital Experiences in Travel Mobile is the default channel for most trip planning and booking in 2026. Many travelers now research options, compare prices, and complete bookings on their mobile devices. They expect fast, intuitive experiences that make it easy to move from inspiration to purchase in just a few taps. If it takes longer or the page lags, travelers are often likely to abandon the booking. Slow load times, clunky forms, required account creation, or missing payment options can cause a drop-off almost instantly. Mobile-first is now a practical requirement for travel marketers. It calls for landing pages optimized for small screens — clear calls to action, low-friction forms, and cross-device continuity. Analytics should focus on where mobile users abandon the process so you can quickly address friction points. Key stats to know: - 74% of travelers book flights through airline websites or apps. - 54% of lodging bookings are made directly through brand sites or apps, rather than third-party channels, per the same report. - 80% of Millennials and Gen Z respondents preferred using travel planning apps or social media to plan trips, and 66% typically download the travel apps they need before departure. ## Trend 3: Leveraging Influencer Marketing and User-Generated Content in Travel Influencer partnerships and user-generated content have become core tools in travel marketing. Travelers rely heavily on real people to show what a destination or experience feels like. Short-form videos, creator-hosted itineraries, and guest photos often outperform traditional brand ads, especially among younger audiences who are skeptical of polished campaigns. Expectations for authenticity are high: Travelers prefer creators who share authentic, realistic travel content, including budget information and honest reviews. In response, brands are building longer-term partnerships with niche creators who speak directly to specific segments and structuring campaigns so that creator content feeds into ad units, landing pages, and email flows. Marketers are also paying closer attention to the emotional impact of travel content. Constant comparison with polished social feeds can leave travelers feeling pressured and dissatisfied with their own trips. In response, marketers are moving toward more transparent messaging, clearer sponsored content labels, and stories that reflect a wider range of budgets, life stages, and travel styles. Think beyond one-off influencer posts: Build ongoing partnerships with creators whose audiences match your key segments. Set guidelines for creators that encourage transparency and authenticity rather than unrealistic highlight reels. Also, add simple prompts for guests to share photos and reviews you can reuse. Key stats to know: - 81% of Gen Z respondents use social media every day. - More than half of Gen Z report spending at least 3 hours per day on social platforms, according to the same survey. - Among Gen Z travelers, 59% look to Instagram, 54% to YouTube, and 47% to TikTok for trip inspiration. ## Trend 4: The Importance of Data and Analytics for Travel Marketing Insights In 2026, data and analytics are central to effective travel marketing. As privacy rules tighten and third-party cookies lose value, brands are relying more on first-party data from bookings, loyalty programs, site behavior, and in-trip interactions. The goal is to understand how people plan and experience travel, and then use those insights to tailor offers, content, and timing. Generative AI is reshaping how both travelers and marketers plan trips and campaigns. Travelers are experimenting with AI tools for destination research, itineraries, and recommendations. On the marketing side, AI can help segment audiences, predict demand, and surface relevant offers almost in real time. When those tools sit on solid data foundations, teams can move from broad targeting to more precise campaigns that focus spending on the segments and offers that convert. Source: Deloitte Travelers expect more control over how brands use their information, too. Clear consent, straightforward privacy language, and a focus on value exchange are critical. Brands that demonstrate how data drives better experiences and deals are more likely to earn the trust needed to build robust first-party profiles. Audit your data, tighten first-party segments, and experiment with AI-driven tools to improve forecasting and personalization. Start with a few clear use cases, such as improving conversion on high-intent landing pages or predicting repeat booking windows, measure the impact on revenue and efficiency, and then expand. Key stats to know: - Among travelers who used generative AI in 2025, 61% used it to research activities and attractions at their destinations. - 15% of travelers used generative AI in their trip planning, per the above report, up from 10% the previous year. - Industry data indicate a 64% annual increase in AI usage in travel. ## Trend 5: The Evolving Role of Social Media and Emerging Platforms in Travel Social platforms are often the starting point for many travel plans. Short-form video, live streams, and interactive content drive destination discovery as much as traditional search. Platforms are also adding more search tools, reviews, and booking features, enabling travelers to move directly from a clip or post into research and reservation flows. Social media functions as an inspiration engine, a reputation hub, and a performance channel. Travelers increasingly treat social feeds and search bars as hybrid search-and-review platforms, making relevance and recency especially important. Consistent content, creator collaborations, and structured user-generated campaigns are essential for staying visible. Social listening and community management also matter more, since comments and messages often reveal objections and questions before they show up in performance metrics. Marketers are also more aware of how travel content affects mental health and expectations. Many people feel pressure to match what they see online, which influences how they evaluate destinations and deals. Brands that acknowledge this reality and focus on honest, grounded storytelling in their social creative are better positioned to build trust, especially with younger audiences who are sensitive to inauthentic messaging. Treat social media as a full-funnel channel: Invest in video-led storytelling and build processes for sourcing and reusing user-generated content, then use listening tools to track shifts in sentiment so you can adjust creative, messaging, and offers before issues turn into reputation problems. Key stats to know: - Social media is the top media channel for Gen Z, with more than half spending at least 3 hours per day on social platforms. - A Pew Research Center survey showed that 84% of U.S. adults use YouTube, and 71% use Facebook. - 50% of adults report using Instagram, while 37% use TikTok, per the above study. ## Key Takeaways Relevance and trust drive travel marketing in 2026. Travelers are seeking personal, intentional experiences that align with their values. They're using mobile devices, social media, and AI tools to plan every step. Younger generations also expect more authenticity, a sharper focus on sustainability, and content that more closely reflects their own experiences. The path forward is to put audience insight at the center of strategy, then use mobile-first design, creator partnerships, smart data use, and honest storytelling to meet travelers where they already are. Teams that regularly test and adapt to these five trends tend to lower acquisition costs, grow direct revenue, and maintain healthy booking pipelines as travel behavior shifts. ## Frequently Asked Questions (FAQs) ### What are the most effective digital advertising channels for travel in 2026? The most effective channels are search, paid social, and your owned channels, such as your site, app, and email. Use search to capture high-intent demand, paid social to generate demand and retarget, and owned channels to convert and drive repeat bookings. ### How can travel brands build loyalty through digital marketing? Build loyalty with a clear program, consistent communication, and offers tied to booking and browsing behavior. Relevance turns one-time bookers into repeat customers. ### What are some successful examples of digital travel marketing campaigns? Jet2 turned an organic TikTok meme around its “Nothing Beats a Jet2holiday” audio into fuel for awareness, then leaned in with platform-native content and challenges to sustain momentum. Wyndham’s “Where There’s a Wyndham, There’s a Way” is a cleaner, brand-led example that unifies multiple hotel brands and Wyndham Rewards under one message and extends it across channels to drive bookings and loyalty. ### How is AI impacting the travel marketing landscape? AI is improving targeting and personalization, accelerating creative testing, and reshaping how travelers research and plan. That makes on-site assistance and smarter offers more important. ### What are the key metrics for measuring success in digital travel marketing? Prioritize cost per booking, revenue per campaign, repeat booking rate, and direct and branded traffic growth. You will see whether efforts are building profit, rather than just vanity metrics. --- ### How to Choose the Right Strategy for Campaign Budget Adjustments URL: https://www.taboola.com/marketing-hub/campaign-budget-strategy/ Last Modified: 2026-03-02 08:39:08 Budget adjustments are a constant part of a performance marketer’s daily work. Getting these adjustments right can mean better conversions, reach, and ultimately revenue, while miscalculating means wasted money that could have been put to better use elsewhere. Bulk budget adjustments and automated adjustments both offer trade-offs in terms of control, speed, and precision. Here, we explore both manual bulk budget adjustments and automated custom rules-based adjustments, and examine how each method helps to control spend and maximize performance. Understanding these differences will help as you’re building a budget strategy for your own performance marketing campaigns. Our Realize experts Melissa Stricker, Associate Advertising Account Manager, Malik Elijah Ward, Senior Advertising Sales Manager, and JeQuan Norris, Advertising Account Manager give us valuable insights from working with performance advertisers on the open web. ## When to Use Bulk Adjustments for Hands-On Control Performance marketers generally choose bulk budget adjustments when they’re making large, strategic shifts. For example, if you’ve been allocated extra budget to use up by the end of year, or you want to test ad copy variations uniformly, you can choose bulk edits. If the data shows that one particular audience segment is outperforming others, use bulk edits to adjust accordingly and improve ROI. It’s also a useful way to change campaign groups or other fields across the board. Bulk adjustments are ideal for speed, uniformity, and hands-on control for making large, strategic shifts. Here’s when to consider this option: ### Applying Uniform Changes Across Hundreds of Campaigns These bulk budget adjustments allow advertisers to apply the exact same change — such as increasing a daily spend limit — across many campaigns simultaneously in a single, error-reducing step. When you need to work quickly, this method is essential for macro-level management. "By making these changes all at once across your campaigns, you ensure your performance isn’t inhibited," says Taboola associate advertising account manager Melissa Stricker. ### Applying Scheduled and Immediate Spending Changes The bulk edit approach is best used for specific, time-bound events, such as an end-of-quarter budget push or reacting to a sudden competitive shift. An advertiser gets immediate, full control over the spend with bulk edits. Performance marketing platforms such as Realize allow users to modify up to 200 campaigns simultaneously with Bulk Edit. This reflects the sheer volume of programs that many performance marketers are managing, and the need to make continual tweaks efficiently. ## When to Use Custom Adjustments for Autonomous Efficiency Performance marketers use custom budget adjustment rules to add automation for speed and efficiency. Performance marketing teams juggle multiple campaigns reaching multiple audiences across formats and channels on the open web, which can easily become complex. Using automation saves time and leads to better outcomes, without depending on human intervention to take advantage of opportunities or pause a losing strategy. Custom, rules-based budget adjustments are a way to add continuous, conditional automation for both budget protection and scaling. Here’s when to consider this option: ### Implementing 24/7 Conditional Logic (Stop/Loss) When you set up custom rules, like in the Realize platform, they operate on predefined metrics (such as CPA) to automatically pause underperforming elements or adjust budgets. You can set the conditions for a pause or budget reduction depending on your audience, campaigns, and goals. The custom rules act as a continuous 24/7 safeguard that protects spend as soon as performance drops — something human teams cannot replicate. “I recommend implementing custom rules to safeguard your budget,” says Taboola senior advertising sales manager Malik Elijah Ward. “If, for example, a site spends 3-4x your daily budget with a CPA over $150 within a seven-day window, it should be blocked automatically. This ensures that even in your high-intent placements, you have protection in place that prevents wasted spend without requiring manual intervention.” ### Adding Automated Scaling for High-Performing Campaigns Custom budget adjustment rules are also essential to capturing campaign momentum and scaling up immediately. These rules detect when a campaign is exceeding performance goals (such as when CPA is below target) and automatically increase the budget to capture more conversions. It’s a hands-off way to scale high-ROI campaigns instantly, and doesn’t require human maintenance when performance starts going up. ## How to Choose the Right Tool: Bulk for the Macro, Rules for the Micro Each of these tools has its place, and performance marketing teams have to decide which to use, and when, for their particular goals. Budget management is part of a holistic marketing strategy that balances human input versus automation. Realize includes both, so marketers can make changes and establish rules all in one place. The most successful budget management strategy is the intelligent integration of both tools. Bulk is for the overall campaign portfolio management, while custom rules handle the continuous, granular, conditional optimization and risk management. To put it another way, bulk rules are for “what now,” while custom rules are for “what if/always.” “Platform tracking gives you great insight,” advises Taboola advertising account manager JeQuan Norris. “I recommend utilizing Custom Rules, which can either block specific elements, or adjust spend based on campaign performance, pushing scale in the moments you need it the most and decreasing spend if ever you see a drop in campaign efficiency. Integrating these rules allows you to optimize as efficiently as possible, ensuring you aren't just reacting to data, but staying ahead of it.” ## Key Takeaways As with performance marketing as a whole, budget management requires the use of the right tools at the right time to succeed. Teams need both bulk adjustments for macro-level changes and custom adjustments for continuous, micro-level optimization. Apply automation when you need continuous, conditional logic, and keep hands-on control for large, instantaneous shifts to strike the right balance between dynamic campaigns and advertising time costs. ## Frequently Asked Questions (FAQs) ### What is the main difference between manual and automated budget changes? Performance marketers work with speed and strategic insight to make sure ads are reaching the right audiences and driving conversions. Budget changes happen frequently, so both manual and automated budget changes play a role in a marketer’s toolkit. Manual, or bulk changes, allow for one immediate update across multiple campaigns at the same time, and users get the benefits of speed and uniform control. Automated changes, or custom rules, apply conditional logic so that budgets are constantly managed based on predefined performance metrics. Modern performance marketing platforms should include both options for flexibility. Realize includes Bulk Edits for users to update multiple campaigns immediately in one step, while Custom Rules runs 24/7 to automate budget management, continually adjusting budgets based on campaign results. ### When should I use the bulk method over automated rules? Using bulk budget adjustments versus automated rule budget adjustments will vary based on performance marketing needs and goals, along with campaign timing. Use bulk adjustments when you need to apply the same immediate change across many campaigns at once. That might be something like, “increase budget by 10%,” along with other scheduled events or large-scale strategic pivots. Within Realize, you can use Bulk Edit to apply immediate changes across up to 200 campaigns. Those changes include modifying budgets and schedules to drive stronger ROI. ### How do custom rules protect against wasted spend? Performance marketers can choose custom rules versus bulk adjustments to carefully protect spend even when they’re not at the computer to monitor performance. Custom, automated rules can be set up to monitor whatever performance metric is essential to the business, such as CPA or conversion rate. Then, the rule can be set to pause or decrease campaign budget when that number stops performing — or increase budget when performance climbs. Custom Rules, such as in Realize, automate campaign management by pausing underperforming ads and adjusting budgets based on campaign results, so performance marketers prevent wasted spend and only scale what’s working. “Instead of manually blocking underperforming sites every day, you should automate this process,” says Stricker. “You can set a rule so that if a site’s CPA hits a certain threshold — for example, over $200 — it’s automatically blocked. This protects your spend around the clock and keeps your focus on scaling what works.” --- ### What Is an Advertiser? Turning Digital Discovery Into Dollars URL: https://www.taboola.com/marketing-hub/advertiser/ Last Modified: 2026-03-19 13:36:32 In the digital world, we’re always being sold something, but who exactly is it that’s doing the selling? In 2026, being an advertiser is a mix of being part data scientist and part storyteller, navigating a chaotic ecosystem of algorithms to turn a casual scroller into a loyal customer. ## What Is an Advertiser? At its core, an advertiser is any brand or person paying to get their message in front of an audience. They represent the “demand side” of the digital economy. While the publisher provides the content, the advertiser provides the ads and the budget to make things grow. ## Types of Advertisers and Their Roles While all advertising may share the same basic end-goal, getting there looks different depending on which type of advertiser is doing it. ### Performance Advertisers For performance advertisers, every ad is a tiny scientific experiment with a clear hypothesis: “If I show this particular creative to this type of person, will they click?” Their entire role revolves around the bottom of the funnel, and they aren’t interested in vanity metrics like views or likes unless those numbers eventually end up at a purchase, a sign-up, or a qualified lead. Data rules everything around them, and they’re constantly tweaking headlines and images to squeeze every bit of value out of their budget. Because of this, modern performance advertising platforms use deep learning to analyze what users are reading on the open web, and help advertisers find discovery-mode audiences who are actually primed to take action, instead of just shouting into the endless void of social media. Performance advertisers use this technology to place their message in high-quality editorial environments where the user is already in a reading and learning mindset, making that final conversion much more likely. ### E-Commerce Advertisers If e-commerce advertisers are the digital storefront owners of the internet, then their primary mission is to turn window shoppers into buyers. They live in a fast-paced world of inventory updates, seasonal sales, and abandoned carts, but their real responsibility is to bridge the gap between a user’s screen and their doorstep. Most often, they’ll do this using dynamic product ads that show you exactly what you were looking at ten minutes ago. For these advertisers, success is measured by the return on ad spend (ROAS) and ensuring the cost to acquire a customer doesn’t eat up their entire profit margin. They’ll often leverage tools like carousel ads or social importers to showcase a variety of products in a single glance. By creating a frictionless path from a discovery click to a checkout button, e-commerce advertisers ensure that the open web acts as a 24/7 global mall. ### B2B and Lead Gen Advertisers While e-commerce is about the quick sale, B2B (business-to-business) and lead generation advertisers play more of a matchmaker role. Their goal isn’t necessarily an immediate credit card swipe; it’s a conversation. They might be offering a whitepaper, a free trial, or a consultation, and their role is to identify high-intent professionals who have a specific business problem, eventually enticing them to find out more. These advertisers focus heavily on targeting by job title, industry, or contextual signals. They use display and native ads to establish authority and trust, knowing that their sales cycle might take weeks or months. Success for them is measured in qualified leads — as in, real people with real budgets who are actually ready to talk to a sales team. ## Advertiser Goals and Objectives Performance advertisers don’t just want digital applause — they’re more focused on bottom-of-funnel metrics like purchases, qualified leads, and sign-ups. If an ad doesn’t move the needle on the balance sheet, it’s back to the drawing board to adjust it and see how to get more clicks. ### Building Brand Awareness This is all about making sure your name is familiar, so when a customer eventually needs what you’re selling, you're the first brand that clicks in their mind. ### Driving Sales and Conversions The heartbeat of performance marketing. Modern platforms use deep learning to find users in a discovery mindset, placing your product in front of them right when they’re actually primed to take action. ### Lead Generation and Customer Acquisition For service brands, it’s about gathering high-intent leads — people who fill out a form or request a quote. By targeting specific reading habits, you fill your sales funnel with people who actually need your service. ### Measuring Return on Ad Spend ROAS is the ultimate scorecard. If you spend $1 and make $5, you’re clearly winning. Performance advertisers live by this number, constantly tweaking their creative to ensure every dollar works as hard as possible. ### Budget Optimization and Performance Scaling Once you find a winning formula, the next step is to scale. AI-powered tools allow advertisers to automatically adjust bids in real-time, helping them reach millions of new users while keeping costs stable. ## Advertisers in the Digital Advertising Ecosystem The digital world is a massive web. That’s why advertisers use a specific tech stack to find their way through it. ### Advertisers and Ad Networks Think of an ad network as a broker. Instead of calling every news site individually, you can use a network to get instant access to thousands of premium sites in one place. ### Programmatic Advertising and DSPs Programmatic is the brain and engine of media buying. Advertisers use Demand-Side Platforms (DSPs) to set their parameters, then let software handle the auction process automatically and more efficiently than humans ever could alone. ### Private Marketplaces and Automated Bidding PMPs are invite-only auctions for premium ad spots. Advertisers use automated bidding to win these high-value placements in milliseconds, ensuring they never overpay for a click. ### Data-Driven Targeting and Audience Segmentation No more spray and pray. By using behavioral signals — like what someone is reading right now — ad platforms help you find users based on their current intent, rather than just their age or gender. ## Advertiser vs. Publisher: Key Differences These two need each other, but their day jobs are very, very different. ### The Goal (Selling vs. Monetizing) Advertisers want to move products, while publishers want to keep people engaged to earn revenue from that attention. ### The Flow of Capital Advertisers are the investors spending a budget, while publishers are the recipients using that money to fund their content. ### Inventory Management Advertisers manage creative inventory (the ads). Publishers manage digital real estate (where those ads sit on the page, or placements). ## Measuring Advertising Success Basically, if you can’t measure it, you shouldn’t be spending your money on it. Advertisers use data to prove that their creative is actually working, which gives a much better indication of where the best investment would be. ### Key Performance Indicators (KPIs) KPIs are the North Star for any campaign. For a performance advertiser, these are usually hard conversion goals. They serve as the primary evidence that a campaign is meeting its business objectives, and help teams stay aligned on what hitting those goals actually looks like. ### Engagement and Conversion Metrics While a click-through rate (CTR) tells you if your ad is interesting, conversion metrics tell you if it’s profitable. Advertisers track actions like “Add to Cart” or “Email Signup” to see exactly where users are dropping off in the journey. ### Attribution Modeling Approaches Attribution is about giving credit where it’s due. Whether it’s First Click or Last Click, advertisers use these models to figure out which specific ad actually convinced the customer to buy, helping them understand the true value of the open web. ### ROI and Cost Metrics (CPA, CPC, CPM) These are the unit economics of advertising. From the cost per thousand impressions (CPM) to the cost per actual customer (CPA), these metrics ensure the advertiser isn’t spending more to acquire a customer than that customer is worth. ## Key Takeaways Advertisers fund the free internet by connecting products with the right people. Performance advertisers are laser-focused on bottom-of-funnel results, i.e., sales and leads, and many use advanced modern performance advertising platforms to help simplify the open web and scale where social media can’t. ## Frequently Asked Questions (FAQs) ### How can advertisers effectively test and optimize creative content? Advertisers can A/B test multiple versions of creatives to see which designs, messaging, and calls-to-action perform best. Continuous testing and data-driven optimization help improve engagement and ROI. On the open web, this is a game where speed is everything. Using the right platform, performance advertisers can test dozens of headline and image combos simultaneously. The AI then automatically funnels your budget into the winners, ensuring your bottom-of-funnel goals are met without the tedious and exhausting manual guesswork. ### What role does audience segmentation play in campaign performance? Audience segmentation allows advertisers to target specific demographics and behaviors, making ads more relevant to people who would buy the product. This precision typically increases conversion rates as well as cost efficiency. For high-level performance, it’s all about intent: Modern platforms target users based on what they’re consuming in that current moment. If they’re reading about fitness, they see your gym supplement ad right then. This hits them in a discovery mindset, which is much more effective than catching them mid-scroll on social media. ### How can advertisers ensure brand safety while running digital campaigns? Advertisers maintain brand safety by using vetted networks and tools that block low-quality content. This is what reduces the risk of appearing alongside inappropriate material. That’s why, when scaling on the open web, sticking to premium environments is key. --- ### Retail Marketing Trends 2026: Understanding What Works URL: https://www.taboola.com/marketing-hub/retail-marketing-trends/ Last Modified: 2026-03-02 08:47:33 Retailers face slower growth, higher costs, and tougher competition in 2026. Shoppers are more price sensitive, journeys are more complex, and every channel is crowded. Online retail already represents trillions of dollars in annual sales and a growing share of total spending, so digital choices have a bigger impact on revenue and profit. The strongest retail brands connect channels, use data responsibly, and take sustainability seriously. They also give shoppers a straightforward path from discovery to purchase. Here are five retail marketing trends shaping how brands plan and measure campaigns in 2026. ## Trend 1: The Blurring Lines of Online and Offline Retail Many shopping journeys span multiple channels rather than remaining in one place. Consumers research on a phone, compare prices on a laptop, and complete the purchase wherever it’s most convenient. Retailers are responding by moving from separate e-commerce and store strategies to unified commerce. Unified commerce means a single inventory view, consistent offers, and a connected identity across web, app, and stores. Hybrid fulfillment options link online ordering with store pickup. Buy online, pick up in store, curbside pickup, and lockers let shoppers combine online convenience with store speed. Digital campaigns are major drivers of store traffic, using geotargeting, dynamic creative, and clear routes into store locators and pickup options. Mobile apps and loyalty programs help connect these touchpoints when data lives in a single system. Stats to know about omnichannel retail: - 96% of global retail executives expect industry revenues to grow in 2026. - Per the same source, 81% of those executives expect profit margins to expand in the year ahead. ## Trend 2: Personalization and Customer Data Platforms in Retail Marketing Retailers collect substantial first-party data from purchases, browsing, and loyalty programs. Turning that data into experiences that feel helpful without crossing privacy lines is the hard part. Shoppers expect brands to recognize them in simple, practical ways, and they notice when offers, recommendations, or loyalty treatment feel random. Customer data platforms (CDPs) aggregate transaction history, browsing behavior, and engagement signals into a single profile. Retailers then build more precise audiences for email, onsite tools, native campaigns, search, and social. They can also layer AI models on top to predict what someone is likely to need next. Retail teams are tightening consent flows, preference centers, and guardrails around how AI uses customer data so targeting gets smarter without eroding trust. What the numbers say about retail personalization: - 75% of shoppers say a consistent experience across retail websites, mobile apps, email, social media, and stores is important, but only 41% say brands deliver it today. - Per the same report, 69% of consumers want retailers to anticipate their needs with relevant offers or information at the right moment, yet only 35% think brands are doing so well. - The report also states that 87% of consumers expect retailers to handle their personal data responsibly and securely, while just 46% believe brands are doing so. - 45% of retailers report using generative AI to manage customer experiences, says the same study. ## Trend 3: The Continued Growth of E-commerce and Social Commerce in Retail Online sales remain a major part of retail growth, even as competition increases. Growth has slowed from past surges, but continues from a much higher base. Retailers are shifting from simply being online to making e-commerce profitable and predictable. Retail teams are tuning site speed and checkout flows, improving product content, and using marketplaces and retail media alongside owned sites. Social commerce also plays a big role: Shoppers discover, validate, and buy directly on platforms such as Instagram and TikTok, often without visiting a traditional e-commerce site. Live shopping and short-form product videos are part of this shift. Retailers are testing creator-hosted streams, shoppable videos, and vertical ad formats to shorten the path from discovery to checkout. Native on the open web remains important for capturing demand beyond closed ecosystems and driving qualified traffic to optimized product pages and content. Source: Adobe E-commerce and social retail by the numbers: - U.S. shoppers spent $257.8 billion online during the 2025 holiday season from November 1 to December 31, an increase of 6.8% year over year. - Consumers spent more than $4 billion online in a single day, on 25 different days during the 2025 holiday season, up from 18 such days in 2024, per the same report. - The report adds that mobile shopping accounted for 56.4% of online transactions during the 2025 holiday season, and 66.5% of online sales on Christmas Day. - Traffic to retail sites from generative AI tools during the 2025 holiday season was up 693.4% compared with the prior year, according to the report linked above. ## Trend 4: The Importance of Sustainability and Ethical Practices in Retail Marketing Sustainability has become a key factor for many shoppers, especially in fashion and beauty. People want clear information: They want to know what materials you use, how you source, how you package, and whether you’re improving. Source: Shorr Retailers that take this seriously are tying marketing to real work in product development and operations and publishing clearer impact numbers. They’re also testing circular models like resale, take back, and repair programs. The challenge is credibility: Shoppers are willing to pay more for sustainable products, but they don’t automatically believe brands’ claims. Brands earn trust when they share clear information and real examples. Digital campaigns can support this by using native articles, video, and sponsored content to explain how sourcing, packaging, and community work are changing. Key numbers on ethics and sustainability in retail: - 54% of consumers report consciously purchasing products with sustainable packaging in the last six months. - 90% of consumers say they are more likely to buy from brands that use sustainable packaging, per the same report, and 43% of consumers are willing to pay more for a product with sustainable packaging. - The report adds that 39% of consumers report switching to a competing brand because that brand offered more sustainable packaging. ## Trend 5: Leveraging Augmented Reality and Virtual Reality in Retail Augmented reality (AR) and virtual reality (VR) are shifting from novelties to useful tools that help shoppers decide what to buy. AR fits naturally into mobile-first journeys, letting people see how a sofa looks in their living room or how glasses fit their face, either at home or in a store. Retailers are using AR for try-ons, room visualization, and richer product information on shelves and packaging. Good AR experiences mean fewer surprises after delivery and fewer returns in categories like fashion and furniture. Most VR use in retail still centers on high-involvement purchases and brand experiences. Broader AR market forecasts and consumer adoption data indicate retailers are testing these tools now rather than waiting. For advertisers, AR and VR show up in shoppable lenses, 3D product previews on mobile landing pages, and interactive ad units that let people try or place products before they click through. These formats can increase time spent with a product and help nudge people closer to purchase. Stats to know about AR and VR in retail: - The global AR in retail market is projected to grow from $7.84 billion in 2024 to $105.87 billion by 2033, at a compound annual growth rate of 32.4%. - In recent retail AR surveys, about 61% of consumers prefer retailers that offer AR experiences. - In the same research, nearly half of consumers say AR content on packaging would increase their loyalty to a brand. ## Key Takeaways Retail marketers need to connect in-store and online experiences, use first-party data more effectively, and keep e-commerce and social commerce profitable. Shoppers also weigh how brands handle sustainability and how helpful tools like AR and VR feel in practice. Strong brands treat these trends as part of a single plan and use them together to make shopping simpler and more useful. ## Frequently Asked Questions (FAQs) ### What are the most effective digital advertising channels for retail in 2026? Retailers see better performance by combining search and shopping ads for intent with social, native, and retail media for discovery. Those channels work best when they drive people to clear product pages or store visits. ### How can retailers combat online competition? Skip the price war. Improve the experience, offer convenient pickup and delivery, invest in honest product content and AR tools, and use loyalty programs that reward repeat buyers. ### What are some successful examples of retail digital marketing campaigns? American Eagle’s “good jeans” campaign with Sydney Sweeney ran across social, video, and stores, generated tens of billions of impressions, sold out key items, and helped lift the company’s stock after strong earnings. E.l.f. Beauty’s “Give an e.l.f.” campaign tied bold creative and high-visibility out-of-home placements to its impact report and multimillion-dollar giving, reinforcing the brand’s values and supporting continued sales and investor momentum. ### How is AI impacting the retail marketing landscape? AI already supports targeting, bidding, recommendations, creative testing, chatbots, forecasting, and pricing. Teams get the most value when they share those insights across channels and set clear rules for data use and privacy. ### What are the key metrics for measuring success in retail digital marketing? Track revenue, margin, and profit by channel, along with customer acquisition cost versus lifetime value. Teams should also watch conversion rate, average order value, returns, key engagement metrics, and, for omnichannel retailers, store traffic and offline sales linked back to digital campaigns. --- ### Mobile Marketing Trends 2026: What's Working, and Why URL: https://www.taboola.com/marketing-hub/mobile-marketing-trends/ Last Modified: 2026-03-02 08:49:36 Mobile is the primary channel through which customers experience your brand, often long before they see it anywhere else. Strong mobile campaigns start with understanding how people scroll, search, and buy on their phones, rather than treating mobile as simply a smaller version of desktop. Here are five trends shaping mobile attention in 2026, and what they mean for your marketing strategy. ## Trend 1: The Primacy of Mobile-First Experiences and Optimization Mobile shoppers abandon sites quickly when design and performance hinder their experience. Slow load times, crowded layouts, or tiny tap targets give people a reason to leave and choose a smoother site. That’s why a mobile-first approach starts with design and performance. Teams should plan layouts, navigation, and forms for small screens first, then adapt them for desktop. Menus stay simple, tap targets stay large, calls to action stay clear, and pages load quickly with lean code and compressed assets. Source: Web Almanac Mobile SEO has become baseline SEO: If your pages are not fast, responsive, and easy to crawl on mobile, it will affect how you show up in rankings, as well as in how often your content appears in features such as carousels and rich results. Voice searches and on-the-go queries only raise the bar further. Stats on mobile usage and performance: - In 2025, consumers spent roughly 5.3 trillion hours in apps on iOS and Google Play, up nearly 4% year over year. - The average smartphone user spends over 3.6 hours per day in apps, across 34 apps per month. - 62% of mobile pages achieved a “good” Largest Contentful Paint score in 2025, compared with 74% of desktop pages. ## Trend 2: Leveraging AI and Personalization in Mobile Marketing Marketing teams lean on artificial intelligence (AI) to plan, create, and adjust campaigns. In mobile channels, it supports ad targeting, recommendations, in-app experiences, and chat-based support. AI also helps tailor mobile experiences in real time. Brands can adjust creative by audience, device, and context, surface products based on recent browsing, and time push notifications to when people are most likely to respond. Predictive analytics can flag churn risk, suggest next-best offers, and guide retention efforts. Dynamic creative optimization can automatically swap in the strongest assets, while responsibly-used location signals help tailor offers to where people actually are. Marketers are relying more than ever on platforms that can test multiple creative variations simultaneously, import social assets into open web campaigns, and let algorithms push spend toward what performs well on mobile. People notice when targeting feels intrusive, or when data use is hard to understand: Overly aggressive tactics and unclear data practices erode trust, so teams that use AI to make experiences more useful, and explain in plain language how they use data, tend to see better long-term results. Key stats about AI and personalization: - In a 2026 marketing survey, about 94% of marketers said they plan to use AI in content creation. - The same survey shows that 45% of marketers use smart image-editing tools, 44% use video or animation generation tools, 42% use smart video or audio editing, and 40% use image or design editors. - The study also makes it clear that nearly 75% of marketers already use AI to create media such as images and videos. ## Trend 3: The Continued Dominance of Video and Immersive Content on Mobile Short-form video has become the main format people watch on their phones. Vertical clips match how people actually use their devices, with quick sessions, optional sound, and heavy scrolling. Augmented reality (AR) and other immersive tools sit alongside that feed and give people ways to try products, rather than just watch them. Shoppers’ mobile journeys often start with short-form social video, before moving into deeper research and comparison. TikTok, Reels, and Shorts all reward content that hooks in the first seconds, uses captions, and feels native to the feed. Brands need focused messages, strong visuals, and landing experiences that match the clip. Retailers can use AR and interactive formats to help people picture products in their own space. Virtual try-ons, room visualizers, and shoppable videos help people to understand and test products without leaving the app. Interactive and gamified units, such as quizzes or playable videos, extend that behavior and keep people engaged longer than static placements. User-generated video and creator partnerships often carry more weight than polished brand spots and can be reused across ads, landing pages, and in-app modules. Mobile video and AR by the numbers: - In Q3 of 2025, Snapchat had 477 million daily active users and 943 million monthly active users. - Per the same source, more than 350 million people used Snapchat’s augmented reality experiences daily during the third quarter of 2025. - In the same period, users played with AR Lenses approximately 8 billion times per day. ## Trend 4: The Rise of Mobile Commerce and Seamless Payment Options Shoppers often discover products, compare options, and check out without ever leaving their phones. When the path feels simple and trustworthy, they complete the purchase on that same screen. Shoppers expect digital wallets and alternative payment options in mobile checkout flows. Apple Pay, Google Pay, PayPal, and local wallets reduce friction by removing manual card entry. BNPL, subscriptions, and saved payment details streamline repeat purchases. Retailers use owned mobile apps to store credentials, run loyalty programs, save preferences, and send push notifications that bring people back to shop again. Social commerce and in-app stores pull even more shopping activity onto phones. Shoppable posts, live shopping, and native storefronts on platforms like Instagram, TikTok, and YouTube allow users to move from inspiration to purchase without leaving the environment. To maintain high conversion rates, brands need clean mobile carts, guest checkout, clear shipping and fee breakdowns, and consistent experiences across mobile web and apps. Key mobile commerce stats:   - From November 1 to December 31, 2025, U.S. online shoppers spent $257.8 billion, up 6.8% from the 2024 holiday season. - During the 2025 holiday period, smartphones accounted for 56.4% of U.S. online transactions. - On Christmas Day 2025, smartphones accounted for 66.5% of online transactions, and on Thanksgiving Day, 61.6%. ## Trend 5: Navigating Privacy Changes and Building Trust in Mobile Marketing Marketing plans must account for what data is collected, how long it is retained, and who can access it. Tracking limits, new privacy laws, and rising user expectations are reshaping how mobile data is collected and used. Regulation is tightening as well. As of late 2025, roughly 20 U.S. states have comprehensive privacy laws in place or on the way, with several more expected to take effect in 2026 and beyond. New rules spell out which companies are covered, how teams must manage consent, and what rights consumers have over their data. Teams rely more on first-party data, clear consent, and privacy-respecting approaches. They use clearer language in consent flows, offer more granular controls, rely more on contextual and cohort-based targeting, and rely less on one-to-one tracking and more on modeled conversions and incrementality tests. Performance partners can rely on predictive, intent-based audiences and modeled conversions, rather than identity-based tracking, to maintain measurement as privacy tightens. Teams that explain what they collect, why they collect it, and how it improves the experience make it easier for people to say yes. Source: Adjust Privacy and consent stats to know: - By Q2 2025, the average app tracking transparency opt-in rate reached 35%, up from 34.5% in Q2 2024 and 34% in Q2 2023. - In 2025, gaming apps saw higher consent rates, per the same source, with opt-in rates of 50% for sports titles, 43% for hyper-casual games, and 40% for action games. ## Key Takeaways Mobile is the primary channel for discovering and buying from brands, so mobile UX and performance come first. AI and personalization support most mobile programs when teams use them transparently to improve relevance. Short-form video and immersive formats, such as AR, capture most of the attention on phones and belong in the core content plan. Phones have become a primary checkout channel, so clean carts, clear pricing, and flexible payment options directly drive revenue. Tighter privacy rules and higher user expectations make clear consent, first-party data, and privacy-safe measurement basic requirements for mobile marketing. ## Frequently Asked Questions (FAQs) ### What are the most effective mobile advertising formats in 2026? In-feed native ads, vertical short-form video, shoppable placements, and app campaigns tend to work best on mobile. Keep them fast, platform-native, and focused on simple mobile landing pages, using rewarded or playable formats only when the value exchange is clear. ### How can brands measure the ROI of their mobile marketing efforts? Tie metrics to clear goals. For acquisition, track installs, signups, or leads. For revenue, track mobile conversion rate, average order value, and revenue by device. Use incrementality tests or lift studies to see what mobile actually adds. ### What are some common mistakes to avoid in mobile marketing? Common issues include shrinking desktop layouts onto phones, overusing pop-ups, sending generic push notifications, and driving traffic to slow or cluttered landing pages. Ignoring consent preferences or hiding privacy controls also hurts trust and long-term performance. ### How is 5G technology impacting mobile marketing opportunities? 5G improves speed and latency where coverage is strong, which helps with high-quality streaming, live video, and responsive AR or interactive content. It enables richer experiences, but does not eliminate the need for performance-friendly pages and apps. ### What are the key performance indicators (KPIs) for successful mobile marketing? Useful KPIs include mobile traffic share, app installs, active users, in-app or on-site time, and engagement metrics such as taps and video completions. On the performance side, track conversion rate by device, cost per acquisition, revenue per user, and retention or churn. Over time, connect these to customer lifetime value to assess whether mobile efforts are attracting the right customers. --- ### Predictive Targeting vs. Contextual Targeting: Which Is Better? URL: https://www.taboola.com/marketing-hub/predictive-targeting-vs-contextual-targeting/ Last Modified: 2026-02-24 12:45:21 As privacy regulations become stricter and user-level tracking becomes more limited, advertisers are rethinking how they reach the right people at the right time. Two approaches have become especially important in this changing environment: predictive targeting and contextual targeting. While both reduce reliance on traditional tracking methods, they don’t work in the same ways. Each approach identifies intent differently, uses different signals to optimize delivery, and scales performance differently, too. As an advertiser, understanding what makes each approach unique is critical to maintaining performance without compromising privacy. ## Predictive Targeting Predictive targeting uses data-driven models to estimate which people are most likely to take action, like clicking an ad, making a purchase, or installing an app. Instead of relying on fixed audience definitions, it uses real-time signals to anticipate what someone is likely to do next. In other words, rather than focusing on who users are, this approach focuses on what they’re likely to do. ### How It Works Predictive targeting analyzes large amounts of de-identified data, including how people engage with content, the context they are in, device information, and past campaign performance. These signals are used to estimate the likelihood that a specific ad impression will lead to a desired outcome. Predictive targeting continuously learns from active campaigns. As new performance data comes in, delivery automatically shifts toward the users, placements, and moments most likely to convert, without requiring advertisers to manually manage audience segments. ### Benefits One of the main strengths of predictive targeting is its ability to adapt quickly. Because optimization is based on live performance data, it can respond to behavior changes due to seasonality, shifts in interest, or creative performance. Predictive targeting also makes scaling easier. Since it doesn’t depend on predefined audiences or persistent identifiers, it can surface high-intent users that traditional targeting methods may miss. This makes it especially effective in privacy-focused environments. ### Considerations Predictive targeting usually requires a learning period. Campaigns often require sufficient time and scale to generate reliable signals before performance stabilizes. Creative quality also matters: Clear messaging, strong value propositions, and clear calls to action help the system recognize meaningful engagement more quickly. ### Use Cases Predictive targeting works well for performance-focused campaigns at scale, including e-commerce acquisition, app installs, and lead generation. It is especially useful when advertisers want to prioritize outcomes rather than audience definitions. It’s also a strong option for brands operating in regulated industries or regions with strict privacy requirements. ## Contextual Targeting Contextual targeting shows ads based on the content being viewed, rather than on user behavior or identity. Ads are placed next to articles, videos, or pages that align with the advertiser’s message, category, or intent signals. This approach emphasizes relevance in the moment, helping ads appear alongside content that feels appropriate and timely. ### How It Works Contextual targeting evaluates signals from the page itself, such as topics, keywords, sentiment, and overall structure, to determine where to place ads. Ads are then served in environments that match the advertiser’s goals and brand standards. Modern contextual solutions go beyond simple keyword matching: They use language analysis to understand content meaning and tone at scale, enabling more accurate and consistent ad placement. ### Benefits Contextual targeting is naturally privacy compliant. Because it does not rely on past behavior or user identifiers, such as an email address or device ID, it aligns well with strict privacy regulations. It also provides strong brand safety and suitability controls, since advertisers can ensure their ads appear alongside relevant, high-quality content that supports brand perception rather than undermines it. ### Considerations While contextual targeting offers strong relevance, it does not automatically optimize for conversion likelihood. Performance can vary depending on how closely the surrounding content aligns with buying intent. Scale can also depend on the availability of suitable content within specific topics or categories, which may be limiting for more niche advertisers. ### Use Cases Contextual targeting is commonly used for brand awareness and consideration campaigns. It’s well-suited for advertisers who prioritize brand alignment, message reinforcement, and privacy compliance. It also performs well in regulated industries such as finance, healthcare, and education, where user tracking may be limited. ## How Does Predictive Targeting Compare to Contextual Targeting? Predictive Targeting Contextual Targeting Privacy Compliance Uses anonymized, aggregate signals without persistent user tracking Fully privacy-safe with no reliance on user data Campaign Goal  Optimized for performance and conversions Focused on relevance and message alignment Setup Complexity  Minimal setup with automated optimization Requires content and suitability parameters Audience Scalability  Highly scalable through continuous learning Limited by content availability and scope Immediate Performance  Improves as models learn Delivers stable, predictable reach early Long-Term Brand Lift  Supports sustained growth through discovery Strong brand association and reinforcement Cost Efficiency  Increases over time with optimization Consistent but less performance-driven Automated AI Integrations Fully leverages AI-driven optimization in platforms like Realize Uses AI for content analysis and matching Brand Safety/Suitability  Balanced through placement-level controls Strong by design A/B Testing  Ideal for creative and performance testing at scale Effective for message-context alignment tests ## How to Decide When to Choose Predictive Targeting Predictive targeting is a strong choice when you need performance and scale. If your goal is to drive conversions efficiently while still meeting evolving privacy requirements, predictive models offer a clear advantage. In Realize, predictive targeting allows campaigns to adjust automatically based on real engagement signals, reducing the need for hands-on optimization. This makes it especially useful when advertisers want to move beyond content alignment and focus on measurable results. Predictive targeting is also well-suited for testing new markets, creative approaches, or offers, when historical audience data is limited or unavailable. ## How to Decide When to Choose Contextual Targeting Contextual targeting is a good fit when privacy, brand safety, and message relevance are priorities. It offers a reliable, compliant option for organizations operating under strict data policies or choosing not to rely on user tracking. It works particularly well for awareness and consideration campaigns, where placing ads alongside relevant content helps build trust and engagement. Contextual targeting can also support performance goals when used as a foundation and paired with additional optimization methods. ## Key Takeaways Predictive and contextual targeting both solve different problems for advertisers. Predictive targeting focuses on driving performance by learning from engagement and continuously optimizing toward outcomes. Contextual targeting prioritizes relevance and compliance by matching ads to appropriate content environments. Your most effective strategy might be to combine both methods. You can use contextual signals to help establish relevance, and predictive models to improve results over time. Platforms like Realize are built to support this balance, allowing advertisers to align privacy, performance, and scale. ## Frequently Asked Questions (FAQs) ### We are a brand with strict privacy standards and do not allow user tracking. I can’t use first-party data, so what is my best performance option? When it comes to privacy, contextual targeting is the safest option, as it doesn’t rely on user data. However, predictive targeting can also support performance goals by using anonymized, aggregate signals rather than individual tracking. ### We are a high-frequency consumables brand looking to time our ads. When is the best time to show my ad to a returning customer? Predictive targeting is better suited for timing optimization, as it analyzes engagement patterns and delivery moments to identify when users are most likely to act, rather than relying solely on content context. ### Which method “learns” faster, predictive or contextual targeting? Predictive targeting generally learns faster to improve performance because it continuously optimizes based on conversion signals. Contextual targeting is more static, learning primarily through content classification rather than outcome-driven feedback loops. --- ### The Importance of Trust and Human Connection for AI-Driven Performance Campaigns URL: https://www.taboola.com/marketing-hub/trust-for-ai-performance-success/ Last Modified: 2026-01-28 08:09:50 In performance advertising, generative AI has moved from an experimental tool to a core capability. In the past, creative production took days or weeks, but it can now happen in minutes at a scale previously unheard of. However, as consumers see more and more AI-generated content, a critical question is emerging: Can you build the same level of trust and emotional connection with machine-generated creative as with work shaped by human hands? Recent research suggests that this question may be framed incorrectly. The real issue is not whether AI can perform, but whether audiences perceive AI-generated creative as authentic. When ads feel artificial or overly synthetic, engagement declines. When AI is used to amplify human-centric creative principles, however, performance does not merely hold steady — it can improve. As it turns out, trust is not a soft brand value in AI-driven campaigns, it’s a measurable performance constraint. ## The Evolution of AI in Performance Advertising The role of AI in performance advertising has expanded in stages, moving from backend efficiency tools to a highly visible force in creative production. Each phase has increased AI’s impact on results, while also raising new questions about how audiences perceive and respond to machine-generated content. Here’s a quick look at the evolution in performance advertising: ### Stage 1: Efficiency & Optimization Initially, AI operated almost entirely behind the scenes. It was primarily used to manage bids, segment audiences, and allocate budgets in real time. These systems improved campaign efficiency and reduced marketers’ manual workload, but they remained invisible to consumers. Because AI did not directly shape what audiences saw, there was little risk to brand trust or creative authenticity. ### Stage 2: Content Proliferation As generative models matured, AI moved closer to the creative layer. Tools such as Realize’s GenAI Ad Maker enabled the generation of thousands of visual and textual variations at virtually no additional cost. This shift democratized high-volume creative testing, allowing advertisers to experiment at a scale that was previously reserved for the largest budgets. For performance marketers, AI became a powerful engine for rapid iteration and optimization, significantly accelerating learning cycles. ### Stage 3: The Authenticity Era (Current) Today, AI sits at the forefront of creative production, generating assets that consumers actively see and judge. As a result, research attention has shifted toward “perceived artificiality” and its effect on engagement. The industry is increasingly recognizing that while AI can calculate, optimize, and scale with remarkable efficiency, emotional connection still depends on human-centric creative cues. In this phase, authenticity is no longer a creative preference but a performance requirement. ## Key Findings: Why Human Connection Wins Large-scale academic research has begun to clarify why human connection remains central to the success of AI-driven advertising. In a recent study led by teams from Columbia University, Harvard University, the Technical University of Munich, and Carnegie Mellon, researchers analyzed live performance data for ads created with Realize’s GenAI Ad Maker. Working in collaboration with Taboola’s in-house creative agency, Creative Shop, the research examined how AI-generated and human-made ads perform side by side across real advertisers, real audiences, and real consumer behavior. Instead of relying on surveys or controlled lab simulations, the study drew on in-market campaign data. This allowed researchers to observe how perceptions of authenticity influence engagement at scale. This real-world approach makes the findings particularly relevant for performance marketers, where creative decisions are ultimately judged by how audiences actually respond, not how they say they might. Here is a summary of some key findings: ### 1. The “AI Penalty” vs. the Disguise Advantage One consistent finding across the research is the presence of what behavioral scientists describe as “algorithm aversion.” Consumers often bring a negative predisposition toward content they believe was generated by a machine, particularly when it appears overly polished or formulaic. This reaction can suppress engagement even when the underlying message is relevant. Overall, AI-generated ads tend to perform on par with human-made ads in terms of click-through rate (CTR). That parity alone is notable given the speed and efficiency advantages AI provides. However, averages obscure an important pattern: When researchers examined performance through the lens of perception, a clear hierarchy emerged. AI-generated ads that were not perceived as artificial achieved the highest engagement rates, outperforming both obviously AI-made ads and traditional human-made creative. Side-by-side comparisons from live campaigns show how closely AI-generated and human-made ads can perform when they follow the same creative principles. In fact, in the following matched examples (see image below), the AI-generated ads achieved slightly higher CTRs without appearing artificial to viewers. The Columbia study backs this up. Across the dataset, which included over 300,000 ads, human-generated ads had a CTR of around 0.65%, while AI-generated ads had a CTR of around 0.76%. The inverse was also true. Ads that felt synthetic were penalized by audiences, regardless of whether they were actually created by AI or by humans. In practical terms, this means performance is governed less by how an ad is made and more by how it is perceived. AI’s advantage lies in its ability to convincingly blend into the visual and emotional language audiences associate with human communication. ### 2. Faces as the Bridge to Trust Among all the variables examined in recent advertising research, few are as consistently powerful as the presence of human faces. From an evolutionary perspective, humans are biologically wired to quickly notice faces and draw emotional information from them. Faces signal intention, relatability, and social presence, all of which help reduce uncertainty in fast-moving digital environments. In performance advertising, this biological bias becomes a functional advantage. Ads featuring clear, prominent human faces are more likely to be perceived as authentic and human-made, which in turn supports stronger engagement. Research examining visual attributes confirms that facial presence is one of the strongest predictors of whether an ad feels trustworthy. Global clothing brand H&M understands this. In 2025, they replaced some of their fashion models with AI-generated versions. However, to ensure the digital models “reflected the real model’s individuality,” they worked with AI specialists and creative teams to design the AI models and build the campaign narrative. This was done to ensure that the technology wasn’t taking away from human artistry, but building upon it. Interestingly, AI-generated creative often includes human faces more frequently than human-designed ads. This is not an accident: AI systems trained on large volumes of high-performing creative tend to replicate the patterns that already work, including the consistent use of faces to establish emotional connection. When applied thoughtfully, this can help AI-generated ads overcome skepticism rather than amplify it. ### 3. Visual Cues of Artificiality While faces help bridge trust gaps, certain visual characteristics reliably signal artificiality and trigger disengagement. Ads that rely on extreme sharpness, heavy color saturation, or highly symmetrical compositions are more likely to be perceived as machine-generated. These traits can create a sense of visual perfection that feels detached from real-world experience. By contrast, ads that incorporate warmth, natural composition, and subtle imperfection tend to feel more human. High realism, rather than hyper-polish, supports relatability. In this context, the goal is not to disguise AI through deception, but to align creative output with the visual cues audiences already associate with authenticity. Avoiding these perceptual markers is often enough to preserve trust and maintain performance. ## Key Takeaways: Trust as a Performance Metric Generative AI is reshaping how performance advertising creative is produced, but it has not rewritten the fundamentals of human psychology. Audiences still respond to authenticity, familiarity, and emotional clarity. The results of the Columbia study, along with other research, show that AI can support these outcomes rather than undermine them, provided it is guided by human-centric design. For advertisers, this reframes the AI debate. The choice is not between human creativity and machine efficiency, but between scaling distrust or scaling connection. Authentic faces, relatable contexts, and emotionally grounded visuals are no longer optional refinements: They are prerequisites for sustainable performance. ## Frequently Asked Questions (FAQs) ### Does using AI in ad creative damage consumer trust? Consumer skepticism toward AI-generated content is real, but it is not absolute. Research shows that distrust tends to arise when ad creative feels robotic, overly perfect, or detached from human experience. According to Frontiers in Psychology, in some functional contexts, such as data-driven messaging or informational content, transparency about AI use can even reinforce perceptions of competence. Problems tend to emerge when emotionally or narrative-driven creative lacks human nuance, resulting in what many marketers now refer to as “AI slop.” In other words, telling people that content was created by AI changes how they react to it. When the content is practical or informational, being upfront about AI can actually help, as people see it as efficient and objective. In fact, a 2024 Sprout Social survey found that 94% of consumers believe all AI content should be disclosed. But, when the content is meant to entertain, inspire, or connect emotionally, revealing that it was made with AI can backfire, making it feel less genuine or engaging. So, knowing when and when not to disclose is critical, but you can also mitigate AI distrust with Realize by using its GenAI Ad Maker tool to prioritize human-centric signals. Research that analyzed large volumes of live ads found that a significant share of AI-generated creative was perceived as human-made, suggesting that thoughtful design can neutralize the so-called AI penalty. By emphasizing elements like authentic faces and natural composition, AI-generated ads can blend seamlessly into human content ecosystems. ### How can AI-generated ads achieve a “human touch”? You need more than prompt engineering to humanize AI-generated creative. Try removing repetitive phrasing, focus on real audience pain points, and allow room for humor or narrative context. In practice, the most successful teams treat AI as a collaborator rather than a replacement, using human judgment to guide tone and emotional direction. Realize embeds this humanization directly into its AI systems by training models on proven creative best practices. When AI is informed by what already resonates across large publisher networks, it can replicate human-like patterns at scale. This includes automatically prioritizing visual elements that audiences associate with authenticity, reducing the need for constant manual oversight. ### What are the performance benefits of using AI for human-centric ads? AI’s primary advantage remains its ability to personalize and iterate at speed. High-velocity testing allows advertisers to identify effective messages and visuals faster than manual workflows permit, improving engagement and return on investment. Importantly, research examining downstream performance suggests that increases in click-through rate driven by AI-generated creative do not come at the expense of conversion quality. On the Realize platform, AI-generated ads that adhere to human-centric principles can outperform traditional creative while maintaining conversion integrity. This combination of scale, efficiency, and trust preservation represents the most compelling use case for AI in performance advertising today. --- ### Contextual Targeting vs. Behavioral Targeting: Which to Choose for Your Campaigns? URL: https://www.taboola.com/marketing-hub/contextual-targeting-vs-behavioral-targeting/ Last Modified: 2026-03-08 09:17:23 The world of performance marketing is constantly shifting, and reaching the right person at the right time has become increasingly challenging. As privacy regulations tighten and third-party cookies crumble, the way we find that person is what’s changing fastest. For e-commerce brands looking to scale, the debate often boils down to two heavy hitters: contextual and behavioral targeting. Deciding where to put your ad spend isn't just about following trends anymore, it’s about understanding the environment your customer is in versus the history they carry with them, in addition to the tools you use that can give you an extra edge. Let’s take a more in-depth look at what this all means for you, your advertising, and your ongoing strategy. ## Contextual Targeting ### Description Think of contextual targeting as a more “in-the-moment” approach. Instead of following a specific user around the web based on their past actions, contextual targeting places your ads on pages that are relevant to the product you’re selling right now. For example, if you’re selling high-end running shoes, your ad might appear next to an article about “Marathon Training Tips.” ### How It Works Contextual targeting scans the actual content of a web page (like keywords, topics, and sentiment) to determine if it matches your ad’s theme. It doesn't need to know who the reader is — only what they’re reading at that second. ### Benefits - Privacy-first: Since it doesn't rely on cookies or personal data, it’s inherently compliant with GDPR and CCPA. This is a huge deal as privacy laws become more widespread. - Relevance: You catch users while they’re already in a specific mindset and ready to look at your product. - Brand safety: You have more control over the types of content your brand is associated with. ### Considerations One of the problems with contextual targeting is that it can sometimes be a little too literal, and if not managed correctly, you might put off a potential customer who’s reading a news article only semi-related to your product. That’s where AI-driven performance advertising platforms help: By using advanced AI to understand contextual signals beyond just keywords, your shoe ad doesn't accidentally end up next to a tragic news story about an incident at a marathon. ### Use Cases - Launching a new product category where you don’t have much historical data. - Campaigns on high-authority news sites where user privacy is strictly protected. ## Behavioral Targeting ### Description Behavioral targeting is an approach that follows the user around more. It focuses on the person rather than the page, looking at a user’s past browsing history, search queries, and purchase data in order to build a profile. ### How It Works Through pixels and cookies, the system tracks a user’s digital footprint. If someone visited your e-commerce store yesterday but didn't buy anything, behavioral targeting (often seen as retargeting) ensures your ad pops up while they’re checking the weather or reading a movie review today. ### Benefits - High conversion intent: You’re reaching people who have already shown interest in your product or similar categories. - Personalization: Ads can be tailored to the specific stage of the customer journey. ### Considerations Attitudes have shifted on these types of ads over the past few years, as they can feel invasive, often not sitting well with customers. Users are increasingly wary of ads that seem to follow them everywhere as privacy concerns grow. Plus, as browsers like Safari and Chrome phase out third-party cookies, the “signal” for behavioral targeting is becoming weaker. ### Use Cases - Abandoned cart recovery. - Loyalty programs and cross-selling to existing customers. ## How Does Contextual Targeting Compare to Behavioral Targeting? Feature Contextual Targeting Behavioral Targeting Privacy Compliance High (no cookies needed) Moderate (requires data consent) Campaign Goal Awareness and consideration Conversion and retargeting Setup Complexity Low (keyword/topic-based) High (pixel and audience setup) Audience Scalability Vast (content-based) Limited (user-list based) Immediate Performance Medium High (intent-driven) Long-Term Brand Lift High Medium Cost Efficiency High (lower CPCs) Variable (higher CPMs for niche lists) Automated AI Integrations Real-time content analysis Predictive user behavior Brand Safety Excellent Variable A/B Testing Focus on creative vs. topic Focus on audience segments ## How to Decide When to Choose Contextual Targeting Contextual targeting sits best as top-of-funnel awareness, so the ideal time to choose contextual targeting is when you want to cast a wide net in a safe environment. Plus, if you’re using Realize, you can use Topics Targeting to leverage contextual segments, reaching users on premium publishers like Apple News or Yahoo without needing deep tracking data. The most important thing is building a brand association with the topics your customers love. ## How to Decide When to Choose Behavioral Targeting Lean into behavioral when you’re optimizing for the final sale. If you have a solid list of past visitors or a well-defined customer persona, behavioral targeting allows you to nudge them across the finish line. Realize also enhances this by using “lookalike” modeling, i.e., taking your best customers’ behaviors and finding new users who act just like them. ## Key Takeaways The most successful e-commerce strategies don't treat these two approaches as rivals, but rather as a partnership. - Start with contextual targeting to fill your funnel and build brand equity. - Use behavioral targeting to capture the low-hanging fruit and retarget those who engaged. - Leverage Realize to manage the technical heavy lifting, from navigating publisher compliance to optimizing your bidding strategy across both methods. ## Frequently Asked Questions (FAQs) ### Can combining contextual and behavioral targeting improve campaign performance? Yes it can, and it’s often called hybrid targeting. By using contextual ads to find new users, and behavioral ads to bring them back, you create a full-funnel experience. This helps brands scale past their initial performance plateaus. ### How do I decide which targeting method will be more cost-efficient for my campaign? In general, contextual targeting offers lower CPCs because the inventory is broader. But, behavioral targeting often has a higher ROAS because the users are warmer. Ultimately, the best way to decide what’s right for you is to run a controlled A/B test to see which delivers the lower effective CPA for your specific product. ### Are there industries where one method consistently outperforms the other? Niche industries (like specialized medical equipment) often thrive on contextual targeting because the content is so specific. High-frequency consumer goods (like apparel or beauty) often rely more heavily on behavioral targeting to stay top-of-mind in a crowded market. --- ### 3 Key Areas to Automate for Maximum Campaign Optimization URL: https://www.taboola.com/marketing-hub/automation-for-campaign-optimization/ Last Modified: 2026-03-19 13:15:21 Modern digital advertising now operates at a speed and scale that manual management simply cannot keep up with. Campaigns generate performance signals in real time, and advertisers need to act on those signals instantly, not hours later when someone logs in to make adjustments. Human teams, no matter how skilled or attentive, cannot watch performance indicators 24/7 or react with the precision and consistency needed to prevent wasted spend. As a result, advertisers are relying on automation more than ever to protect campaign performance and leverage opportunities as soon as they appear. In this article, I’ll explore three key areas where automation is no longer optional if you want to optimize effectively, and how these automated systems reduce risk and scale impact. ## 1. Continuous Risk Management: Stop/Loss Automation ### KPI Monitoring and Performance Protection Automation plays its most critical role in protecting campaigns from reduced performance and financial loss. Stop/Loss rules act as an early-warning system by continuously scanning key performance indicators (KPIs) such as cost per acquisition (CPA), cost per lead (CPL), cost per click (CPC), or spend thresholds. When a campaign exceeds a predefined limit, the rule triggers an immediate action. This action might be pausing a campaign, stopping an ad group, or triggering a budget reduction. The key advantage is reaction time: Automation can respond to emerging issues the moment they happen, not hours later. For example, let’s say a CPA threshold is set at $50 and a campaign suddenly spikes to $80. In that situation, a Stop/Loss rule can respond instantly, even if the advertiser is offline or busy managing other campaigns. This immediate action prevents runaway spending and preserves budget for higher-performing campaigns. ### Conditional Pausing of Underperforming Ads Stop/Loss automation doesn’t just operate at the campaign level — you can also set it to evaluate each creative asset or ad group individually to identify poor performers before they affect the campaign’s overall metrics. When a specific ad starts to show high CPA, low click-through rate, or insufficient conversion volume, automation can pause it automatically. This conditional pausing ensures that the remaining budget is redirected to stronger performing ads without having to wait for a manual review. This helps maintain healthier CPA averages and frees teams from constantly checking which creatives need to be swapped out. Instead, they can focus on producing better assets, rather than monitoring failing ones. ## 2. Dynamic Spend Control: Budget Adjustments Automation doesn’t just mitigate risk, it also supports growth by reacting instantly when campaigns deliver strong performance. Advertisers can use automation to scale what works and reduce what doesn’t through conditional logic. ### Scaling Success With Automated Budget Increases If a campaign’s CPA sits below the target for a sustained period, an automated rule can increase the daily budget by a fixed amount or percentage. This ensures that the campaign receives more funding as soon as it achieves optimal efficiency. Without automation, advertisers may not notice these movements in time, or they may not be able to react quickly enough to maximize the opportunity. ### Containing Failures With Automated Budget Decreases Just as important as scaling success is containing potential losses. When performance trends downward, budget automation can reduce daily caps to minimize unnecessary spend. This helps advertisers control costs without shutting down the entire campaign too early. Automated decreases also ensure that the advertiser’s spend remains aligned with KPIs even when performance shifts unexpectedly. It keeps budgets flexible and responsive, something advertisers rarely achieve through manual real-time adjustments. ## 3. Proactive Optimization: Real-Time Compliance and Recommendations Automation isn’t limited to ad budgets and performance metrics: Today’s platforms offer real-time assistance in areas where human teams often struggle to keep up, such as compliance and strategic optimization. ### Real-Time Ad Compliance Checks Ad compliance issues can disrupt campaigns if they aren’t identified quickly, whether they involve creative content, landing page responsiveness, or platform policies. Automation helps solve this problem by continually evaluating ads and landing pages to identify issues before they lead to rejections or downtime. Automated compliance systems can ensure that a landing page loads correctly, that ad text adheres to guidelines, and that creative assets meet required specifications. When issues arise, the system can quickly flag or correct them. This reduces disruptions and saves advertisers from losing impressions or performance momentum due to preventable compliance errors. Real-time monitoring also means that marketing teams don’t have to perform these checks manually, which would be a significant burden. ### AI-enabled Recommendations Many advertising platforms now offer AI-powered optimization recommendations. These suggestions translate complex performance data into concrete actions, like reallocating budget, creating new audiences, adjusting bids, or resolving performance bottlenecks. For example, Taboola’s AI ad assistant, ABBY, can evaluate campaign behavior, analyze performance signals, and surface actionable insights. Instead of spending hours combing through dashboards, advertisers can rely on AI to identify the adjustments that will most likely improve performance. I should note that these recommendations don’t replace strategic thinking; they accelerate it. Advertisers still need to decide which suggestions to apply, but automation drastically reduces the time spent identifying issues, testing ideas, or troubleshooting constraints. ## Key Takeaways Clearly, automation is now a core requirement for high-performance campaign management. With so many variables shifting in real time, human teams cannot rely solely on manual oversight. Automated Stop/Loss rules and conditional budget adjustments ensure campaigns receive round-the-clock protection and support. This prevents wasted spend and allows advertisers to scale success instantly. With custom rules, advertisers can delegate routine optimization decisions to automated systems. By relying on automation, they can spend more time focusing on creative development, audience exploration, and higher-value strategic work. ## Frequently Asked Questions (FAQs) ### What is a key benefit of automating campaign controls? Automation provides continuous, 24/7 oversight and protection. Instead of waiting for a human to detect performance problems, automated systems react immediately to safeguard the budget and maintain KPIs. For example, Realize’s Custom Rules automatically pause underperforming ads and adjust budgets, preventing wasted ad spend. ### How does automation handle “stop/loss” for budgets? Automated rules monitor KPIs such as CPA, spend, or conversion volume and trigger predefined actions, such as pausing an ad or lowering the budget, when indicators exceed a set threshold. This functionality protects your ad spend and ensures that your campaigns stay closely aligned with your performance goals. ### How can automated rules help with scaling successful campaigns? Automation can identify campaigns that are outperforming expectations. For example, suppose your campaign’s CPA is lower than expected. In that case, it will automatically increase the daily budget to capitalize on strong results without waiting for the advertiser to enter an increase manually. --- ### 7 Must-Know Mobile UX Best Practices for 2026 URL: https://www.taboola.com/marketing-hub/mobile-ux-best-practices/ Last Modified: 2026-03-31 13:07:21 The mobile-first mantra has evolved. Do you still need a responsive site? Yes, of course. But, the experience on a mobile phone or smartwatch must be equally fluid, so that a busy executive can approve a $50,000 contract while waiting in line for their latte. Despite the fact that mobile users have been the online majority for a decade or more, many B2B platforms still feel like desktop software crammed into a pocket-sized screen. There’s no need for that when you absolutely can bridge the gap between business logic and mobile interaction. Here’s your comprehensive guide to the mobile user experience (UX) best practices that will define B2B success in 2026. The key? Convenience, responsiveness, and ease of use. ## Why Does Mobile UX Matter? The short answer? The buyer has changed. The long answer? It’s a mix of psychological shifts, artificial intelligence (AI) integration, and a loyalty tax on slow experiences. Acquiring a B2B customer costs more, but mobile UX can help with retention. According to Forrester, organizations adopting rigorous user testing for their digital experiences see revenue retention improvements of up to 10.8% over three years. The line between a B2B buyer and a B2C consumer has blurred, too. UX — not price or product features — drives 80% of B2B purchases. Your prospects use Uber, Instagram, and Airbnb in their personal lives; they have little patience for a clunky, 2015-style enterprise portal when they switch into work mode. Research shows that 94% of first-impression assessments are design-driven. On mobile, that judgment happens even faster. Over 60% of consumers discover brands and products on mobile devices before ever logging onto their laptops (and 75% complete their entire purchase journey on a smartphone or tablet). If your mobile experience fails, you lose that mobile user and the lead before they even reach their desks. Good UX builds habitual ties. A business tool that works well on a smartphone can become the default choice for users, creating a moat that competitors can’t easily cross with lower prices alone. In an era where AI generates infinite content, the only thing you can’t easily replicate is experience. If your mobile UX is a barrier, you’re worse than invisible — you’re irrelevant. ## Mobile UX Red Flags To help you audit your current mobile experience, use this red flag checklist. Find yourself nodding along to more than three of these warnings? Your 2026 growth strategy might be leaking leads. - If your user has to zoom in to read your “Request a Quote” form or click a text link, you’ve failed the most basic mobile UX test. Content must be natively legible. Use a minimum 16-point font for body text and ensure your layout is fluid, not merely shrunk. - Requesting a phone number, job title, industry, and company size on the first mobile screen? B2B marketers often prioritize data over the user. On mobile, every form field increases the chance of abandonment by 10%. Use social sign-on such as LinkedIn or Google, or a single-field email capture to start. - B2B designs often favor thin, elegant fonts and small icons. However, evolution has not caused the human thumb to shrink. If your buttons are clustered too closely, users will accidentally hit Cancel instead of Save. Maintain a minimum scale of 44x44 pixels for all interactive elements. - Does your menu contain 15 different sub-categories? Desktop mega-menus are an organizational disaster on mobile. If users must scroll through a massive list to find your contact info, spoiler alert: they won’t. Use a hamburger menu to prioritize the top 4 tasks, and a search bar for everything else. - If your platform uses AI to generate B2B insights but simply drops a number on the screen without context, users will doubt the data. A lack of transparency is a UX failure. Add a “Why am I seeing this?” tooltip that explains your data sources in a sentence or a few bullet points. ## Best Practices for Mobile UX Success ### 1. The “Thumb Zone” and One-Handed Mastery Ever try to hit the Submit button in the top-left corner of a smartphone (even a Pro-sized one) while holding a briefcase? It’s a recipe for a dropped phone and a lost lead. Mobile users often operate devices with one hand in crowded or transit-heavy environments. To accommodate this typical use case, your most critical actions — like “contact sales” or “view demo” — should live in the natural thumb zone (the screen’s bottom two-thirds). #### Key Layout Tactics - Bottom navigation: Use a tab bar at the base of the screen for primary navigation. - Large tap targets: Ensure buttons are at least 44x44 pixels to prevent large-finger errors. - The center-lower priority: Put high-stakes interactions in the center-bottom section for max accessibility. ### 2. The 60-Second Onboarding Rule In B2B, time is money and the ultimate friction point. If your onboarding process is cumbersome, your bounce rate will reflect it. High-quality onboarding (including a seamless process) is a requirement for user acquisition in 2026. #### How to Accelerate Time-to-Value (TTV) - Embrace progressive disclosure: Don’t ask for a company’s entire history up front. Request the bare minimum to start and gather more data as they engage. - Show, don’t tell: Use subtle animations to guide the eye. Micro-interactions can confirm actions and create forward momentum without heavy text. - Offer optional sign-ups: Whenever possible, let users explore your platform’s value (like browsing a library or seeing a dashboard preview) before requiring a login. ### 3. Designing for Explainable AI and Agentic UX Potentially the biggest disruptor for 2026 is the continued rise of Agentic UX. According to PwC, 88% of business leaders plan to increase their AI budgets for agentic capabilities. What that means for mobile UX is that it must now account for how users interact with AI agents that perform tasks on their behalf. #### Bridging the Trust Gap - When an AI tool suggests a B2B strategy or report, display its reasoning in clear, accessible language. The explainable AI market is projected to exceed $33 billion by 2032 because users won’t trust systems they can’t understand. - Not everyone knows about Control+Z. Provide emergency exits or undo buttons. If an AI agent drafts an automated email to a client, the mobile interface must make it easy for the human to intervene before it sends. ### 4. Speed as a Feature Goose and Maverick once said, “I feel the need. The need for speed.” Did you know that 38% of first-time visitors look at your navigation menu the moment they land? If that menu takes more than a few seconds to load, you’ve lost them. Performance is the backbone of a trustworthy and positive B2B experience. #### Technical Best Practices - The 1-3-10 rule: If an action takes more than 1 second, you don’t need an indicator. If it takes 1-3 seconds, use a spinner. For anything longer than 3 seconds, use a progress bar. - Skeleton loaders: Use skeleton screens (neutral placeholders) instead of blank white screens to make the app feel faster than it actually is. - WebP and lazy loading: Use WebP for images to balance quality and speed, and implement lazy loading so the above-the-fold content appears instantly. ### 5. Consistency Across the Ecosystem Consistency reduces cognitive load. In a B2B context, you might switch from a desktop browser to a mobile app five times (or more) a day. If the iconography or submit button behavior changes between devices, you’re forcing your audience to relearn your product every time. Cross-platform unification is a top trend for 2026. It’s simple: you maintain a uniform visual language (colors, typography, and spacing) across the web, mobile, and even wearables. Consistency matters because it: - Builds familiarity. Users learn patterns once and apply them everywhere. - Reduces support costs. Clear, consistent user interface (UI) leads to fewer support tickets. - Strengthens brand. A cohesive look boosts brand identity and professional trust. ### 6. Accessibility-First Isn’t Optional With about 16% of the global population living with some form of disability, accessibility has become a legal and ethical requirement. In 2026, B2B platforms should comply with WCAG 2.2 (Web Content Accessibility Guidelines) standards without question. #### Accessibility Checklist Your platforms should: - Ensure a high contrast ratio for text and backgrounds. - Use correct HTML structures (semantic headings) so that assistive technology can navigate your app. - Use vibrations to confirm successful actions for users with visual impairments. ### 7. Micro-Copy Packs a Punch In mobile design, every word must earn its keep. B2B jargon often clutters small screens. Skip the technobabble in favor of active voice and benefit-driven copy. - Bad: The system is processing your data synchronization request. - Good: Syncing your data. Almost there! Focus on what the user gets, not what the feature does. This approach reduces the mental effort required to navigate complex business tools on a 6-inch (or smaller) screen. Feature The Old B2B Way The 2026 B2B Standard Onboarding 10-page manual/long video Interactive, three-step walkthrough Loading Blank white screen Skeleton loaders and first response Interactions Click-heavy Gesture-based (swiping/haptics) AI Integrations Chatbots that don’t help Proactive AI Agents with reasoning Accessibility Afterthought/compliance only Core design pillar (WCAG 2.2) ## Future Trends in Mobile UX Design Trends in mobile UX design will keep evolving as devices themselves evolve and user behaviors change. Potential future trends in mobile UX design include: - Hyper-personalized and predictive systems: AI-powered on-device intelligence will adapt interfaces to user behavior, providing smart suggestions, changing menus, and creating invisible UX where the app anticipates needs. - Multimodal and spatial interfaces: UIs will continue to integrate voice (VUI), gesture control, and spatial/augmented reality (AR) to create immersive experiences, as seen in smart glasses technology. - Functional minimalism and clear design: We continue to move away from decorative, heavy, or complex, visually-dense design toward cleaner, faster, more purpose-driven interfaces that improve readability and performance. - Sustainable and ethical UX: Design will prioritize digital well-being, environmental sustainability, and ethical AI usage to deliver functional and responsible interfaces. - Advanced visual techniques: Expect more use of 3D elements, kinetic typography, and progressive blur effects to create more dynamic and engaging (but still clean!) visual experiences. - Persistent and adaptive dark mode: Dark mode is becoming a standard and highly optimized experience designed for battery efficiency and comfort. ## Key Takeaways With over 80% of purchases driven by experience quality, not just price, B2B mobile UX design remains a key factor for customers who rely on your app. Today’s mobile platforms must bridge the gap between business logic and interaction by recognizing and prioritizing how people use their devices: - A thumb zone for one-handed use. - A 60-second, easy-peasy onboarding rule. - Robust technical performance (and no lag time). Most importantly, mobile apps should feature a clean visual design, be accessible to all users, and deliver a consistent experience across smartwatches, smartphones, tablets, laptops, and desktops. ## Frequently Asked Questions (FAQs) ### How do you do UX research, and why is it important? UX research involves studying user behaviors and pain points through usability testing, heatmaps, and interviews. The research matters because it removes any guesswork. In B2B, where acquisition costs are high, research helps you identify the features users want and the features you might think are cool, but that users will ultimately ignore. ### What is the difference between UX and UI design? UX, or user experience, focuses on an app or website’s logic and feel. Is the journey logical? Is it easy to find the checkout or other information? UI, or user interface, focuses on the look — the colors, fonts, spacing, and overall visual aesthetic. ### What are the best mobile UX design tools? The industry leaders in mobile UX design tools are: - Figma: The gold standard for collaborative design and prototyping. - Adobe Express/XD: Great for quick iterations and integration with the Creative Cloud. - Spline: Increasingly popular for 3D elements and kinetic typography. - Prototyper.io: Used for advanced gesture-based animations. ### Who is the inventor of UX? Industry experts credit Don Norman with coining the term “user experience” in the 1990s while working at Apple. He argued that the experience covers all aspects of a person’s interaction with the system, including the manual and the packaging. ### What is agentic UX? The magic of agentic UX happens when the UI is designed for AI agents that act on a user’s behalf. Instead of someone clicking through five screens to generate a report, they ask an agent, and the UX focuses on showing the agent’s process and reasoning for the final output. --- ### Why AI Ads Offer Superior Cost-Efficiency Without Sacrificing Performance URL: https://www.taboola.com/marketing-hub/ai-ads-cost-efficiency/ Last Modified: 2026-01-28 08:08:35 For chief marketing officers, chief financial officers, and performance marketing leaders, the question surrounding generative artificial intelligence is no longer whether it works, but whether it improves return on investment. Creative production has become one of the most costly parts of digital advertising, and performance expectations continue to rise: Advertisers are under pressure to test more variations, personalize at greater scale, and maintain efficiency in increasingly competitive auctions. Against this backdrop, artificial intelligence (AI)-powered ad creative has emerged as a potential solution, but it brings perceived risk — namely, that if AI-generated ads cause skepticism or disappointment, any efficiency gains will quickly disappear. That challenge is exactly what a recent large-scale study set out to explore. Conducted by researchers from Columbia Business School, Harvard Business School, the Technical University of Munich, and Carnegie Mellon University, the study examined whether AI-generated ads — created by Realize’s GenAI Ad Maker — could deliver results without increasing costs or sacrificing quality. Their findings challenge several assumptions that have shaped how leaders think about AI creative. ## Debunking the Curiosity Myth An ongoing concern with AI-generated ads is the idea of “curiosity clicks.” The worry is that ads created by artificial intelligence may attract attention simply because they stand out as unusual or novel, rather than because they resonate with genuine user intent. Over time, this type of engagement could undermine not only ad performance, but overall brand trust. The research disputes this. What matters most, per the data, is perceived artificiality, not how an ad was actually made. Researchers describe this reaction as algorithm aversion, which is a negative predisposition toward content consumers believe was machine-made. Ads that look synthetic are penalized, even when they’re created by humans. On the other hand, AI-generated ads that convincingly blend in with surrounding content avoid this penalty entirely. Those AI-created ads didn’t just go undetected alongside the human-made ones, though: The study found that, when indistinguishable from their human-made counterparts, these ads achieved what they called “superhuman” click-through rates (CTRs), exceeding traditional ads. This is where cost-efficiency enters the picture. Tools such as GenAI Ad Maker allow advertisers to scale creative output rapidly, but scale alone is not the advantage: The real advantage is that quality and authenticity can be preserved even as volume increases, as long as the right creative conditions are met. ### Human Cues: The Face of Success Among all the visual characteristics examined, one factor stood out above the rest: Ads that featured large, clear human faces were far more likely to be perceived as human-made and, as a result, achieve higher engagement. The finding reinforces what performance advertisers have seen for years across eye-tracking studies, platform benchmarks, and creative effectiveness reports. Human faces play a critical role in fast-moving, feed-based environments because they: - Capture attention quickly. - Convey emotion and relatability at a glance. - Establish trust in crowded content streams. What surprised the researchers was how consistently AI-generated ads applied this principle. In the dataset, creatives produced with GenAI Ad Maker were more likely to include prominent human faces than traditional human-made ads. That emphasis helped AI-generated ads blend naturally into surrounding content. It also explained why many of these ads matched or exceeded human performance when they avoided overly stylized or artificial aesthetics. The takeaway isn’t that AI changes the rules of effective ad creative. Instead, it shows that AI can apply longstanding, human-centered best practices reliably and at scale, reinforcing performance instead of undermining it. ### Matching Human Results at Scale The conclusions in this study aren’t based on small A/B tests or limited pilots. Rather, they come from real-world performance data analyzed at significant scale, offering a rare view into how AI-generated creative performs in live campaigns. Across the full dataset, the results were clear: - 300,000+ live ads were analyzed across multiple industries. - AI-generated creatives averaged a 0.76% CTR. - Human-made ads averaged a 0.65% CTR. These numbers suggest a performance advantage for AI-generated creative. The research went a step further, though, to ensure those results weren’t influenced by external factors such as targeting differences, timing effects, or campaign structure. To accomplish that, the researchers applied a “sibling ads” methodology. In this approach, AI-generated and human-made ads were matched so that they: - Came from the same advertiser. - Ran in the same campaign. - Launched on the same day. - Shared the same objective and landing page. Under these tightly controlled conditions, AI-generated and human-made ads showed a performance that was statistically comparable. For decision-makers, this distinction is important. Matching human performance under strict controls shows that AI-generated ads aren’t simply benefiting from looser conditions or novelty effects. Instead, they can deliver human-level outcomes reliably and at scale, creating room for efficiency gains without risking performance. ### Efficiency for the CFO: High Performance, Low Cost From a financial standpoint, the promise of AI-powered creative is no small improvement: It signals a substantial shift in the way ad spend is allocated. Traditional creative production relies on human labor, agency workloads, and fixed costs that scale poorly. A recent survey of marketing leaders found that 76% are investing in generative AI solutions, with more than half specifically targeting creative production to cut costs and speed up workflows. AI tools such as GenAI Ad Maker simplify creative generation. They enable rapid creation of ad variations at a fraction of the cost and time associated with manual production. Most importantly, the study found no evidence that higher AI-driven CTRs came at the expense of conversions: performance held steady even as creative volume increased. For chief financial officers (CFOs) focused on efficiency and risk management, the combination of stable conversion outcomes while cutting costs is appealing. It allows teams to shift resources away from production and toward testing and improving ad performance. ## 5 Ways AI-Creative Campaigns Offer Scalable Results at a Lower Cost Cost efficiency in advertising isn’t about cutting corners, it’s about removing friction from the parts of the workflow that can slow learning, limit testing, and consume budget without improving outcomes. Below are five ways AI-driven creative campaigns help advertisers achieve stronger performance at lower cost, without sacrificing quality or control. ### 1. Faster Time to Learning In performance marketing, speed to insight is often more valuable than perfection. Multiple industry reports show that creative is one of the biggest bottlenecks in optimization cycles. This isn’t because teams lack ideas, it’s simply because producing, testing, and deploying new variations takes time. Unfortunately, regular testing is still crucial to success. Studies find that increasing the velocity of creative testing leads to better performance outcomes, with brands that A/B test ad creative on a weekly basis seeing 31% higher conversion rates than those that test less frequently. AI-powered creative workflows compress those learning loops. By generating and launching multiple creative variations quickly, advertisers can gather useful insights early in a campaign’s lifecycle. That feedback helps teams refine messaging, visuals, and formats while budgets are still flexible, rather than discovering issues after the money has already been spent. ### 2. Lower Marginal Cost per Variant With traditional creative production, scaling comes at a cost. Each new asset requires additional design time, revisions, approvals, and coordination across internal or external teams. For many organizations, that means the marginal cost of testing “one more idea” is high enough to limit experimentation. Organizations cite creative production as one of the most resource-intensive aspects of modern digital marketing, particularly as teams attempt to deliver personalized and multi-format campaigns at scale. AI shifts that equation by making it easy to generate additional assets without requiring more resources. This allows advertisers to expand coverage without increasing headcount or spend, improving cost efficiency while also encouraging experimentation. ### 3. Reduced Creative Fatigue Risk Creative fatigue shows up in a simple but costly way: people tune out. When audiences are repeatedly served the same visual and message structure, engagement drops because the ad no longer feels like fresh information. One industry report found that 93% of customers skip or block ads. In crowded, feed-based environments, that means advertisers are often paying for impressions that don’t drive results. A few things typically contribute to creative fatigue: - Repeated exposure to the same visuals. - Messaging that becomes predictable over time. - Lack of variation across formats or placements. AI-driven creative helps reduce this fatigue by making refresh less of a burden. Instead of restarting the production process for every update, teams can generate and rotate new variations frequently while preserving the same offer and core message. As a result, brands can keep things fresh without constant manual effort. ### 4. Consistent Application of Proven Creative Signals Long before generative AI entered the conversation, advertising research consistently pointed to the same conclusion: Human cues drive attention. Clear focal points, realistic imagery, emotional relevance, and human presence get results. That includes the use of faces in ads, which studies have shown perform better than other visual stimuli. Key findings from a recent study reveal: - 91.7% of ads featuring a human face attracted more attention than ads without faces. - Faces are detected at least twice as fast as many other visual elements. - When models make direct eye contact, viewers are more likely to perceive trust and emotional connection. - Viewers subconsciously follow the direction a person in an ad is looking. AI systems trained on large datasets apply these principles consistently across hundreds or thousands of assets. Instead of relying on individual designers to interpret best practices asset by asset, AI ensures human-centric cues are repeated reliably at scale. That consistency helps protect performance as volume increases, which is something that’s difficult to maintain manually. ### 5. Budget Reallocation Toward Media Performance The shift toward AI doesn’t just reduce production costs, it frees up funds that businesses can shift toward other efforts. When creative absorbs a large share of the marketing budget, it constrains how much can be invested in reach, testing, and optimization. Since marketing leaders are under increasing pressure to demonstrate short-term return on investment, it’s important to assess how much spend is tied up in fixed operational costs rather than flexible, performance-driven investments. Lowering creative overhead lets organizations redirect spend toward the levers that most directly influence outcomes, including: - Expanding reach into new or under-tested audiences. - Increasing testing velocity across creatives, formats, and placements. - Funding faster optimization through bidding and budget adjustments. - Extending the lifespan of high-performing campaigns without creative bottlenecks. By reducing the fixed cost of creative production, AI enables this shift without forcing trade-offs in quality or volume. For business leaders, the result is improved capital performance. The same overall budget delivers more testing, more reach, and more opportunities to optimize. ## Key Takeaways Generative AI is changing the economics of performance marketing. When applied strategically, AI-driven creative enables faster experimentation, broader coverage, and sizable cost efficiencies without sacrificing effectiveness. At scale, AI-generated ads can perform on par with human-made creative while removing many of the operational constraints that traditionally limit testing. The advantage doesn’t come from automation for its own sake: Results improve when AI is used to reinforce established creative best practices, producing ads that feel authentic, credible, and designed for how people actually engage. ## Frequently Asked Questions (FAQs) ### Do AI ads drive accidental or low-quality clicks? Some advertisers worry that AI-generated ads may attract attention because they feel unfamiliar or “uncanny,” leading to clicks driven by curiosity rather than genuine interest. In real-world campaigns analyzed in the study, though, higher engagement from AI-generated creative (in this case, from GenAI Ad Maker) did not come at the expense of outcomes. The study found no evidence that increased click-through rates led to weaker conversion performance, indicating that AI-driven engagement reflected real intent rather than accidental interaction. ### What visual “red flags” make an ad look like AI to consumers? Audiences tend to disengage from creative that feels overly artificial or synthetic. When visuals appear too perfect or unnatural, viewers may subconsciously distrust the content, leading to lower engagement and weaker brand perception. Ads that trigger that reaction often feature exaggerated polish, odd proportions, or human elements that feel subtly “off.” Insights from campaigns analyzed in the study highlight specific design patterns that increase perceived artificiality. Heavy color saturation and highly stylized, glossy visuals were most likely to signal AI-generated content. In contrast, ads featuring large, clear human faces consistently appeared more authentic to viewers and earned stronger click-through performance. ### What visual features make an ad look too much like AI? When creative leans too far into artificial-looking elements, audiences may instinctively pull back. Content that feels synthetic or exaggerated can trigger skepticism, reducing trust and engagement even before the message is processed. Analysis from campaigns found that excessive color saturation and overly polished design were the strongest indicators that an ad would be perceived as AI-generated. Creative that emphasized larger, more natural-looking faces and simpler, less saturated visuals was far more likely to be seen as authentic. --- ### 5 Winning Elements of High-CTR AI Ads: What's Working URL: https://www.taboola.com/marketing-hub/ai-ads-that-work/ Last Modified: 2026-01-28 07:14:30 The past few years have been a living experiment on broadly adopting a new technology — generative AI — for businesses and individuals. It’s largely been a blank canvas, with people exploring how to use AI to create written drafts, videos, imagery, and more, alongside process automation, data exploration, and other practical applications. AI usage in advertising only continues to rise, with 83% of ad execs saying that their company has deployed AI in the creative process (that’s up from 60% in the same source’s 2024 study). Now, we’re finally capturing data to truly understand how gen AI compares to solely human-created work. In the world of digital advertising, a new study from Columbia, Harvard, Technical University of Munich (TUM), and Carnegie Mellon, found that AI-generated ads perform at the same level as, or better than, human-made content. The researchers analyzed more than 300,000 ads and more than 500 million impressions. The AI-generated ads studied were created using Realize’s GenAI Ad Maker, which lets advertisers create ads using AI alongside their traditional, human-made content. The study found some key parameters to help advertisers understand how to use AI and human-made work in tandem, and how to incorporate the benefits of AI to serve up great ads. Use these best practices for creating and testing AI-generated ads that connect with your audience. ## 1. Prioritize Clear Human Faces in the Frame When it comes to AI-generated ads, the single most powerful visual feature for driving engagement is the presence of faces. Consumers recognize this signal of humanness, and they strongly associate larger displayed faces with human-made content. That perception increases trust and drives a higher click-through rate (CTR). Design elements matter quite a lot for performance when creating ads, per AdExpert AI: Genuine smiles increase engagement by 27% compared to neutral expressions, and authentic, candid shots outperform professional studio photos by 41%. Newer AI tools are already incorporating this parameter into their creative work. Realize’s GenAI Ad Maker, for example, is actually more likely to include human faces than traditional human-made ads, aligning with long-standing creative best practices. When prompting AI on imagery, specify "close-up" or "mid-shot" views to ensure facial features are prominent and clear. ## 2. Master the Looks-Like-AI Spectrum Our human brains often gauge imagery as AI-generated based on a quick glimpse, leading to the “uncanny valley” effect. That’s when there’s a drop in human empathy and comfort upon seeing artificial beings that are almost human, but not quite. It can apply to robots, CGI, or even dolls, and it’s important to consider in AI-generated ads. To make sure imagery is comfortable to consumers, try these tips: ### Avoid over-saturation Consumers often identify ads as AI-generated if they feature intense color saturation or overly polished aesthetics. ### Aim for realism Higher levels of perceived artificiality significantly lower an ad’s performance. AI ads that were not perceived as AI achieved the highest CTR of all groups in the Columbia study, scoring even above those that were actually made by people. ### Disguise through quality Use features that consumers mistakenly associate with humans, such as high image clearness and sharpness, to "disguise" the AI's origin. Ultimately, AI-generated ads perform best when they feel authentically human. ## 3. Consider Strategic Industry Application Advertisers and performance marketers work in incredibly varied product categories, and they’re not all created equal. The effectiveness of AI varies significantly by product category, so advertisers should use AI where it shows the strongest performance. The research study found that AI-generated imagery has delivered especially strong results in personal finance, pets, real estate, and food and drink, so try AI-created ads in these high-performing verticals. Conversely, categories like education haven’t seen the same uplift, suggesting that audience expectations for human expertise are higher in these fields. Recent data has also found that retail and e-commerce is the top industry for use of AI in advertising, with 26% using AI, more than financial services, media, healthcare, and others. Consider AI ads with contextual sensitivity based on what you know about your audience. ## 4. Use the Sibling Ads Testing Framework In the study, the researchers used a “sibling ads” approach, comparing matched pairs of AI-generated and human-made ads created by the same advertiser, for the same campaign, on the same day, with the same objective and landing page. This approach let them isolate the impact of AI-generated visuals while controlling for all the many other variables that typically influence ad performance. For your own ad testing and to maximize ROI, move beyond simple A/B testing and try the sibling ads approach for faster iteration. Launch AI-generated and human-made variations simultaneously within the same campaign, sharing the same landing page and objectives. Because AI lets you create dozens of ad variations in minutes, you can identify the winners faster, without the creative burnout of traditional design cycles. ## 5. Drive Clicks Without Sacrificing Conversions One recent concern for performance marketers is that AI ads might only drive curiosity clicks that don’t lead to true engagement. But, in reality, the data shows that AI-generated visuals increase or maintain CTR without reducing downstream conversion performance. There’s a lot of potential for AI to improve ad performance through engagement, too. Microsoft found that users of Copilot engage much more with ads, driving 73% higher CTR and 16% stronger conversion rates. Plus, Copilot shortens customer journeys by 33%, reducing the number of steps to conversion versus traditional search. It’s also important to remember that, because AI ads can be produced at zero or near-zero cost, the on-par performance results in a significantly higher overall ROI for the campaign. ## Key Takeaways As AI usage continues to grow in creative work like advertising, there’s more and more data available on how AI-driven ads perform compared to human-created ads. The future of high-performance advertising will rely on a strategic blend of AI efficiency and human-centric design. Advertisers can continue improving ad effectiveness and cutting down on production overhead by prioritizing authentic human cues, avoiding perceptual markers, and using elements that consumers are comfortable with. As models continue to improve, the most successful brands will be those that use AI to enhance, not replace, the human feel of their creative work. ## Frequently Asked Questions (FAQs) ### Does using AI-generated imagery negatively impact my ad's click-through rate (CTR)? There have been concerns among advertisers that algorithm aversion will cause users to ignore or distrust AI content. But, within digital advertising, performance depends heavily on the quality of the output and how well it mimics human-made design principles. The Columbia study, which studied Realize’s AI ad creator specifically, showed that AI-generated ads actually achieved a higher average CTR (0.76%) compared to human-made ads (0.65%) in the full dataset. Using the sibling ad method with identical settings, AI ads performed the same or better against human-made ones, proving they do not sacrifice performance for efficiency. ### Is there a risk that AI ads will attract junk clicks that don't lead to actual sales? Another recent industry concern is that AI content might drive curiosity clicks, where users click on an ad simply because the image looks strange or novel, leading to high bounce rates and low conversions. In the research study, however, the data found no evidence of this curiosity click phenomenon. AI-generated visuals on the platform increased or maintained CTR without reducing downstream conversion rates, meaning advertisers gained scale without trading off quality or ROI. ### What is the most effective way to ensure an AI-generated ad performs well? Successful AI ads should continue to follow established marketing psychology. That includes using clear focal points and high-quality resolution, while avoiding the uncanny valley effect where images look nearly, but not quite real to human viewers. AI ads perform best when they do not look like AI to the consumer. Those using Realize for AI ad creation know that the art of disguise is key to success: Make sure that ads built with AI include large, clear human faces, which is a feature that consumers strongly associate with human-made content, and that drives significantly higher engagement. --- ### Ad Personalization: Relevancy Without Overreach URL: https://www.taboola.com/marketing-hub/ad-personalization/ Last Modified: 2026-02-03 08:06:24 With personalized ads outperforming the more traditional kind, ad personalization has become the standard for e-commerce brands. At the same time, though, some consumers are resistant to the perceived intrusion that often comes with it. How can marketers balance ad personalization with data privacy? Let’s explore the basics of ad personalization and then dive into the best ways to deploy it in a rapidly changing digital landscape. ## What Is Ad Personalization? Ad personalization is the practice of tailoring ads to specific consumers based on their demographic data, past online behaviors, or interests. It goes beyond simply addressing a prospect by their name in an email, or recommendations of relevant products that may match thematically with items already in their online shopping cart: Ad personalization serves up ads based on context, lookalike audiences, and other demographic data to deliver highly relevant content when consumers are ready to buy. Although, as stated, these ads can rub consumers the wrong way — on average, 45.6% of consumers across generations in the U.K. and the U.S. have a “negative” reaction toward personalized ads, according to a survey from Verve and Censuswide — they’re still highly effective. Indeed, the same survey shows that 76% of consumers said they pay attention to relevant ads, while 66% said they’ve helped them discover products. Per Attentive, 81% of shoppers said they ignore ads that aren’t relevant to them, while an overwhelming majority (96%) said they are more likely to purchase from a brand that shares a personalized message. ## How Does Ad Personalization Work? Ad personalization works slightly differently based on the platform or media, such as social media vs. display ads, SMS, paid search, paid social, or other formats. Essentially, though, ad personalization collects data, often using cookies or pixels to track user behavior and actions across platforms. In some cases, it can bridge the gap between social media, e-commerce sites, the open web, and mobile. Ad platforms and AdTech tools collect demographic, geographic, and contextual data to serve relevant ads at the right time. This way, featured ads closely match users’ interests and behaviors. ## Benefits of Ad Personalization Ad personalization has proven to be an effective means to reach consumers with relevant messages, especially as AI and large language models (LLMs) become better at understanding how specific content relates to user data. For instance, through contextual clues, LLMs can determine whether an article about “apple” refers to a computer, the tech company’s stock, or a fruit. In relation to advertising, highly relevant ads can reflect a specific location, time of year or season, or product category. Ad personalization can also reflect broad or specific knowledge about the user, such as their age, interests, gender, occupation, or income level. ### Greater Audience Engagement As noted above, more than three-quarters of consumers in the U.S. and U.K. pay attention when ads are relevant to their interests, while 81% ignore irrelevant messaging. Capturing engagement in today’s fast-paced, scroll-happy world represents a chance at success versus languishing in obscurity. ### Improved Conversion Rates Once you’ve captured attention, you’ve increased the odds of a successful conversion — whether that means an opt-in, a share, or a sale. When brands use personalized ads, conversion rates can rise by 10% up to 275% or more. Amplifon (a hearing device retailer), e.g., saw a 29% boost in conversion rates using personalized ads. Publishers hosting ads for brands also see improved results through AI personalization. For instance, the USA Today Network increased their click-through rate by 47% using personalization features. ### Customer Loyalty The benefits of ad personalization can also be measured through intangibles, or metrics that are slightly harder to quantify, like customer loyalty. Brands who use personalization often see more repeat customers, per LinkedIn. ## Types of Ad Personalization ### Predictive Personalization Predictive analytics use past consumer behavior and AI to deliver relevant ads based on predictions that the user will behave in the same way in the future. For instance, someone who just purchased an electric bicycle might have a high likelihood of also buying an app-enabled smart helmet to go with it. Predictive personalization can also work well for upsells; someone who just purchased season tickets to a regional theme park may also be interested in a meal package. E-commerce retailers can use predictive personalization to suggest clothing of the right size and style based on past purchases. The more historical data you have, and the better the analysis technology, the more relevant your results. ### Lookalike Marketing Lookalike marketing is a variant of predictive personalization, where marketers look at various characteristics shared by current customers, and serve ads to similar consumers in the hopes of expanding reach. Lookalike audiences can be used for ad personalization, but the success of a campaign hinges on whether or not consumers in similar demographics behave the same way. ### Behavioral Targeting Behavioral targeting is another form of predictive personalization based on a user’s past behavior, such as opened emails, abandoned carts, web browsing, or prior ad clicks. ### Retargeting Retargeting is a form of ad personalization based on user behavior. It focuses specifically on abandoned carts, people who engaged with an ad but didn’t convert to becoming a buyer, or someone who visited your website and didn’t make a purchase. Marketers can track their actions on-site to target them with highly relevant, personalized ads and even money-saving deals. ### Contextual Targeting Contextual targeting can help marketers serve up personalized ads without any specific demographic or user behavioral data. Instead, these personalized ads appear adjacent to related content. For instance, an ad for a Nissan Pathfinder might appear next to an article highlighting the 10 best SUVs for families. ### First-Party Data Personalization First-party data personalization relies on information collected from the brand, publisher, or website without relying on the use of cookies or tracking pixels. First-party data eliminates privacy concerns and enables highly targeted advertising even in cookieless environments. ## Where to Use Ad Personalization ### Search Engine Results Pages You can personalize Google Ads campaigns, as well as paid search on other search engines. Google allows you to personalize campaigns based on keywords, audience demographics, user preferences, web activity, location, and more. Google also allows publishers and users to change their ad experience in My Ad Center; people can select interests, limit sensitive ad topics, or even turn off their personalized experience. By allowing users greater control of their ad experience, it can help advertisers serve the right ads to Google, YouTube, and Maps users. ### Social Media LinkedIn, Facebook, Instagram, and other social channels also allow marketers to personalize ads through retargeting or based on specific demographics. It often takes trial and error for marketers to dial in on the right parameters: If a group is too large, you’ll target users who don’t care about your brand; if the group is too small, you won’t see any results. ### Your Website Personalization on e-commerce websites takes the form of product recommendations and upsells, as well as messages enticing consumers to revisit their abandoned cart. ### The Open Web While search and social platforms operate within closed ecosystems, the open web enables advertisers to reach users across a broad network of trusted publishers and environments. Personalization for programmatic ads, for example, leads to better results by serving up highly relevant content at the time a user is most likely to make a buying decision. Realize, as another example, uses powerful AI algorithms that rely on context cues to place relevant, personalized ads without needing to harvest user data. ## How to Personalize Ads ### Understand Your Demographic Before you can find relevant users to share your ads with, you have to understand their shared characteristics. Understanding your audience is key to a successful campaign: Dive deep into your ideal buyer personas, including age, location, interests, income, and the type of content they typically consume. ### Choose How You’ll Find Your Ideal Audience Which characteristics matter most to find your ideal audience online? Are you personalizing ads based on the context of information your ideal consumers are reading or viewing? Are you serving up personalized ads based on their interests or other demographic data? Are you using cookies to track consumers across the open web? Or, are you relying on lookalike audiences to find people with similar interests and characteristics as your best customers? ### Develop Creative That Resonates With Your Audience Once you’ve found your audience, the key to a successful campaign is creating content that resonates with them. It doesn’t end with the ad in their social media feed or on a search engine results page: The landing page connected to the ad should be compelling, easy-to-read (and act on!), and should mimic the look and feel of the ad so users feel a sense of continuity and familiarity. ### Transparency Is Key to Building Trust Remember that stat saying that 45% of web users react negatively to personalized ads? If ads hit too close to the mark, they can feel creepy — almost as if advertisers are following a consumer’s every move. Yet, the vast majority of consumers also acknowledge that highly relevant ads are useful. Advertisers need to find a balance, delivering the exact content shoppers need without seeming intrusive. By revealing how you’re collecting data and what you’re using it for, you can set people’s minds at ease. As ever, transparency builds trust. ## How to Leverage Personalized Ads for Cross-Channel Campaigns ### Design a Consistent Cross-Channel campaign Consistency is key to brand recognition and building trust. Your cross-channel campaign should use a similar look and feel across channels, including the same branding, a cohesive color palette, and similar copywriting. ### Use Programmatic Advertising to Reach Audiences on Multiple Channels Programmatic advertising follows users across the open web based on contextual signals. For instance, someone researching new SUVs might visit sites like Consumer Reports, Edmunds, and car review websites. Ads placed on each of these sites build brand familiarity; if someone researching new SUVs sees the Nissan Pathfinder enough times, they are likely to consider purchasing the vehicle. Data-driven insights can help marketers scale their most successful campaigns across channels to expand reach, reach consumers in the decision stage of the sales funnel, and increase their click-through rates and ROAS. ## Key Takeaways Personalized ads in multichannel campaigns reach audiences in the action phase of the sales funnel, when they are ready to buy. Transparency in how you’re harvesting and using consumer data is important to maintain trust. Ultimately, consumers agree that highly relevant advertising is effective, but they don’t want to feel as if brands are tracking their every move, either. ## Frequently Asked Questions (FAQs) ### Where does personalized ad data come from? Personalized ad data can come from cookies or pixels placed on websites, first-party data collected by the brand, or demographic data collected by social media sites or search engines. ### What are some marketing tools used for ad personalization? Platforms like Realize are powerful marketing tools for ad personalization across channels, including the open web. Realize offers AI-driven Audience Matching to identify and serve ads to the audiences most likely to convert; Predictive Audiences to target high-intent users; and Dynamic Creative Integration to deliver tailored creative formats (display, vertical video, etc.) across diverse placements, which helps to match creative to user context and preferences. --- ### AI vs Human Generated Performance Creatives: Synergy, Not Rivalry URL: https://www.taboola.com/marketing-hub/ai-vs-human-creatives-myth/ Last Modified: 2026-07-06 09:00:49 Now that the novelty of generative AI is wearing off, advertisers have gotten more serious about its usage — and are asking questions about its true effectiveness as they build a long-term strategy. As generative AI technology keeps maturing, we’re more able to gauge its impact across industries, refining its use and better understanding how to apply it. For advertisers, “artificial intelligence” is often synonymous with “artificial creativity.” The goal, then, is to find the balance between generative AI creative work and human-generated creative work, which means gathering more information about how consumers perceive and respond to AI-made ads. To that end, a new study conducted by researchers from Columbia, Harvard, Technical University of Munich (TUM), and Carnegie Mellon, using data from Realize (Taboola's performance marketing platform), has found that AI and human ads perform remarkably well when run in tandem. Below, I’ll explain how best to use them together for maximum impact. ## Why Performance Advertisers Should Balance AI Creative Work with Human Insight By now, it’s clear that AI won’t replace human creativity and ingenuity. What it can do — and do well — is augment human time and effort, when used correctly. The study found that AI-created ads can even outperform human-created ads, which serves as a good reminder of AI’s incredible power and scale when harnessed intelligently. The findings echo what many advertisers and marketers are finding in their day-to-day work: Generative AI can bring lots of benefits, but it needs human guidance to create the right outputs. For creative work, those outputs are ads that feel authentic and human — something that advertising and marketing teams can gauge. Keep an eye on these trends as you’re using, measuring, and refining your own AI use in advertising. ### 1. AI-Generated Creatives Bring Efficiency Efficiency is one of the top reasons why generative AI took off so quickly over the past few years, across industries. Advertising and marketing teams have embraced AI for creative work, even if they haven’t always fully understood how performance compares to human-generated work. For many of these teams, AI is the best option to help them keep up with competitors: 78% of respondents with ad budgets under $10 million are using AI tools for brainstorming and to develop initial basic concepts. Cost efficiency is also now seen as a top benefit of AI, according to 64% of respondents in an Interactive Advertising Bureau (IAB) survey. In addition, 61% of respondents say that AI’s creative innovation is an advantage. AI-generated creatives can be produced at a fraction of the cost — estimated at just a few cents per request — and in significantly less time than human-generated work. Users are still exploring and understanding the nuances of using AI for core advertising and marketing work beyond efficiency. ### 2. AI vs. Human Ads Show On-Par Performance Whatever the company size, team size, available resources, and current use of AI, advertisers and marketers have the same goal: improving performance metrics. What ultimately matters is how an AI-generated ad performs when compared to a human-generated ad. In the controlled setting of the paper’s research, AI-generated ads maintained performance that was on par with human-made ads. The Columbia study found that ads using AI-generated images achieved “human-level” click-through rates (CTRs). And, if an AI-generated image doesn’t “look” like AI, the CTRs become “superhuman.” (Helpfully, the study also explored the boundaries of what’s perceived as artificial by ad viewers, such as overly stylized imagery.) ### 3. The AI-Human Relationship Is Still Evolving AI use is extremely common in advertising work: 85% of those in the field are either using or planning to use gen AI to build video ad creative, with 85% using AI for social media ads, 73% for display, and 56% for TV ads. In the meantime, consumers have strong points of view on the use of gen AI ads: 60% think that the use of generative AI in ads should always be disclosed to consumers. AI concerns aren’t unfounded, of course: One IAB survey found that 70% of marketers reported at least one AI incident, such as hallucinated outputs, biased or inappropriate content, or off-brand or offensive material. Because of those incidents, 40% of respondents had to pause or pull ads, while more than a third dealt with brand damage or PR issues. Other studies have found that the disclosure of AI-generated content can negatively impact perceived credibility. These areas of exploration align with AI’s maturity curve, and offer plenty of opportunity for human advertisers to refine their work with generative AI in ads. According to one industry expert, the best-performing ad campaigns still rely heavily on human judgment, strategy, and experience. ### 4. Some Industries Are Adopting AI Tools Faster for Advertising The Columbia study observed that AI adoption is a strategic choice, often led by performance-oriented advertisers. The AI image generation feature studied, Realize’s GenAI AdMaker, has been adopted early by high-performance industries, in particular: - Pets: 15.6% adoption rate. - Business: 4.68% adoption rate. - Technology and Computing: 3.73% adoption rate. - Real Estate: 3.52% adoption rate. - Travel: 2.9% adoption rate. In general, AI usage in advertising is set to grow, with 58% of respondents in one survey planning to increase the use of AI for creative generation in the next year. AI usage is expanding in terms of formats and workloads, too: Beyond developing concepts, AI can create net-new images and even video. Another IAB survey found that when it comes to ad creative, AI is mostly being used for scriptwriting (75%) and generating visuals (55%). ## Key Takeaways Generative AI usage in advertising has moved beyond initial experimentation into a phase of refining and strategizing, and a new study shows that AI-generated ads can match or exceed the performance of human-generated ads. For performance marketers, applying gen AI tools to ad creation allows them to scale massively beyond what a human team can do, by using data and previous learning to refine prompts and parameters to guide AI creation. Advertisers can build a parallel human-and-AI approach to refine their ad strategies and achieve new levels of performance. ## Frequently Asked Questions (FAQs) ### Does AI-generated imagery perform better in specific industries than others? The Columbia study examined different industries and found major differences in which ones are adopting and benefiting from AI. Sectors like personal finance, food and drink, and pets all saw high adoption rates for using AI to create ads. Categories like education saw a much smaller adoption rate. This suggests that audience expectations and the typical look of an industry can influence how those audiences receive AI-generated content. ### Can consumers actually tell when an ad is AI-generated? The research found that nearly half (45%) of AI-generated ads were perceived as definitely or likely human-made, reflecting inaccuracies in consumer assumptions about AI creative. Raters identified 24.87% of actual human-made ad images as being AI-generated. This suggests that the perceived artificiality of an image — i.e., how much it looks like AI — is a more powerful driver of consumer behavior than the actual source of the image. ### What are the specific visual tells that trigger Algorithm Aversion? The study identified certain features in advertising that consumers tend to think are AI-created. Consumers generally associate high aesthetics, intense color saturation, and strong symmetry with AI-generated content. If an ad is overly polished, consumers may subconsciously or consciously sense its artificial origin and react negatively. But, AI-generated ads also disguise themselves by using features that consumers believe are human, such as larger facial areas and image clearness. Those features are actually more common in GenAI Ad Maker outputs than in human-made ones. --- ### 12 Best Performance Marketing Attribution Softwares to Use With Realize URL: https://www.taboola.com/marketing-hub/best-performance-marketing-attribution-software/ Last Modified: 2026-01-20 13:25:46 More and more, performance advertisers are using Realize to improve their campaign efficiency, tracking accuracy, and open-web visibility. What often gets overlooked, though, is that Realize is not itself an attribution platform. While it does a great job of collecting high-quality, first-party event data and providing actionable insights to optimize Realize campaigns, Realize’s reporting is designed to help advertisers understand and improve Realize campaigns, and not to replace full analytics systems like GA4 or advanced attribution tools. Full attribution systems allow you to apply last-click, multi-touch, or data-driven models. They can track multiple channels in parallel, and they understand user behavior across sessions and devices. Realize should therefore be used as the optimization layer, while the right attribution software provides a complete picture of your performance across the marketing mix. With that in mind, I’ve compiled the best attribution platforms to use alongside Realize. Each tool serves a different type of advertiser, and each helps fill measurement gaps that Realize intentionally does not attempt to solve. ## Best Performance Advertising Attribution Software to Use With Realize Again, the best attribution tool for your campaign will depend on your specific use case. Before reviewing individual platforms, it helps to understand where advertisers typically pair Realize with additional attribution software. For example, some advertisers need B2B-focused tools that connect early Realize-driven engagement to downstream CRM outcomes, while others, such as e-commerce brands, need software that connects spend to purchases, LTV, and cohort behavior. Each of the tools below strengthens Realize’s role in the measurement ecosystem and expands your ability to see and optimize the whole customer journey. Software Name Best For Leverages Realize With Pricing Adobe Marketo Measure Enterprise B2B organizations Maps Realize click data to CRM/sales stages and pipeline. Starts ~$40k/year; can exceed $200k/year. Dreamdata B2B SaaS and ABM teams Uses Realize clicks to initiate account-level journey timelines. Starts at $750/month; custom plans up to $50k+/year. Cometly Mid-market B2B advertisers Server-side tracking connecting Realize to lead quality/deals. $500–$1,000/month after 14-day free trial. Triple Whale Shopify/DTC merchants Ties Realize spend to Shopify purchases, LTV, and renewals. $149–$219/month (Free Basic plan available). ThoughtMetric Scaling DTC brands Connects Realize traffic/spend to revenue and retention. $99–$1,500/month after 14-day free trial. AnyTrack Performance marketers/affiliates Captures Realize clicks for CAPI and server-to-server tracking. $100–$300/month after 14-day free trial. Voluum High-volume media buyers Enables A/B testing and rule-based routing of Realize traffic. Individual: $119–$299/mo; Business: $539–$7,999/mo. Hyros Subscription/long-horizon brands Uses an "identity spine" to track long-term Realize LTV. Shopify: ~$69/mo; Business: $230–$1,499/mo. Singular Multi-channel advertisers Aggregates Realize cost data with other platforms for ROI. Growth: $0.05/conversion (Free plan available). Adverity Advanced analytics/BI teams Ingests Realize API data into data warehouses (Snowflake, etc). Enterprise pricing (custom quotes only). Google Analytics 4 Universal baseline tracking Tracks Realize sessions via UTMs and event tags. Free (Standard version). Adjust Mobile app marketers Handles app installs and SKAdNetwork attribution for Realize. Custom pricing quote only. ### 1. Adobe Marketo Measure (Formerly Bizible) Adobe Marketo Measure is one of the most advanced multi-touch attribution platforms for B2B organizations, with use cases for customer lifecycle engagement, cross-channel personalization, sales and marketing alignment, and proving marketing impact. It takes in Realize click data as early-touch engagement and maps it across marketing automation workflows, SDR processes, sales interactions, and CRM opportunity stages. By stitching together both anonymous and known interactions, it provides a complete view of how Realize influences pipeline development and revenue generation across long and complex buying journeys. Special Features  - Multi-touch attribution models (W-shaped, full-path, custom modeling). - Deep Salesforce integration and revenue-based reporting. - Identity stitching and touchpoint deduplication. - Account-based attribution across teams and channels. Pricing  Pricing is customized and not published on Adobe’s website, but generally starts around $40,000/year for basic features, and can exceed $200,000/year for large enterprises. Pros - Gold standard for B2B attribution. - Excellent CRM alignment. - Highly customizable modeling. Cons - Expensive. - Requires significant setup and data hygiene. ### 2. Dreamdata Dreamdata is explicitly designed for B2B teams that need visibility into long, touch-heavy journeys. It reconstructs account-level timelines from first engagement to closed revenue. Realize top-funnel clicks often initiate the timeline, then Dreamdata adds context by layering in website activity, email marketing, product usage, SDR outreach, and sales stages. The result is a granular, multi-touch understanding of how Realize contributes to pipeline velocity and deal creation. Special Features - Account-based journey mapping. - Cohort-level revenue analytics. - Custom models for pipeline influence. - Automated data stitching across platforms. Pricing For smaller businesses, the Foundational Plan starts at $750/month. Custom pricing is available for larger firms, but it can be up to $50,000/year or more. All paid plans include a free guided trial. Pros - Excellent for SaaS and ABM. - Strong visualization of multi-touch journeys. Robust identity resolution. Cons - Requires onboarding time. - Overkill for simple funnels. ### 3. Cometly Cometly is a lightweight yet powerful attribution tool designed for mid-market advertisers who need accurate alignment between their CRM and ad platform. It uses server-side tracking to connect Realize clicks to downstream events such as lead quality, pipeline stage progression, and closed deals. It is especially useful for advertisers who need a reliable, cost-effective alternative to enterprise tools. Special Features - Server-side conversion tracking. - CRM integrations. - Cross-channel attribution dashboards. Pricing After a 14-day free trial, pricing generally ranges between $500-$1,000/month, with custom pricing available for Enterprise-level volumes. Pros - Affordable. - Easy to implement. - Ideal for mid-market B2B. Cons - Less sophisticated than Marketo Measure or Dreamdata. Limited customization options. ### 4. Triple Whale Triple Whale is purpose-built for Shopify merchants, providing a unified performance hub for ROAS, multi-touch attribution, LTV, and cohort behavior. When you integrate Triple Whale with Realize, it ties Realize clicks and spend to Shopify purchase events, subscription renewals, and downstream customer value. Triple Whale also analyzes creative performance and identifies which ads, platforms, and campaigns contribute the most incremental revenue. Special Features - LTV-based attribution modeling. - Shopify-native cost and revenue aggregation. - Creative performance dashboards. - Pixel + server-side tracking. Pricing $149-$219/month (2 months free with annual payment), with a free Basic plan and add-ons available. Pros - Best-in-class DTC insights. - Strong creative analytics. - Easy Shopify integration. Cons - Only suitable for e-commerce. - Advanced features may require higher-tier plans. ### 5. ThoughtMetric ThoughtMetric is a streamlined attribution platform designed for e-commerce marketing teams that want fast, clear visibility into cross-channel performance. It connects Realize traffic and spend to purchases, revenue, and retention signals. The platform emphasizes simplicity, with intuitive dashboards that reveal which channels and creatives are driving the most effective growth, without requiring deep analytics expertise. Special Features - Straightforward multi-touch attribution. - Creative and campaign-level insights. - E-commerce funnel analysis. Pricing After a 14-day free trial, plans range from $99 to $1,500 a month (save 18% with annual billing). Custom pricing is available for pageviews over 3MM. Pros - Very easy to use. - Affordable. - Perfect for early or scaling DTC brands. Cons - Not as comprehensive as Triple Whale. - Limited advanced modeling. ### 6. AnyTrack AnyTrack automates server-to-server tracking and sets up Conversion API pipelines for Meta, Google, TikTok, and other platforms. It offers solutions for e-commerce, affiliate marketing, ad agencies, and lead generation. It captures Realize clicks and enriches conversion events before pushing them downstream, ensuring attribution accuracy even when browsers block cookies or users opt out of tracking. This makes it ideal for performance advertisers navigating privacy-driven signal loss. Special Features - Automatic CAPI integrations. - S2S event tracking. - Cross-funnel tracking automation. Pricing After a 14-day free trial, plans range from $100 to $300/month, with discounts available with annual billing (2 months free). Pros - Simple server-side implementation. - Great for performance marketers. - Strong automation features. Cons - Not designed for deep analytics. - Limited to tracking and passback roles. ### 7. Voluum Voluum is a powerful tracking and optimization platform favored by affiliate marketers and high-volume media buyers. It enables detailed tracking of Realize traffic, advanced A/B tests, and rule-based traffic routing. Voluum can automatically optimize workflows by adjusting traffic distribution based on performance metrics, fraud signals, or cost-efficiency thresholds. Special Features - Rule-based optimization and automation. - Split testing and path-based routing. - Bot/fraud detection. Pricing A wide range of plans for individuals and businesses. Individual plans are $119 and $299 monthly, while business plans are $539-$7,999 monthly. Discounted annual pricing is available for all plans. Pros - Ideal for large-scale performance buying. - Excellent experimentation tools. - High-speed reporting. Cons - Overkill for advertisers not running complex funnels. - Not built for B2B revenue attribution. ### 8. Hyros Hyros uses a server-side “identity spine” to track user behavior across sessions, devices, and channels. This allows it to reliably capture first-touch Realize engagement and attribute conversions that might occur weeks or months later. Hyros is popular among subscription-based and long-horizon advertisers who need rich LTV modeling and persistent identity resolution. Special Features - Server-side identity tracking. - Cross-device attribution. - LTV modeling and cohort analysis. Pricing Hyros offers a broad range of pricing for various use cases. You can get Shopify pricing as low as $69/month, paid annually, while business pricing plans range from $230-$1,499/month. Custom pricing is also available. Pros - Exceptional signal accuracy. - Reliable long-term attribution. - Strong for subscription or info-product models. Cons - Expensive. - Requires deeper technical setup. ### 9. Singular Singular is a unified marketing intelligence platform designed for advertisers who need cross-channel cost aggregation, attribution modeling, and ROI analysis. It pulls in spend and performance data from Realize alongside dozens of other ad platforms, analytics tools, and mobile partners, enabling advertisers to evaluate Realize performance in a broader context. With Singular, Realize’s click, cost, and conversion metrics become part of a complete multi-channel dataset that fuels unified marketing ROI (UM-ROI), media mix insights, and budget allocation decisions. Special Features - Unified cost aggregation from Realize, Meta, Google, Programmatic, and more. - Attribution modeling across mobile, web, and cross-platform campaigns. - Fraud detection and quality filtering tools. - Custom report builders and API access. Pricing Limited Free Plan available, but the Growth Plan is $0.05/conversion. Custom pricing is available on the Enterprise plan. Pros - Excellent for multi-channel advertisers. - Highly customizable dashboards. - Strong mobile + web + programmatic integration. Cons - Requires mature analytics workflows. - Setup complexity may be high for small teams. ### 10. Adverity Adverity is a powerful data integration and analytics platform used primarily by mid-market and enterprise advertisers who want to centralize marketing performance data in a warehouse for advanced analytics and modeling. Adverity ingests Realize API data along with social, search, CRM, and e-commerce systems to deliver a harmonized dataset that analysts can use for BI dashboards, attribution modeling, MMM, and forecasting. For advertisers who prefer to control their own attribution logic, Adverity offers unmatched flexibility. Special Features - ETL (Extract, Transform, Load) data pipelines. - Warehouse integrations (BigQuery, Snowflake, Redshift). - Automated data harmonization. - Customizable dashboards and analytics. Pricing Enterprise pricing, not published. It will vary widely depending on connectors and data volume. Pros - Ideal for advanced analytics teams. - Fully customizable data modeling. - Scales easily across global organizations. Cons - Not a plug-and-play attribution tool. - Requires internal data expertise. ### 11. Google Analytics 4 (GA4) GA4 is the foundational attribution and analytics platform for many advertisers. It collects Realize data via UTM parameters, event tags, and enhanced measurement, enabling advertisers to track Realize-driven sessions, conversion paths, and multi-touch attribution. While GA4 is not an enterprise-level attribution tool, it provides a critical baseline: how users behave after arriving from Realize, whether they engage with additional content, and how Realize interacts with other acquisition channels. Special Features - Data-driven attribution modeling. - Cross-device and cross-platform identity. - Event-based tracking architecture. - Custom funnels and audience builder. Pricing Free (GA4 Standard). Pros - Universal, widely adopted. - No cost. - Strong baseline for multi-channel visibility. Cons - Limited for advanced attribution. - Requires configuration for best results. ### 12. Adjust Adjust is a leading mobile measurement partner (MMP) that provides attribution, fraud prevention, and analytics for app marketers. When Realize drives app installs or in-app events, Adjust handles the proper attribution, SKAdNetwork compliance, and post-install tracking that Realize alone cannot measure. Mobile advertisers rely on Adjust to maintain Realize performance in line with Apple/Google privacy frameworks and to understand long-term app user value. Special Features - Mobile attribution and SKAN support. - Fraud prevention and anomaly detection. - Deep-linking and deferred deep-linking. - In-app event tracking and cohorts. Pricing Users will need to contact Adjust for a custom pricing quote. Pros - Ideal for mobile-first advertisers. - Strong fraud prevention. - Works seamlessly with Realize-driven app campaigns. Cons - Cost varies by usage. - Limited relevance for web-only advertisers. ## Key Takeaways Realize is a powerful performance-tracking and optimization tool, but you need to combine it with dedicated attribution software to fully understand your customer’s journey. For example, B2B advertisers often pair Realize with platforms like Adobe Marketo Measure or Dreamdata to link early-funnel engagement to CRM revenue. E-commerce brands use tools such as Triple Whale or ThoughtMetric to connect Realize spend to purchases, LTV, and creative performance. Server-side solutions such as AnyTrack, Voluum, and Hyros help capture Realize events more accurately in a privacy-sensitive environment. Multi-channel teams rely on Singular or Adverity to unify Realize data with other sources for holistic ROI measurement, while GA4 and Adjust provide foundational web and mobile attribution. Choosing the right attribution partner ensures advertisers can see Realize’s true impact across their entire marketing mix. ## Frequently Asked Questions (FAQs) ### How can performance advertisers ensure accurate and reliable conversion data despite privacy restrictions? As browser-based tracking becomes less reliable due to cookie deprecation and privacy constraints, advertisers are increasingly shifting toward server-side tracking and first-party data to preserve conversion accuracy. Moving measurement from the browser to secure server environments, often via Conversion APIs (CAPI), helps maintain signal fidelity, reduces data loss, and ensures compliance across platforms. Realize is designed to support this modern measurement approach. Its codeless conversions make it easy to capture key funnel events without heavy instrumentation, while integrations with Google Tag Manager and third-party tracking tools enable server-side or hybrid setups. Additionally, Pixel Audiences leverage first-party data to strengthen targeting and attribution even as third-party signals disappear. By pairing industry-standard server-side practices with Realize’s built-in data capture capabilities, advertisers can maintain accurate, resilient performance measurement in a privacy-restricted ecosystem. ### How can platforms move beyond simple conversion counting to optimize for customer lifetime value and ROAS? The advertising industry is increasingly shifting from basic conversion counting to value-based optimization, where platforms use machine learning and Value-Based Bidding (VBB) to prioritize customers who deliver higher revenue or long-term value. Instead of optimizing solely for CPA, advertisers send conversion values and customer signals so algorithms can predict Return On Ad Spend (ROAS), evaluate historical behavior, and allocate spend toward audiences most likely to drive profitable growth. Realize is built to support this value-driven approach through its Performance AI Toolkit. With Maximize Value (ROAS) bidding, advertisers can optimize directly toward high-value outcomes, while Maximize Conversions (CPA) offers efficiency-focused optimization with optional CPA targets. Realize’s Predictive Audiences further enhances this strategy by identifying users most likely to take valuable actions using machine learning models trained on source data. Together, these capabilities help advertisers move beyond volume-based metrics and align Realize campaigns with revenue, LTV, and profit goals. ### How can performance marketers proactively manage media spend and prevent wasted spend on low-quality placements? Efficient media management increasingly depends on automated protections, rule-based optimization, and transparent supply controls. Across the industry, advertisers use machine learning, historical performance data, and real-time quality indicators to pause underperforming ads, shift budgets, and block low-quality or invalid traffic. This reduces waste and improves overall campaign efficiency. Realize integrates these best practices directly into its optimization framework. SpendGuard automatically monitors recent performance to detect and minimize inefficient spend on poor-quality placements, while Custom Rules allow always-on automation to pause ads or adjust budgets based on key performance indicators (KPIs). Realize also provides granular supply controls, including site blocking, keyword blocking, and pre-bid filters, so marketers can proactively maintain inventory quality and protect brand safety. Combined, these capabilities ensure smarter, more efficient spending without relying on manual intervention. --- ### Four Ways GenAI Ad Maker Improves Creative Performance URL: https://www.taboola.com/marketing-hub/genai-ad-maker-for-image-performrance/ Last Modified: 2026-03-15 14:22:57 In performance marketing, your creative is either working or it’s costing you money. With engagement dropping off sooner than ever, marketers need a way to deliver a steady stream of fresh, high-performing assets across platforms and formats. Generative artificial intelligence (GenAI) has made it easier than ever to produce images at scale. Instead of waiting days or weeks for new concepts, marketers can now create and test original image assets on demand. The newest evolution in this space is AI-driven motion generation, which lets you apply animation and micro-movement to any static image. This capability offers a faster, more scalable way to refresh ads, boost click-through rates, and combat the creative fatigue that slows performance. Below, I’ll break down the three critical ways that GenAI Motion Ads — part of Realize’s game-changing GenAI Ad Maker product suite — empowers advertisers to create, refine, and test image and motion assets at a pace that matches today’s performance marketing demands. ## 1. Accelerated, On-Demand Asset Generation Long production timelines are a major hurdle for performance advertisers. Even a simple motion asset can demand the work of designers, animators, and editors, with each draft requiring approvals and revisions. By the time a new variation is ready, performance patterns may have already changed. GenAI Motion Ads solves that slowdown by making it quick and easy to transform any static image into a polished motion ad. Instead of relying on storyboarding, motion graphics work, or multistep video editing, you can upload a single still image and let the AI add dynamic movement in seconds. ### Instantaneous Motion Ad Creation The core innovation of GenAI Motion Ads is its ability to use AI to generate motion in a static image. This feature lives inside the AI tab, allowing you to move from concept to motion-ready creative without leaving the platform or dealing with multiple production tools. The process is intentionally streamlined: - Upload an image: You can upload a graphic or provide a URL, which makes it easy to repurpose existing brand assets. - Add prompts: You can provide short descriptions of what you’re looking to achieve. These will help guide the AI and better ensure that the finished product aligns with your campaign. - Apply motion: Once you’ve uploaded an image, the system analyzes it and applies motion intelligently, typically generating the finished product within 30 seconds. - Test your image: Your ready-to-test assets feature animation that’s strategically guided by performance patterns that historically lift engagement. AI-powered motion tools are especially impactful because motion ads consistently outperform static images. Even subtle animation can add life to an existing asset without requiring a redesign. This allows teams to extend the value of assets they already have in their media library, converting them to high-engagement formats that strengthen the testing funnel. ### Eliminating Production Lag Historically, creative teams had to choose a smaller number of motion variations due to limited bandwidth. This slowed down testing and put a cap on optimization potential. By removing that friction, GenAI Motion Ads lets you: - Refresh underperforming creatives quickly. - Build motion variants for top-performing static assets. - Produce multiple versions of an idea to compare different movement styles, which can be refined with short prompts. - Keep pace with rapid shifts in performance trends. Motion creative is no longer a special project: With tools that streamline asset creation, you can optimize continuously without slowing your workflow. ## 2. Performance-DNA: Generative AI Built on Real Network Signals While general-purpose AI tools focus on aesthetics, GenAI Ad Maker on Realize is fundamentally different because it is engineered for outcomes, not just images. Its output is directly informed by the performance patterns of one of the world's largest advertising datasets. - Trained on Engagement Signals: Unlike generic models, Realize’s AI is integrated with a performance toolkit shaped by real-world engagement signals—including over 500M impressions and 3M clicks analyzed in the Columbia study. - Built-in Creative Best Practices: The tool automatically prioritizes elements that the study identified as high-performing, such as prominent human faces, which were found to be the single most influential factor in driving engagement and building trust. - Performance Alignment by Design: Every motion and image asset is generated to align with specific campaign objectives—whether awareness or conversion—ensuring that the creative is "production-ready" for the open web from the moment of creation. - Predictive Success: By incorporating long-standing Creative Shop best practices into its algorithms, the GenAI Ad Maker produces assets that are statistically more likely to match or exceed human-made benchmarks. ## 3. Policy-Aligned and Quality-Controlled Assets High-quality creative matters, but quality without compliance likely won’t get approved, and assets that don’t meet platform policies can stall campaigns or hurt performance. GenAI Motion Ads was built to generate original assets geared toward both strong performance and adherence to advertising policies. ### Higher Performance and Quality Vetting The platform’s AI engine has undergone extensive testing, with the results demonstrating that the motion creative it generates can significantly improve performance metrics. When testing AI-generated motion against a static baseline, social trading platform eToro saw a 26% lift in viewable click-through rate (vCTR) and a 29% increase in its conversion rate (CVR). These results reinforce that motion created with GenAI Motion Ads is more than a novelty: It dramatically improves how users interact with your ads and how often those ads convert. ### Built-In Compliance and Policy Awareness Realize’s GenAI Ad Maker product suite stands out for its policy-aware design, having been trained to understand the creative guidelines governing advertising approval, ensuring that the generated output aligns with typical platform rules. That includes: - Avoiding misleading imagery, inappropriate themes, or graphics associated with restricted product categories. - Guiding layouts, messaging cues, and visual treatments toward formats most likely to pass review. - Producing creatives aligned with performance and targeting requirements, supporting compliance across Meta, Google, TikTok, and other major channels. - Conducting automated policy checks to minimize downtime from rejected ads, keeping campaigns live and performing while protecting your brand reputation. By producing assets tailored to your campaign objectives and built to comply with policies, you can avoid wasting time on ad rejections and resubmissions. The result is faster deployment, more flexible testing, and quicker learning. This built-in quality control is especially valuable if your team is managing fast-paced campaigns. You simply can’t afford the downtime that comes from asset issues. ## 4. Enabling High-Velocity A/B Testing and Optimization In modern performance marketing, testing is no longer a quarterly task: Successful marketers make testing a part of their processes. Winning ads often come from testing dozens of variations to find the right combination of format, motion style, visual composition, and messaging. The GenAI Ad Maker suite is designed to fuel that pipeline by making high-volume creation iteration easy, affordable, and fast. ### Fueling the Creative Testing Funnel The ability to generate multiple original creative assets on demand can turbocharge your testing strategies. Instead of waiting for new visuals, you can quickly spin up sets of assets that explore different motion styles, colors, focal points, or visual treatments. This ease of image creation helps with: - Combatting creative fatigue: When audiences see the same visual assets over and over, performance starts to drop off. Being able to generate low-cost variants means you can battle that fatigue and keep engagement strong. - Finding winning combinations: Pairing AI-generated visuals with different headlines or calls to action (CTAs) allows for high-velocity experimentation across every element of the ad. - Scaling quickly: Rapid testing means faster identification of top performers, which also helps you quickly iterate on what’s working. - Maximizing return on investment: More testing leads to more optimization opportunities, which leads to a better overall return on your ad spend. All of this is especially relevant if you’re already using generative AI tools or motion formats and want to deepen your testing strategies. When you leverage AI to add motion to your still images, you can generate a steady but diverse flow of creative for ongoing experimentation. ### Performance-Driven Creative at Scale GenAI Motion Ads doesn’t just speed up your processes, it also helps you produce testable creative assets that can noticeably improve your results. Each asset is generated with performance in mind, so your visuals will be closely aligned with your campaign objectives. This works well for marketers who want: - A scalable way to power continuous A/B testing. - The ability to rapidly adapt to incoming performance data. - A creative workflow without constant bottlenecks. - Assets designed specifically to drive conversions or engagement. Instead of rebuilding the creative process for every iteration, these tools create high-quality assets to feed your campaigns. This, in turn, lets you shift your focus to analyzing results and tweaking your strategy in response. Part of the GenAI Ad Maker product suite, GenAI Motion Ads represents a major shift in how creative teams approach performance marketing. Instead of relying on expensive and time-consuming manual processes, you can now generate dynamic, conversion-ready assets in seconds. This instant static-to-motion capability offers a faster route to higher engagement and stronger CVRs, eliminating the bottlenecks that once made video creative cost-prohibitive, backed by real-world results. A second, crucial benefit comes from the scalability that generative AI brings to testing creatives. Armed with the capability to produce a steady stream of fresh assets, you can run high-velocity A/B tests, fight creative fatigue, and adapt to performance trends in real time. For performance marketers, the message is clear: AI-driven creative generation is not only an advantage, it also provides a solid foundation for sustainable and scalable growth. ## Frequently Asked Questions (FAQs) ### What measurable performance benefits can I expect when using GenAI to add motion to my static images? Adding motion to static images typically leads to stronger engagement and conversion performance. Motion creatives catch attention in crowded feeds, increasing the likelihood that a user will not only notice it, but take action. Many advertisers see double-digit lifts in vCTR and CVR when they test motion variants against static assets, making it one of the most reliable ways to refresh creative and boost campaign efficiency. Realize has measured this impact directly through internal testing. Clients using the GenAI motion capability have seen measurable performance lifts. As mentioned above, eToro recorded a 29% increase in CVR when comparing motion variants generated using GenAI Motion Ads to its static images. These improvements help advertisers drive more conversions without increasing spend. ### How can I efficiently scale motion creative for testing and iteration without massive production costs? Generative AI tools make it possible to scale motion creative quickly and affordably by automating the production process. Instead of working through lengthy design timelines, your team can instantly generate creative variants or animate existing still images. This reduces your reliance on specialized design teams, effectively lowering the overall cost of producing each testable asset. The biggest benefit of this is that you can continuously experiment without stretching your limited resources. Realize offers the GenAI Ad Maker product suite — which includes GenAI Motion Ads — to make it easy to instantly produce motion variants from existing assets or text prompts, enabling rapid testing at a fraction of traditional costs. With Realize, teams can maintain a steady flow of fresh creative for experimentation without adding headcount or outsourcing animation work. ### How do I ensure the images I create with AI are ready for immediate use and testing? The key to producing production-ready creatives is to use AI not just for asset output, but also for performance alignment. The most effective tools generate visuals that match your campaign goals, whether you’re aiming for awareness, traffic, engagement, or conversions. This ensures each asset is ready to use as soon as it’s produced. When your outputs are built to perform, they can feed directly into your A/B testing environment without requiring heavy editing. Realize’s GenAI Ad Maker integrates with its performance toolkit. That means every generated asset is shaped by the same performance signals that guide campaign delivery, which ensures that your creative aligns with your advertising objectives. This keeps your entire optimization cycle strong with a steady stream of creatives for you to test. --- ### 5 Ways Realize Will Scale Performance on the Open Web in 2026 URL: https://www.taboola.com/marketing-hub/how-realize-scales-performance-on-open-web/ Last Modified: 2026-01-09 14:09:13 It’s another new year, with lots of new opportunities in the world of digital advertising. As you’ll have noticed, the market is more fragmented than ever, with overcrowded, expensive social media and search channels and fatigued users. For advertisers and performance marketers, taking a fresh approach is essential to grow audiences and conversions. Instead of reusing the same playbook, the open web offers a path for advertisers to reach high-intent users where they actually spend time, whether that’s reading news, researching products, or engaging with content. Modern platforms offer lots of options for advertisers to deploy their resources more intelligently and strategically than in years past. The new year is a good time to test new platforms as you increase incremental reach and optimize for performance. That’s where performance advertising platform Realize can help. Realize incorporates deep machine learning and robust first-party data to serve as the go-to solution for brands that demand measurable growth. Customers report increases in return on ad spend (ROAS), reach, and time on site, along with reduced cost-per-click numbers and other essential performance metrics. Here are five reasons why Realize should be at the heart of your 2026 media mix, with details from users on how they’ve refined their approach with Realize and found success on the open web. ## 1. Capture Attention and Increase Reach and ROAS with AI-Powered Bidding Whatever your industry or marketing goals in 2026, efficiency is likely top of mind. Instead of guessing at what ad spending will be weekly, monthly, or quarterly, try a better way. Realize’s Maximize Conversions bidding technology adjusts bids automatically in real time, so you can meet specific goals and use money wisely without overspending. Realize users see improved reach and ROAS by using this feature. Case in Point: Hospitality group Minor Hotels activated AI-driven native ads and automated bidding in Realize as part of their strategy to drive property bookings beyond search and social. Minor Hotels optimized for high-intent travelers and achieved 5x ROAS using Realize’s features. Likewise, social publisher and entertainment network Ströer achieved a 3x increase in reach when they adopted Realize’s Maximize Conversions feature. They also saw a 13% increase in page views per session and a 2x increase in revenue per click. ## 2. Triple Your Lead Generation While Reducing Costs Lead acquisition on the web happens on an entirely different scale from search engines, which can easily stagnate or return mismatched leads. With Realize, it’s possible to quickly test thousands of creative variations and targeting segments to see what’s working and choose the most efficient way to capture a lead, without going over budget. Built-in generative AI tools mean you can use Realize to test copy, CTAs, and creative separately or in different combinations, to choose the right variations for the audience. Realize users can better target audiences and test hypotheses faster. Case in Point: Security provider Verisure chose Realize to expand their reach in Latin America. They used both native and display formats with ABBY, Realize’s gen AI creative tool, and tripled lead volume while reducing cost per lead (CPL) by 69% over five months. ## 3. Reach High-Intent Audiences Beyond the Typical Walled Gardens Intent is essential when you’re looking for new leads and conversions, and it’s often missing from social and search channel advertising strategies. Open web users are generally in discovery mode, rather than mindlessly scrolling on social media channels. Realize uses first-party behavioral signals to find customers who are ready to engage, based on what they’re actually consuming. Realize users end up with better leads that align with their performance goals. Case in Point: Energy solutions provider Livguard aimed to find and engage audiences with true consideration intent, rather than the shallow traffic they’d seen from social and over the top (OTT) content delivered directly to users. Livguard chose Realize, and found that users coming from Realize stayed on the site 16% longer than those from paid search, and that those users took 29% more actions per session than those from social media traffic. For TripAdvisor’s Cruise Critic review platform, using Realize to diversify their media mix led to a 67% lower cost per click (CPC) than their Meta advertising. Realize was also 78% more efficient than Cruise Critic’s CPC benchmark. ## 4. Outperform Other Performance Channels Marketers and advertisers have spent years trying to eke out small gains as legacy performance channels grow oversaturated and user journeys become more complex. Taking better advantage of the open web’s opportunities can lead to greater growth numbers that meet or exceed business goals: Realize consistently proves its worth by delivering lower CPAs and higher returns than legacy performance channels. Case in Point: PortAventura World, one of Europe’s largest leisure destinations, used Realize to increase hotel bookings. They chose the Taboola Pixel for precision retargeting and automated bidding, and achieved a 44% higher ROAS and a 47% lower CPA — that’s compared to all their other performance marketing channels in 2024. ## 5. Boost Engagement and Revenue Through Better User Experiences While performance marketers can find a lot of new success with Realize features, it’s also designed to help create better, more seamless user experiences. Realize serves ads that feel like a natural part of the content that users are already consuming, leading to higher engagement rates and increased brand affinity. Case in Point: Broadcasting network Channel A chose the continuously scrolling Taboola Feed, powered by Realize, to reverse a decline in monthly active users and also increase revenue. The Feed brings a non-disruptive, personalized experience to users, and Channel A saw a 60% increase in RPM and a 40% increase in CTR with Realize. Meanwhile, grocery store chain Lidl Hellas, along with Project Agora, grew their user engagement metrics when they implemented an always-on Taboola campaign. They saw a 3.6x above average CTR, an increase of 548% pages per session, and a growth of 442% average session duration after choosing Realize. ## Your New Year’s Resolution: Scale Smarter in 2026 So, which increased performance metrics will you celebrate at the end of 2026? Whether you’re looking for a 5x ROAS like Minor Hotels, a 69% lower CPL like Verisure, or deeper engagement like Livguard, Realize offers the tools to win on the open web. Rather than sticking with oversaturated, underperforming search and social channels, find performance and scale where engagement is highest. --- ### 3 Key Elements to A/B Test for Maximum Conversion Uplift URL: https://www.taboola.com/marketing-hub/ab-test-key-elements/ Last Modified: 2026-05-24 07:56:16 When you’re running a digital advertising campaign, there’s a ton of choices to make — creative, strategic, and design to name just a few — and it can get overwhelming right from the start. Optimizing digital ad and landing page performance is less about luck and more about data-driven testing, specifically, forming a comprehensive strategy focused on A/B testing the three most impactful ad elements: headlines, images, and calls-to-action (CTAs). Why these three? Well, first off, as a copywriter, I can tell you that they’re the highest-visibility components that define a visitor's first and last impression. That matters, because a well-executed test focused on these critical variables can deliver statistically significant results, leading to maximum conversion rate uplift and better return on investment (ROI). ## The Core A/B Testing Process A/B testing is a method of comparing two or more versions of a single element to see which performs better against a specific goal, such as conversion rate or click-through rate (CTR). It may seem like a simple concept, and it kind of is at its core, but it’s an absolute game-changer. A/B testing is a structured process that helps remove guesswork from your optimization strategy, and that’s worth a lot. Here are the steps to doing it effectively: ### Formulating a Strong Hypothesis Every successful test starts with a clear hypothesis — a prediction about how a specific change will affect user behavior. A strong hypothesis has three parts: - Identify a clear problem or challenge: For example, "The sign-up rate for our ad is low." - Offer a precise solution: The change you will make, e.g., "Changing the CTA button color from blue to green." - Describe the expected impact: The predicted effect on user behavior, e.g., "This change will increase click-through rates by 10%." ### Setting Up the Test To ensure your test gives you reliable data, follow these guidelines: - Test one element at a time (isolate variables): This is crucial. If you change both the headline and the image simultaneously, you won't know which change caused the impact. - Establish a clear goal and baseline metric or key performance indicator (KPI): Define exactly what success looks like (e.g., conversion rate, CTR, or bounce rate). - Run the test for a sufficient amount of time: Testing typically requires two to four weeks to gather enough data for statistical significance and accurately measure the true conversion lift, filtering out daily fluctuations. ### Analyzing Results and Scaling Wins The final step is translating raw data into actionable growth: - Compare the performance: Look at the variation against the control (original version) across your defined KPI. - Look for a statistically significant uplift: Ensure the difference in performance (e.g., the 10% increase in CTR) is reliable, usually requiring a 95% confidence level or higher. Marketers can ensure this by using online statistical significance calculators or relying on the built-in reporting tools of modern ad platforms, which often provide the confidence score as a percentage, or flag a test as statistically significant. - Implement the winning version and scale: Once validated, replace the original ad with the winner and immediately start a new A/B test to continue the cycle of optimization. ## Headlines: The First Impression Test Headlines are the large, lit-up sign over the main entrance to your ad or content. ### Why Headlines Matter If the headline misses, most users won't stick around to click on anything beyond it. Testing the headline is critical because it's the first thing your audience sees, capturing attention and enticing them to engage with your content. ### What to Test (Variables) - Length: Try a longer, more descriptive headline versus a short, punchy one. - Tone/emotion: Test headlines that express negative or positive emotions, or try posing a direct question to the user. - Content: Test different value propositions, or try incorporating a testimonial directly into the headline for social proof. - Clarity: Ensure headlines clearly and instantly communicate the value of your content to the casual scanner. ## Images: The Visual Hook Test Before even reading the copy, images and visual assets are often the first elements users take in. When everything aligns, testing them can drastically improve click-through rates and directly increase sales. ### Why Images Matter Images are the primary visual hook that captures attention in a busy feed. Testing the right visual elements ensures that the "look" of your ad immediately resonates with your target audience. ### What to Test (Variables) - Relevance and seasonality: Test images to ensure they are aligned with the target market’s current mood, circumstance, or season. - Diversity and representation: Make sure your images reflect an equal representation of diversity — especially within your target audience, but also as a good general practice. - Product focus: Use at least one clear image of the product in use (Hero Product Image), showing the positive benefits, rather than just the object. - Visual type: Test a static image against a short, dynamic video, or against a specialized Motion Ad format (short, dynamic, motion-based creative) to capture more attention. - Visual appeal: Test images that feature "pretty people." Seriously — studies suggest people are more likely to trust attractive individuals. ## Calls-to-Action (CTAs): The Action Trigger Test The CTA is the moment of truth. It's an action that needs to inspire the customer to take charge, filling in the final blank on whether a user will convert. ### Why CTAs Matter The CTA is your ad's final instruction. Even a perfect headline and image will fail without a compelling and clear action trigger to guide the user to the next step. ### What to Test (Variables) - Language/text: Use active, powerful verbs, or try clear action phrases against value-first messages (e.g., "Shop Now" vs. "Get My Free Guide"). - Color/style: Test different color palettes to find one that stands out against the website's color scheme, ensuring the button is highly visible. To go even deeper, research the effect that colors can have on users. Choosing a bright red can make a noticeable difference than, say, a soothing blue. It’s worth testing and seeing what works. - Placement: Test the position on the page, such as putting the CTA immediately after the value proposition, versus placing it after social proof or testimonials. ## Enhancing the A/B Testing Lifecycle Manual A/B testing is powerful, but it's slow. To achieve the fastest conversion uplift, modern marketers integrate artificial intelligence (AI)-powered optimization tools that speed up every step of the testing process. ### Leveraging Campaign Performance Data After running a test, analyze the data — not just for the winner, but for specific audience segments. For example, an ad showing a 5% overall conversion rate might show a 15% conversion rate (CVR) among users who clicked through from a financial news site, but only 2% from a sports blog. This insight allows you to stop running the ad on the sports blog (an underperforming segment) and scale up the ad on financial sites. Use this data to refine targeting and optimize ad spend across your entire campaign ecosystem. ### Creating Multiple Creative Variations Quickly Manual testing is limited by the time it takes to design new assets. That’s where tools like Realize are game-changers, as they allow you to use AI to quickly generate and optimize creative variants. New variants enable faster A/B testing on different visual and copy combinations. ### Streamlining Test Setup and Optimization An AI-powered assistant can also automate campaign setup and media planning, applying best practices right from the start. This allows you to quickly create multiple landing pages and ad variations — often using simple prompts — to enable faster A/B testing without the tedious manual setup. ### Automating A/B Test Outcomes The fastest way to scale a test is through automation. Employ Custom Rules to automatically pause underperforming ads and scale successful campaigns, allowing the learning from A/B tests to be instantly applied to optimize budget and performance. This ensures you're always running the highest-converting variations. ## Key Takeaways Focusing your A/B testing efforts on headlines, images, and CTAs offers the most direct route to boosting your conversion rates (uplift). By using a structured, data-driven approach on these three high-impact components, marketers can drop the guesswork and achieve measurable, dependable results. This continuous cycle of testing and analysis, particularly when amplified by platforms like Realize, is essential for maximizing your return on investment (ROI) and fueling business growth. ## Frequently Asked Questions (FAQs) ### How can I quickly generate creative variants for A/B testing? You can manually create variations or use third-party AI tools for basic design and copy generation. This often involves juggling multiple tools and manually uploading assets into your ad platform, which can be time consuming, and slow down your testing cycle. Within the Realize platform, the GenAI Ad Maker creates and optimizes creative variants directly. This tool utilizes generative AI to swiftly make free and original creative assets, including images and titles, based on current network best practices and your campaign goals. It enables you to generate multiple variations from a single prompt or image in seconds, achieving enhanced performance and results instantly. ### How can I test creative changes in a low-risk environment before launching them widely? Use A/B testing platforms to split a small portion of your live traffic (e.g., 50/50 split) between the control and the variation to minimize revenue risk. While effective, this still exposes your campaign to potential losses if the new creative performs poorly, and it requires a long waiting period for statistical significance. With Realize, the Performance Simulator (currently in beta) allows you to model and reduce the risk of budget changes by testing different scenarios and gaining visibility into expected conversions before adjusting budgets. This unique tool provides predictive testing before scaling, using your campaign’s historical data (with at least four spending days in Maximize Conversions) to forecast the impact of budget increases or decreases. Doing this helps you find the optimal investment level with a confidence score before you expose your campaign to any real risk. ### How can I simplify the process of setting up conversion tracking for my A/B tests? You typically need to add specific code (e.g., event snippets or URL tracking parameters) to your website or landing pages to measure conversions accurately. This process often requires technical knowledge (or a developer) to implement the Taboola Pixel and then add subsequent event code snippets to track specific actions, like form fills or downloads. The Realize platform offers a Codeless Conversions solution to simplify setup. Once the basic Taboola Pixel is implemented, you can create event and URL-based conversions directly in the platform without technical setup. This tool is crucial for A/B testing because it ensures you can quickly and accurately track key metrics — such as button clicks or page visits — without delays from coding, enabling faster optimization decisions. ### What kind of visual formats can I A/B test for performance uplift? You can test static images, GIFs, and videos on your ad creatives and landing pages. Standard A/B testing usually compares images against each other or against longer video assets to see which style is most engaging for your audience. With Realize, you can A/B test a wide range of formats, including Native Ads, Motion Ads, and Vertical Ads, all within the platform. Beyond standard static Native Ads (which blend seamlessly), you can test Motion Ads (short, dynamic, motion-based creatives that drive higher CTR) or Vertical Ads for mobile-first, full-screen impact. You can even use the GenAI Ad Maker to turn a static image into a Motion Ad with a text prompt, rapidly generating a new format variant for your test. --- ### Static Images vs. Motion Ads: Strengths, Weaknesses, and How To Know What to Go With URL: https://www.taboola.com/marketing-hub/static-images-vs-motion-ads/ Last Modified: 2026-04-23 11:19:30 It’s a performance-driven ad landscape out there, and creative formats have become just as critical as channel selection and audience targeting these days. As marketers push beyond overly saturated search and social environments, they need formats that not only capture attention but also convert — efficiently, repeatedly, and at scale. That’s where the two dominant creative formats stand out on the open web: static image ads and motion ads. Each plays a unique role across the funnel, yet not all formats provide equal conversion impact. Let’s explore how these formats perform, why motion is increasingly outperforming static, and what tools performance advertisers now have at their disposal to optimize for higher conversion rates (CVR) — especially in environments where costs are rising, attention is fragmented, and traditional performance channels are becoming less reliable. ## Why Creative Format Matters More Than Ever The choice of ad creative isn’t just a matter of preference, it’s an important performance lever, particularly for marketers operating on the open web. Here are a few reasons why your format choice is crucial: - Artificial intelligence (AI) disruption to search is reducing dependability. - Rising costs-per-acquisition (CPAs) and costs-per-mille (CPMs) are forcing advertisers to squeeze more value from each impression. - Performance marketers need formats that convert — not just reach audiences. - Attention scarcity is at an all-time high. ## Static Images: Strengths, Limitations, and Best Uses The static image ad is the workhorse of digital advertising, defined by its simplicity and universal acceptance. It’s a classic, like Coca-Cola’s instantly recognizable red “Share a Coke With…” ads, or Chik-Fil-A’s grammar-challenged cows, but it has its strengths and weaknesses: ### Benefits of Static Image Ads - Easy to produce and deploy across channels. - Ideal for product-led messages and direct response. - Widely supported across ad networks. - Strong familiarity = faster approvals. ### Where Static Falls Short - Lower attention capture vs. motion. - Increasing banner blindness. - Limited emotional impact. - Fatigue sets in faster without variation. ## Motion Ads: The New Creative Performance Standard Motion ads, which include short video and animated GIF formats, are quickly becoming the new baseline for performance on the open web, including on platforms like Realize. ### Why Motion Ads Drive Higher CVR Motion ads are built to address the attention deficit that plagues static formats: - Dynamic movement captures attention faster. - Communicates value fast — ideal for mid-lower funnel. - Improves click-through rate (CTR) and CVR. - Drives engagement and scroll depth, improving audience qualification. ### Motion Ads in Performance Use Cases Motion ads are a powerful choice for driving conversions, since they grab attention fast and clearly explain value. That’s exactly what people need when scrolling through content. This dynamic format is excellent at getting users to pause and click, transforming casual viewers into qualified leads by providing quick, engaging explanations. They also consistently boost CVR and CTR by making the ad's benefit obvious in a matter of seconds. A couple examples showing how major companies made use of motion ad strategies and saw measurable performance results: 1. One Zero Bank The Goal: As Israel’s first digital-only bank, One Zero Bank needed to expand its reach and acquire high-quality leads in a competitive financial market while maintaining strict cost-per-acquisition (CPA) targets. How Motion Ads Demonstrated Effectiveness: - Lowering Performance Costs: By utilizing motion ads, the bank achieved a 40% lower CPA than their target price. The looping visuals captured attention more efficiently than static images, allowing the campaign to scale while remaining cost-effective. - Driving Higher Conversions: The "thumb-stopping" nature of motion assets led to a 20% increase in conversions. This suggests that motion ads are highly effective at moving users from the "awareness" stage to "action" in a financial context. 2. Motor Culture Australia The Goal: This automotive enthusiast community aimed to drive new member sign-ups and entries for their high-value car giveaways, seeking to outperform their results on traditional social media channels. How Motion Ads Demonstrated Effectiveness: - Outperforming Traditional Social Channels: Motor Culture Australia found that motion ads delivered the highest ROAS (Return on Ad Spend) in their marketing mix, beating out both Facebook and Google. - Boosting Account Creation: The campaign resulted in a 27% increase in sales and a 35% increase in unique accounts created. - Visual Storytelling for High-Value Leads: Because their offer involved visually stunning custom cars (e.g., Audi RS6, Ford Ranger Raptor), the motion format was essential for showcasing the excitement of the prizes, leading to more engaged, high-intent users. 3. Hyundai The Goal: Hyundai sought to promote its IONIQ 5 electric vehicle by maximizing brand exposure and generating qualified leads in a crowded automotive market. How Motion Ads Demonstrated Effectiveness: - Balancing Awareness and Performance: Hyundai used motion-based native advertising to solve the "awareness vs. lead gen" dilemma. The format allowed them to drive millions of impressions while simultaneously securing hundreds of cost-effective conversions. - Improved Retention and Recall: In related digital-first campaigns using short-form motion content (like TikTok-style "react" ads), Hyundai saw an 18% increase in 6-second view-through rates and an 8.5x increase in ad recall compared to industry benchmarks. - Engagement-to-Conversion Pipeline: The motion assets acted as a bridge, engaging users with dynamic visuals of the car’s features and then funneling them toward high-value actions like the brand's first-ever interactive auto configurator. ## Real Performance Gains: Static vs. Motion on the Open Web The performance difference between static and motion is increasingly pronounced, and this lift isn't limited to brand awareness. - Motion formats increasingly outperform static image ads. - Lift doesn't only sit at the top of the funnel — conversion impact is measurable. - Works especially well when paired with AI-driven optimization and first-party data targeting. - Motion plus high-intent audiences equals some of the highest converting combinations at scale. ## How Modern Performance Platforms Power Better Creative Outcomes Platforms like Realize address the traditional hurdles associated with motion ads — namely, production complexity and optimization difficulty — making them accessible and scalable for all performance marketers. ### Motion Ad Creation at Scale Realize provides tools to make motion simple: - Ability to auto-generate motion from static assets: Advertisers can easily convert existing static images into dynamic, engaging Motion Ads using Motion Ads Studio, without needing external video production. - Creative studios and GenAI support: GenAI Ad Maker offers robust support for creating performance-ready motion and static assets instantly, ensuring a constant supply of fresh creative. - Built-in compliance and instant launch workflows: The platform automates checks and streamlines the process to get high-performing motion ads live quickly. ### Smarter Audience Qualification with Motion Realize uses AI to maximize the value of every motion impression through: - Real-time CVR optimization: The platform continuously adjusts bidding and delivery to optimize for the ads and audiences most likely to convert. - Engagement-based signals: The platform reads how users interact with the motion creative (e.g., watch time, engagement) as a signal for purchase intent, improving qualification. - Motion creative mapped to purchase intent: The AI ensures the highest-performing motion ads are shown to the most qualified users. ### AI-Driven Bidding to Maximize Motion Performance Realize's bidding strategies are designed to boost CVR with benefits like: - Maximize conversions (with or without pixel): Automated bidding aims to drive the maximum number of conversions at the most efficient price. - Maximize value (ROAS-led bidding): For advertisers focused on revenue, the system optimizes for return on ad spend (ROAS). - Automated pacing plus SpendGuard to reduce waste: These features ensure budget is spent efficiently on high-converting motion impressions, minimizing wasted dollars. ## When to Use Static, and When to Use Motion Choosing the right format depends entirely on the campaign goal, the audience stage, and the resource budget. Here’s when it’s recommended to use each. ### Use Static Ads For: - Fast deployment. - Single-claim messages. - Budget-limited testing. - Evergreen, lower-funnel retargeting where the user is already highly qualified. ### Use Motion Ads For: - New product launches. - Consideration-stage users. - Audiences requiring persuasion. - Converting high-intent users at scale. ## Key Takeaways Choosing between static image and motion ad creative isn’t about selecting one format over the other, but rather about deploying the right format at the optimal time to achieve maximum conversions. While static images provide reliability and are easier/faster to produce, motion ads consistently demonstrate superior engagement and deliver more efficient conversions. For advertisers struggling with high costs and creative saturation on traditional channels, motion creative offers a significant advantage: it captures more attention, drives deeper user interaction, and ultimately secures more conversions. It’s also important to note that the integration of modern tools like AI optimization, automated bidding systems, and GenAI creative generation (all offered in Realize) has substantially lowered the barrier to entry for motion ad production and scaling. In a competitive performance environment where minimizing wasted spend is crucial, motion is evolving from a high-cost upgrade to the new performance default for achieving sustained, conversion-led growth. ## Frequently Asked Questions (FAQs) ### Do motion ads always outperform static images in terms of conversion rate? Not always. Dynamic creative doesn't automatically guarantee superior results in every instance: Ad performance is dependent on a whole bunch of contextual factors, including the specific audience segment being targeted, the placement of the ad (in-feed vs. sidebar, for example), and the consumer's position within the marketing funnel. However, as an overall general rule, motion ads are inherently better at cutting through the noise and capturing a user's attention in crowded environments. Getting that higher initial engagement typically translates into a better-qualified click and a subsequent uplift in conversion rates. As mentioned, motion ads running on Realize generally experience higher CVR compared to their static counterparts. It’s a notable performance boost that’s achieved because Realize's proprietary technology is designed to maximize the format's potential. Realize turbo-charges high-engagement creative by combining it with sophisticated AI-driven bidding and targeted audience qualification, ensuring that the dynamic ad is delivered specifically to users who have a high-intent profile and are most likely to complete a purchase. ### Are motion ads harder or more expensive to create? Historically, motion creative was considered a greater challenge and a higher expense, and demanded specialized graphic design or video production teams, leading to lengthy production cycles and substantial resource allocation. The complexity of the process often limited motion's use to large, high-budget brand campaigns rather than day-to-day performance marketing, but that’s changing rapidly. Realize has basically eliminated those traditional difficulties of motion ad creation, making it a scalable option for all performance marketers. The platform offers seamless capabilities to transform static images into engaging, dynamic motion ads automatically using the Motion Ads Studio. Additionally, the built-in GenAI Ad Maker provides immediate support for rapidly generating various high-quality performance assets. It’s this integration that allows advertisers to easily and cost-effectively scale their dynamic creative testing without racking up the traditional costs or delays associated with external video production. ### How can I test whether motion ads really work better for my campaigns? To determine the true value of motion for your specific product or service, you need to implement a rigorous A/B testing framework. That involves isolating the creative format variable by testing the static version against the motion version within identical audience segments. After that, you’ll need to measure key performance indicators (KPIs) like CVR, cost-per-click (CPC), and CPA to draw statistically significant conclusions about which format drives more efficient results. Realize provides a comprehensive suite of automated testing tools that streamline this process. You can conduct predictive testing using the Performance Simulator before allocating large budgets, ensuring you only scale high-potential formats. You can also utilize the Maximize Conversions bidding strategy to allow the platform's AI to automatically identify and prioritize the highest-converting formats in real time. Real-time reporting paired with the ABBY assistant offers automated recommendations on which specific creatives should be paused or scaled. All of this testing is done without disrupting or pausing your existing, live campaigns, ensuring a continuous (and data-informed) optimization cycle. --- ### Three Ways Motion Ads Drive Higher ROAS URL: https://www.taboola.com/marketing-hub/motion-ads-higher-roas/ Last Modified: 2025-12-30 10:43:04 Today’s performance marketers face a familiar but growing series of frustrations: The creative well is running dry, costs keep climbing, returns keep shrinking, and the sparkle of static images — those tireless workhorses of digital advertising — has begun losing its shine. Audiences scroll past without a second thought, creative fatigue is hard to avoid, and campaigns risk stalling before they even scale. Is it a media problem or an algorithm glitch? Neither: It’s a creative crisis. Static visuals once did the job, but they can’t keep pace with the attention economy. A photo that pops today is relegated to wallpaper tomorrow. That fatigue limits engagement, caps growth, throttles ROAS, and turns optimization into a losing battle. The truth? Performance marketing can’t rely on one-dimensional creative to drive multi-dimensional results. Enter motion. Dynamic ad formats do more than move: They perform, capturing attention before a thumb can swipe, deliver more story per second, and bridge that gap between art and analytics. These ads also unlock a smarter framework for growth — one built on the pillars every marketer should master, i.e., creative power, scaling efficiency, and AI optimization. Movement has evolved beyond aesthetics to strategy. ## The Challenge: Static Fatigue and Stalling Conversions Performance marketing doesn’t typically fail overnight — it slowly stalls over time. Initially, a sharp, static image and a robust offer look like winners. Then, as frequency rises, results soften, and suddenly that same top performer quietly begins to drag down your ROAS. Creative fatigue has crept in, and the algorithms merely reflect what your audience has already told you: Your ads have become overly familiar, and they’ve stopped “seeing” them. ### Static Fatigue in the Feed Static banners and images have become the baseline of the attention economy, but rich, motion-based formats routinely deliver higher engagement and click-through rates than static ads because they feed the eye and brain. Studies on animation and motion in advertising show that moving creative is up to 3x more effective at capturing and re-capturing an audience’s attention than static visuals, which quickly become background noise once viewers have seen them a few times. Static ads hit creative invisibility quickly. CTR declines, engagement drops, and platforms issue warnings about fatigue or limited creative performance. Even when your targeting is spot-on, the same repeated image trains users to scroll right past it, diminishing your effective reach and forcing you to pay more (yet still be ignored). ### Conversion Plateaus and Rising CPAs When creative stops evolving, it struggles to move users through the funnel, from a glance to real consideration to actual purchase. Because they’re often limited to a single frame and a single moment of persuasion, static formats make it hard to answer objections, show value, and nudge someone to take that final step across the finish line. By contrast, motion and video formats can drive 20% to 30% higher click-through and conversion rates, underscoring how their dynamism translates into more people taking action. As static performance erodes, CPAs climb. Fatigued creative delivers fewer clicks, lower-quality traffic, and weaker conversion rates, so each acquired customer costs more than the previous one. Well-executed motion or video ads can: - Drive several times as many clicks as their static counterparts. - Reduce cost per click and cost per lead by hundreds of percentage points (directly improving ROAS). A study conducted by Ocean and Neuro-Insight found that full-motion digital out-of-home (DOOH) ads deliver 2.5x the impact of static ads. ### Why Static Alone Can’t Scale Welcome to the trap for performance marketers: What used to be a “safe” static workhorse has become an invisible budget leak at scale. Once fatigue sets in: - Platforms down-rank the creative. - Frequency rises. - You pay more for impressions that generate fewer conversions. The combination of declining CTR, rising CPA, and flat or falling conversion volume strangles growth and forces teams into constant, reactive tweaks instead of strategic, scalable optimization. In that context, hunting for a new audience or better bid strategy becomes a band-aid for the symptoms. Without more dynamic creative that can reset attention, refresh engagement, and send AI more high-quality signals to optimize against, static-heavy accounts become stuck in a loop: - More spend. - More fatigue. - Less return. Suddenly, motion-based formats start to feel like less of a “nice-to-have” and more like the only sensible next step. ## The Strategic Solution: Unlocking Performance With Motion Is motion a “nicer” format? Sure. But, it’s also the performance lever tying creative, operations, and AI together. Dynamic motion ads (short, narrative animations built from the same assets you already use) give campaigns what static alone can’t — a way to: - Consistently win attention. - Move people from curiosity to action. - Feed modern bidding algorithms with richer, higher-quality engagement signals. Want real results? Don’t try video as a one-off experiment. Use motion as the creative backbone of your creative ecosystem, and reap the benefits. ### Benefit 1: Elevate Conversion Rate Through Dynamic Storytelling Who doesn’t love a story? Motion formats bring static images to life, transforming a single frame into a short sequence that conveys context or use in a few seconds. That extra moment of story — before/after, product-in-use, social proof, or value proposition — helps users understand why the offer matters. It’s exactly what you need when optimizing for consideration and conversion, and not just impressions. Because motion is built for these mid- and lower-funnel goals, the clicks it earns tend to be more intentional. Advertisers using animated or motion-based formats report higher engagement and stronger conversion lifts versus static alone. Retargeting already increases ad engagement rates by 400% in some cases, and has an efficiency rate of over 500%, so imagine the results you could achieve by adding motion to retargeting or direct response flows. In practical terms, the same budget is driving fewer empty clicks and more people primed to complete a purchase, pushing CVR up instead of slowly fading over time. ### Benefit 2: Unlock Production Efficiency to Scale Testing A common fear is that “motion” means big shoots, long timelines, and a full (and expensive) video team, but modern performance setups look very different. Many platforms now offer lightweight motion ad studio tools that can automatically assemble existing static images, headlines, and brand elements into native motion units sized for each placement. You use the same asset library of product shots, lifestyle images, and brand colors, while the system handles transitions, pacing, and animations that work with your feed. Once you template your motion ads, testing becomes easier. You can spin up multiple motion variants from a single base image, changing sequences, overlays, CTAs, and value propositions, and launch A/B tests in days, not weeks. No separate video production element needed! That efficiency keeps a constant stream of fresh, optimized creative in rotation, which is critical for fighting fatigue, protecting CPAs, controlling budgets, and sustaining scale. ### Benefit 3: Maximize ROAS With Performance AI Alignment You realize the final benefit when motion creative meets performance AI and value-based bidding strategies. Automated systems work best when they have clear, consistent signals about which impressions drive valuable outcomes. Motion ads supply those signals by generating stronger engagement and more meaningful post-click behavior per impression. AI then uses those more robust signals to prioritize serving motion-led placements to users showing the highest likelihood of converting at a profitable value. Here are a couple of case studies: 1. Matas Retail Performance Growth: Optimizing for Value Through the strategic use of motion creative paired with high-intent audience targeting and Maximize Value bidding, a leading retailer achieved substantial performance gains. - Initial Goal: To validate motion creative as a primary performance driver for lower-funnel objectives. - Quantifiable Achievements: The brand achieved a remarkable 124% increase in ROAS over a five-month period. 2. Chery Automobile: Boosting Lead Generation and Engagement Chery, a China-based automobile brand, implemented a multi-format content promotion strategy to elevate their campaign results on the open web. - Initial Goal: To improve lead generation and engagement through a diversified mix of Image, Motion, and Video Ads. - Quantifiable Achievements: 12% higher CVR on days featuring video ad content. - 35% lower average CPA on days when video ads were active. - 25% higher average vCTR compared to standard benchmarks. ## Key Takeaways Performance marketers have hit a wall with static creative. This creative crisis requires a permanent fix, and motion delivers completely: - Sharper engagement that pulls users into the story. - Production tools that simplify scaling. - AI signals that turn automated bidding into an ROAS machine. What may start as a simple animation can rewrite your entire performance equation. Static images cap out fast, but motion resets attention, fuels endless testing variants, and gives platforms the dynamic data necessary to optimize for value over volume. Campaigns that integrate motion can see conversion rates climb by 20% or more. ## Frequently Asked Questions (FAQs) ### What’s the typical performance lift I can expect when switching from static images to motion ads? Motion ads pack a punch over static ads, grabbing attention with movement that static can’t match. Brands that switch strategically often see an engagement jump: 87% of marketers report that motion ads, such as video, generate more leads, drive more sales, and increase conversions by up to 80%. Realize elevates your results with crystal-clear reporting that instantly compares the ROAS of motion ads to your static benchmarks. That immediate visibility and insight empowers you to shift strategy or budget to winning motion creative quickly, so you can realize benefits like those of Chery and other companies featured in our case studies. ### I don’t have a large video team. How can I efficiently scale motion ad creation for rapid testing? No video squad? No problem. Modern platforms let you spin motion from existing product shots and other assets, creating optimized loops or clips for quick A/B tests without the production grind. With these platforms, you can turn static assets into variants that test hooks, CTAs, and flows in days (not weeks), to ward off fatigue and keep scale growing. Realize’s GenAI AdMaker suite is purpose-built for iteration. It effortlessly transforms static images into motion formats inside the platform. The suite takes core brand assets (colors, logos, copy, and existing top-performing ads) as input, then uses machine learning and genAI to generate a high volume of new ad variations. It alters elements like visuals, video clips, voiceovers, music, headlines, and CTAs for motion ads. Then, it automatically deploys these ads to target platforms, allowing you to test dozens of variations from one asset library. You’ll slash time-to-market, eliminate the need for expensive teams, and still gain agile, data-driven iteration. ### How do I ensure my motion ads drive purchases — not just clicks? Motion shines at engagement, but purchases demand pairing it with smart, conversion-tuned bidding that reads user signals and creative performance. These systems spot high-intent moments, analyze user behavior and engagement, and serve your best motion to people primed to buy, turning scrolls into sales. Realize integrates your motion creative’s performance with SmartBid (conversion-optimized bidding). SmartBid bids higher for users most likely to convert, and motion ensures the creative shown is the best possible version to drive that conversion, creating a data loop that accelerates scaling and improves ROAS by aligning automated bidding with creative effectiveness. The platform prioritizes top performers among users most likely to purchase, so increased engagement generated by the creative drives higher conversion rates and ROAS. --- ### The Human Edge: Why Expertise Trumps Self-Service in Complex Channels URL: https://www.taboola.com/marketing-hub/self-service-vs-managed-accounts-performance-advertising/ Last Modified: 2025-12-21 07:04:06 In today’s rapidly evolving performance marketing landscape, the debate between self-service execution and dedicated managed support has never been more important. We spoke with Nadim Batista-Kuttab of Xevio, one of the industry’s leading native advertising specialists, to discuss how artificial intelligence (AI) platforms are reshaping what’s possible. Nadim’s team manages multiple large enterprise accounts, giving them a unique view on where self-service excels, where there are limitations, and why human expertise remains the ultimate competitive advantage. Today, Nadim shares why enterprise brands often struggle with DIY approaches, how agencies use Realize differently than internal teams, and how dedicated expertise consistently delivers stronger outcomes in terms of scale, speed, and long-term profitability. ## The High-Budget Dilemma: Why Self-Service Fails Enterprise Self-service platforms base their models around simplicity — put in a headline, upload a creative image, set a budget, and go from there. For small advertisers or those testing new channels with limited spend, this can be a more cost-effective entry point, but for enterprise-level customers, the limitations of a purely DIY approach quickly become unavoidable. As Nadim sees it, there’s a disconnect with large advertisers consistently struggling, not because the tools are insufficient, but because the operational complexity of performance advertising makes it impossible to scale without a deep level of expertise. “I think self-service has never worked for big advertisers,” Nadim explains. “Mainly because there are so many different pieces to a typical campaign that without some sort of internal support, you're going to have a really hard time navigating policy issues, creating white lists or blacklists for sites, and understanding the trends of the platform, which are changing very quickly.” This complexity becomes especially challenging for advertisers more familiar with walled garden sites like Meta, Amazon, or Google, where platforms are more automated and far narrower in format constraints. Those environments create a false sense of ease, and advertisers branching out onto the open web often underestimate the depth of manual optimization required for a successful campaign. This assumption can become costly, as contextual inventory, publisher environments, and rapid policy shifts all play into these overarching costs. Even more importantly, the open web rewards experienced users who know how to leverage audience signals, publisher-level insights, and ongoing trend data. Without this understanding, brands risk misallocating budgets, misinterpreting early data, and assuming that short-term performance spikes represent long-term viability. Nadim notes that, for consistent results, having an account manager is the best option. https://youtu.be/SNck87ia2To ## The Strategic Advantage of Team Dedication The biggest differentiator between self-service advertisers and those working with an agency or dedicated in-house team comes down to focus. A strong performance organization doesn’t just know performance advertising, but lives it daily. This singularity in expertise can’t be replicated by brands splitting attention and staff resources across channels. “We have 50 people here that do nothing but Taboola and the other native and performance channels on a daily basis,” Nadim says. “If you assign one or two people from internal teams to compete, it’ll be very difficult.” This isn’t a commentary on the skills of your internal talent, but simply an acknowledgement that the sheer volume of learning required to manage and scale these campaigns is immense. ### Creative Volume Performance advertising relies heavily on creative iteration. A larger team naturally produces more testable assets, such as variations in imagery, hooks, or landing page experiences that expand the options for exploration and accelerate the journey to profitability. With Realize, this scale becomes even more powerful as agencies can run structured creative experiments, feed the algorithm with richer signals, and capitalize on insights from multiple client accounts. A solo marketer, no matter how talented, will always be limited in their volume of experimentation. ### Competitive Intelligence Running dozens of accounts simultaneously generates pattern recognition that no individual brand can replicate. Agencies see emerging trends, publisher shifts, inventory changes, and seasonal behaviors before they ever appear in public documentation. Nadim describes this as a near-instant advantage: Because the team at Xevio is working with the data every day, they can make adjustments in real time. ### Efficiency Every advertiser on the open web eventually pays a “learning tax” through the costs of testing, failing, adapting, and scaling. Agencies, though, have already absorbed that cost across extensive historical datasets. They’ve seen thousands of examples of what works and what doesn’t, they know and understand the pitfalls, and they can interpret early performance data more accurately. All of this translates into less wasted spend. As Nadim puts it, this allows brands to essentially “skip steps” and reach performance goals faster, reducing both total cost of learning and time to profitability. ## The Power of Proximity: Shaping the Platform’s Future Perhaps the most overlooked advantage of a dedicated account partner is their access, not only to data but to the platform itself. Managed service teams and scaled agencies like Xevio operate with a direct line to Taboola’s internal teams, allowing them to resolve issues, request insights, and help shape the platform’s development roadmap. “As a company, we have several direct lines to Taboola itself,” Nadim confirms. “We can suggest products directly to the product teams at Taboola, which essentially helps us shape the direction Taboola is taking toward what we need as one of the bigger spenders on the platform.” ### Policy Resolution Ad rejections and policy flags are inevitable on any performance platform, but on the open web, where publisher diversity is significantly higher, policy issues can decelerate momentum quickly. Agencies can bypass the standard queues and work directly with policy teams to resolve issues within minutes, rather than days. This keeps campaigns more stable, maintains learning continuity within the algorithm, and prevents disruptions to Realize’s optimization process. ### Product Input Because agencies represent some of the platform’s most advanced advertisers, their feedback influences product development. Whether it’s advocating for new targeting capabilities, refining bidding models, or providing suggested enhancements to Realize’s predictive intelligence, agencies play an active role in steering the evolution of advertising products. This is a significant advantage over self-service users working on the platform. ### Early Access Agencies and managed account partners frequently receive early access to beta products, audience models, ad formats, or new inventory sources. These advantages can translate into meaningful campaign gains, particularly during competitive cycles or peak seasonal periods. For enterprise brands, this early access can make the difference between scaling profitability and falling behind competitors who leverage new tools faster. ## AI as an Accelerator: Blending Generative Tools with Human Expertise AI has already transformed nearly every aspect of performance marketing, from creative ideation to landing page production and data analysis. But Nadim is clear that while AI benefits marketers with its ability to improve workflows, it does not replace the strategic, human-driven thinking required to interpret data, understand user psychology, and build comprehensive value streams. “Our recommendation right now, especially on the content side, is to use AI as an accelerator, not as a replacement,” he says. ### Creative Generation Generative image models and automated thumbnails significantly reduce the need for extensive design resources, as with a few prompts, advertisers can generate dozens of variations ready for testing on Realize. This democratization of the creative process opens new possibilities for scaling. However, AI will never replace the strategic thinking behind what should be tested. The best results come from when humans are guiding the design, ensuring that headlines, visuals, and concepts align with the brand goals. ### The Content Fallacy AI can produce content quickly, but what it produces should always be considered the draft, not the finished piece. “Just because an AI tool can write you a landing page in 10 seconds does not mean that landing page will perform remotely as well as a page that has a couple of hours of several people’s time put into it, to really fine-tune those edges and make it perform better,” Nadim says. Human teams are still essential for tweaking nuance, providing emotional resonance and brand cohesion, and understanding conversion psychology. ### The Full Value Stream The performance of advertising campaigns is never determined by a single asset: Success requires synchronicity across the entire chain from click to content, to product and purchase. AI can assist with ideation and drafting, but aligning all components in the advertising puzzle still requires human expertise. As Nadim puts it: “The hard part is really just getting the entire value stream right.” This is precisely where dedicated teams outperform individuals, regardless of the tools available. ## Key Takeaways In complex performance environments like the open web, success is defined by the ability to use tools like Realize with expertise, precision, and understanding of how these tools function to make the best strategic recommendations. While self-service platforms create low-friction pathways for advertisers getting started, they’re not built for enterprise-level budgets or more complicated revenue models. The risk of going it alone increases with spend, as policy errors become costlier, creative limitations become more restrictive, and the inability to test, interpret, and adjust quickly can drain budgets long before campaigns reach their full potential. The human edge remains the ultimate differentiator for enterprise brands. Dedicated account managers, specialized agencies, and deeply experienced advertising teams bring a level of strategic focus that’s difficult for internal teams to match. They offer creative volume, real-time competitive insights, and the ability to navigate performance fluctuations with confidence in little time. This means that the learning curve is typically shortened, protecting budgets from inefficiencies, and always optimizing towards a stronger return on investment (ROI). For advertisers looking to unlock the potential of the open web, attempting to scale with large budgets through self-service alone is no longer the answer. Instead, tools like Realize reward strategic testing and high-quality data inputs thanks to engaged experts who have a deep understanding of the platform. --- ### Unlocking Sustainable Performance ROI by Maximizing Conversion Value URL: https://www.taboola.com/marketing-hub/optimize-roi-with-max-conversion-value/ Last Modified: 2025-12-18 12:29:21 It’s a tale as old as time: Your cost-per-acquisition (CPA) is fantastic, and maybe even falling to a historic low. Conversions are pouring in, filling your pipeline. But, then the inevitable occurs: You look at the monthly revenue reports and notice that your return on investment (ROI) has flatlined. So, what gives? Welcome to the ROI illusion. Many performance teams become so tunnel-visioned on the immediate, easy-to-measure metrics — like lowering CPA or boosting conversion volume — that they completely miss the bigger, more critical picture. You’ve acquired an influx of customers, but they never buy again, return the product, or churn quickly. Your tale includes a problem: You’ve optimized for acquisition cost, not for lifetime economic value. You can change your story’s ending by embracing a value-first approach. This strategic philosophic shift prioritizes the lifetime economic value of an acquired customer over the immediate cost of acquisition. In plain English, an acquired customer who costs you $100 but spends $1,000 over their lifetime is obviously better than one who costs $50 but spends only $60. Making this shift actionable at scale requires modern, intelligent tools that use predictive artificial intelligence (AI) and advanced attribution to identify future value right now. ## Conversion Volume vs. Conversion Value: Making the Critical Switch The difference between a successful, profitable marketing team and one that just spends money? Your answer to one fundamental question: Are you optimizing for volume or for value? ### Volume: The Quantity Trap Focusing on volume means chasing metrics like CPA, click-through rate (CTR), and the sheer number of leads or conversions. It’s a comforting strategy because the numbers rise, and it’s easy to report. It’s also a short-term strategy meant to fill the funnel. But, if your funnel is filled with tire-kickers and one-time discount shoppers, you’re paying for a lot of noise. ### Value: The Quality Imperative Shifting to value requires a deeper commitment (and patience). This strategy’s metrics include customer lifetime value (CLV), average order value (AOV), and the actual post-conversion margin. This approach doesn’t just fill the funnel, it fills it with your best customers — those who make repeat purchases, buy premium items, and become brand advocates. ### The 2025 Mandate Look around: Competition is soaring, and so is the cost of digital media, with ad spend hitting $137 billion this year, a 12% increase from 2024. Relying on mass, low-value acquisition isn’t financially sustainable. The mandate for every growth team in 2026 is simple: Low-value acquisition won’t work, and teams must prioritize customer quality to make ad spend profitable. ## The Non-Linear Customer Journey and Value Attribution The customer journey is no longer a straight line. Your customer might see a top-of-the-funnel vertical video ad on TikTok (the awareness touch), two weeks later click on a Google Search ad, and finally convert after clicking a retargeting banner (the conversion touch). If you’re only giving credit to that very last click, you’re making a mistake: You can’t discount the initial, high-reach video that began this journey and culminated in the retargeting banner’s success. Accurate, cross-channel attribution is non-negotiable. You need a system that can properly assign value across the entire messy, multi-touch customer journey. This approach gives your awareness and high-intent conversion campaigns the credit each deserves, enabling you to sustain the strategy that drives high CLV. ## How to Leverage Predictive Intelligence to Find High-Value Users If you can’t accurately predict which users will be high value before they convert, you’re just guessing. Here are a few tips to master the shift. ### Stop Using Basic Lookalikes Most standard platform lookalike audiences are relics. They’re typically based on recent activity, like “users who purchased in the past 30 days,” but they lack intelligence about future behavior. These audiences are fine for volume but struggle (or fail entirely) to predict long-term value. You’ll end up bidding heavily on a lot of one-time buyers. ### Target Users Based on Predictive Intent The most successful, high-growth marketers use sophisticated artificial intelligence, like Realize’s Matchmaking AI, to analyze deep, multi-year behavioral, transactional, and contextual data to create high-probability segments. This AI can look at thousands of data points and say, “This user, based on their browsing patterns and past purchases of users like them, has an 80% chance of becoming a high-CLV customer.” This analysis shifts the focus from those who recently converted to those likely to be worth more in the future. ### Widen the Net with Confidence Once you have predictive confidence, you can break out of your comfort zone — profitably. This intelligence empowers you to scale campaigns into new, unsaturated channels and demographics that may have seemed too risky or expensive before. You’re no longer bidding for a click or generic lead; you’re bidding for a pre-qualified, high-quality future customer. That’s how you escape the crowded, low-ROI bidding wars and achieve true, profitable scale. ## How to Deliver the Right Value Proposition at Scale Predictive targeting comprises only part of your story. Once you find a high-value user, you must speak to them in a way that resonates with their needs and budget. ### Customize the Hook A user with a high-CLV potential wants something fundamentally different than a user searching for a coupon code. Top-of-the-funnel creative for a high-value customer should address the long-term pain point and the transformative benefit of your product. Bottom-of-the-funnel creative should reinforce its long-term benefit and premium quality, not just a fleeting deal. Use creative messaging that justifies a higher purchase price and a sustained relationship. ### Maximize Asset Utility Creative testing can cause the biggest bottleneck in scaling. You spend weeks perfecting a fantastic vertical video for Instagram, but it takes forever to adapt it for display banners, YouTube, or native ads. You need tools that allow you to maximize asset utility by efficiently repurposing top-performing assets (social videos, carousels, testimonials) into multiple ad formats. From there, it’s easy to test value propositions across different channels quickly. ### High-Visibility Formats High-value messages don’t work unless they’re seen. Use modern, high-visibility formats, like vertical and shoppable video and interactive displays, that capture attention and clearly communicate your value proposition. Put your “value first” message front and center, so it’s highly visible and engaging for your targeted, high-intent audience. ## How to Use AI to Protect Your Value-First Strategy You’ve launched a beautiful, value-optimized campaign. Now, you must protect it from budget erosion. ### Automatic Value Monitoring Forget monitoring just clicks and CPA. Once your campaign launches, your system should prioritize metrics directly tied to future value: - AOV. - Repeat purchase rate signals. - Projected CLV. If you’re spending money, make sure that money flows toward the most valuable outcomes. ### AI-driven Course Correction No human can (or should!) monitor a campaign 24/7 — but an AI assistant can. This AI can continuously track performance; you can program it to immediately flag and potentially pause spending when key value metrics drop, or when a low-value segment starts draining too much of your budget. This guardrail against budget dilution helps protect your ROI’s health. ### Iterate on Insights An AI assistant doesn’t just function as an alarm bell — treat it as a strategic partner that will generate summaries and actionable opportunities. For example, “The segment of users viewing Product X and who read a corresponding case study is yielding 30% higher CLV. Increase your bid on this lookalike by 15%.” Use these concise, data-driven insights to make continuous, profitable decisions that maintain the highest possible average conversion value. ## Key Takeaways It’s time to shift your focus: Stop optimizing only for low CPA and start optimizing for high CLV. Leverage AI to predict future customer value and bid more aggressively on the few high-value prospects, and ensure your ad creative speaks to the value and long-term benefit that high-value customers seek. Implement AI monitoring to immediately flag and stop budget spending on segments that deliver high volume but low value. ## Frequently Asked Questions (FAQs) ### How does Realize's predictive targeting specifically help me bid more aggressively for high-CLV customers? Realize’s predictive targeting, powered by Matchmaking AI, works from over 17 years’ worth of proprietary behavioral data to assign a probabilistic CLV score to prospective users before they click your ad. Instead of bidding on a generic lookalike of all buyers, you’re bidding on a lookalike of the top 5% of lifetime value customers. This confidence allows you to increase your bid for those specific high-probability users; it’s not a generic bid increase but a precise, high-ROI bid increase. ### I have a winning TikTok ad. How quickly can I leverage that successful creative across other high-visibility formats in Realize? With effective asset utility tools, you can repurpose your top-performing creative almost instantaneously. Realize includes tools like the Social Importer, which allows you to quickly adapt, resize, and import winning assets into other formats, to rapidly test the value proposition on new channels without the typical production bottleneck. ### How does Realize's Matchmaking AI help me scale beyond my current audience and accurately forecast high-value users? Matchmaking AI uses your historical high-CLV customer data to identify hidden, common characteristics that simple lookalikes miss. Then it scans the entire available media universe for users exhibiting those same complex patterns. The AI can identify high-value users in new, untapped channels or geographies where you previously have lacked conversion history. Equipped with a high-fidelity forecast of profitable users, you gain the confidence to cast a wider net and scale into new markets. --- ### Creating High-Performing Segments with Realize Audiences URL: https://www.taboola.com/marketing-hub/creating-high-performing-segments-with-realize-audiences/ Last Modified: 2025-12-18 12:18:50 Marketers across various industries face similar challenges in understanding and targeting their audiences. Ad fatigue in a crowded digital market affects every industry, and typical search and social channels aren’t bringing the returns they used to. Performance marketers have to rethink how they’ll find the right users, and where they should allocate budget to adapt to new practices. Platforms like Realize are built for this new approach. If you’re already using Realize, you’ve probably explored its audience segmentation capabilities: These allow marketers to create high-performance segments by utilizing its AI-powered Predictive Audience feature alongside strategic manual targeting. Once you add your company’s first-party conversion data, you can build segments that attract the high-intent users you need to meet your goals faster and with improved cost per acquisition (CPA). ## General Audience Segmentation Tips As with any ad platform, Realize requires some time for the algorithm to train on your particular market and goals to optimize recommendations accordingly. These data-driven strategies can help as you’re getting started, whatever your industry: - Start with a strategic test period to gather crucial data. - Prioritize a wider audience to ensure sufficient scale. - Use robust tracking and artificial intelligence (AI) tools to automate and optimize campaigns. Once you’ve moved through the test period, you can expect improved performance and have a clear justification for scaling. Your confidence in moving forward will be backed by the data you’ve gathered. Again, while these tips are useful regardless of your industry, I’ve also consulted multiple Realize experts to capture some common challenges and guidance in a few specific sectors: financial, real estate, housing for people with disabilities, e-commerce, and education. You’ll hear from these experts in each of those areas on what they’ve seen work best, and get their specific advice to inform your own performance marketing strategies. ## Tips for Financial Sector Audience Segmentation Performance marketers in the financial sector can consider capitalizing on a few current trends. “Anything referencing AI and how it’s changing investments is big right now,” says Jeremy Bade, advertising sales manager, growth, at Taboola. “Saving money is always a big topic, but especially now with the economy, more people are looking for ways to save. Even with offers like car warranties, we’re seeing the saving money angle as a very effective, evergreen angle.” In terms of connecting with your audience, Bade advises that, “You need to give your audience something before asking for information or pushing a sale. I see the most success with headlines and images that seem like a leg up in our economy. For investing, or anything that’s referencing the future, AI, past large financial events, and calling out certain age groups has done well.” As an example, Bade shares that one of his clients is a lender with a HELOC offer, looking to make an impact in the current sea of lenders. “They’ve followed best practices and have continually hit better CPAs than Google or Meta,” he says. “With some small tweaks to their current campaigns, they could hit even better CPAs. They’re using dynamic keyword insertions and calling out homeowners to ensure they have high intent and relevant clicks.” ### Challenges for Marketers Limited budgets and high expectations: Marketers in financial services often use Taboola to reach a target demographic of pre-retirees and retirees with a significant amount of investable assets, but their campaign budgets are often very small. Scaling from narrow targeting: Starting a campaign with a specific, local audience can severely limit the data collected, making it difficult to scale and optimize performance. ### Recommendations Start with a wider geographic net: Instead of a hyper-local campaign, launch at a statewide level. This ensures a broad reach, allowing the Realize algorithm to collect enough data points to identify which sub-segments are performing best. Use creative as the primary filter: Initial ad creative should be designed to attract the specific audience, e.g., by featuring content about retirement or tax planning. This helps pre-qualify the audience before they even click. Leverage dynamic value optimization: Implement server-to-server (S2S) tracking to pass back the monetary value of each conversion. This allows Realize’s "max value" bidding strategy to optimize not just for the number of leads, but for the most valuable ones. ### Expected Outcomes Data-driven scalability: The initial broad campaigns will provide the necessary data to intelligently scale the campaign. Improved lead quality: By optimizing for value, the campaign will attract higher-quality leads, reducing the effort and cost of nurturing them into paying clients. Halo effect: By reaching people on the open web, the campaigns will also positively impact performance on other channels like Google and social media. ## Tips for Real Estate Audience Segmentation “In my experience, having analyzed effective content distribution across native platforms like Taboola, the key strategy is always to prioritize educational content with real-life testimonials and a strong CTA to capture leads early in the funnel,” says Chloe Tai, advertising sales manager at Taboola. ### Challenges for Marketers Programmatic burnout: The client's customers have had negative experiences with programmatic advertising in the past and are highly skeptical of the channel. Long conversion cycles: The path from lead to sale in real estate can take months or years, making it difficult to demonstrate performance and justify ad spend in the short term. Prove value on a tight budget: The client needs to see results from a small initial test budget. ### Recommendations Start broad with whitelisting: The campaign should begin with a broad reach but be immediately whitelisted to ensure ads only appear on high-quality, trusted news and finance sites. This counters the client's previous negative experiences. Use the Audience Exploration tool: After the initial learning phase, use Realize’s Audience Exploration tool to identify the top-performing audiences that the client may not have considered. Track the full funnel: Implement tracking for every conversion point, from the initial lead form fill to a phone call or appointment booking. This allows you to prove the campaign's value even if a final sale is months away. Below, you’ll find Tai’s explanation of how you should think about the funnel, as well as other invaluable tips from her on getting the most out of your campaign: #### Top-of-Funnel: Awareness/Education Content: Evergreen, educational, and intriguing articles. Examples: - "5 Hot Neighborhoods in Poised for 20% Growth in 2026" - "Why Your Property Tax Bill Is Changing (And What to Do About It)" - "The Ultimate Checklist for First-Time Homebuyers in " Goal: High click-through rate (CTR) and an engaged reader (high time-on-site). #### Middle-of-Funnel: Consideration/Lead Capture Content: Specific, high-value lead magnets that require a form-fill (gated content). Examples: - "Download the Full 2024 Local Market Investment Report" - "Access Exclusive Floor Plans for The Development." Goal: Optimize for cost per lead (CPL). #### Bottom-of-Funnel: Conversion/Action Content: Headlines and images should look more like news articles than banner ads. Use numbered lists, question headlines, or shocking statistics. Examples: - Bad Headline: "Buy a Condo at The Riverwalk Now and Save!" - Good Headline: "The Unexpected Amenity Driving Up Condo Prices" Goal: Conversion. #### Getting the Most From Your Campaign Use people in images: Images featuring people, especially looking directly at the camera, tend to perform exceptionally well on native platforms. Use authentic, high-quality images that don't look like stock photography. Motion ads: Utilize Taboola's Motion Ads (short, looping videos/GIFs). Real estate is highly visual, and subtle movement draws attention in a way a static image can't, often leading to a better CTR and lower CPL/CPA. Sequential retargeting: Set up a chain of ads with the following aim: - User clicks on your educational article (ToFu). - User is then retargeted with an ad to download a specific "Buyer's Guide" (MoFu). - User who downloads the guide is then retargeted with ads showing actual listings or an offer for a "One-on-One Consultation" (BoFu). Geo-targeting and zip code specificity: Real estate is local. You must target by highly specific demographics and geographic areas (zip code, radius around a property) to ensure you’re only paying for clicks from relevant markets. Lookalike audiences: Upload your CRM list of past buyers and current qualified leads to Taboola. The platform's AI can then create lookalike audiences based on that data to find new high-quality prospects. Rigorous A/B testing: Your success hinges on testing, so try the following: - Test multiple combinations: Test a minimum of 2-3 images with 5-10 headlines for every content piece. Taboola's algorithms will automatically optimize delivery toward the winning combinations. - Test landing pages: Don't send all clicks to your home page — create dedicated, conversion-focused landing pages that mirror the headline/ad copy the user clicked on. ### Expected Outcomes Achieve desired cost per lead: The campaign will quickly achieve a healthy cost per lead (CPL) in the client's desired range. Build trust with data: By providing clear, data-backed reports on the quality of the leads and their progression through the funnel, you can rebuild the client's trust in programmatic advertising. Justify budget increases: The successful test will serve as a proof-of-concept, encouraging the client to increase their budget and shift ad spend from other channels. ## Tips for Housing for Disabled People Audience Segmentation ### Challenges for Marketers Legally restricted targeting: Direct targeting of users with disabilities is not possible due to legal and privacy concerns. This makes it difficult to reach the core audience. Audience skepticism: It isn’t always clear whether the target audience spends time on premium news sites, which are the primary publishers on the Realize platform. Small initial budget: A small monthly budget makes it difficult to justify a larger, managed service agreement. ### Recommendations Focus on contextual and topic targeting: Instead of targeting the user directly, target the content they are consuming. For example, run ads on articles related to disability support, community resources, or accessible housing. Use the first two weeks as a learning phase: Allocate the initial budget to a broad "prospecting" campaign. The goal is to collect data and insights on which content categories and publishers deliver the best results, rather than to achieve immediate conversions. Implement the pixel for retargeting: Ensure the Taboola pixel is implemented immediately. The data from the prospecting phase will build a retargeting audience of people who have already shown interest in the ads. ### Expected Outcomes Data-backed confidence: The initial prospecting phase will provide clear data that proves the target audience is active and engaged on the Realize network, which has alleviated the client's skepticism. Effective lead generation: After the learning phase, the campaign can be optimized for leads, leading to a stable and lower CPL. The retargeting campaign will then serve as a high-performing channel for driving conversions. Path to growth: By demonstrating success on a small scale, the client will be more likely to increase their budget and move to a managed service, which offers access to exclusive, premium inventory. ## Tips for E-commerce Segmentation Success Performance marketers for e-commerce brands can use Realize in many ways, from reporting all the way to testing creatives, says Taboola advertising manager Julius Dunkin. “The secret is to educate and inform on your product, then lead prospects to the purchase,” Dunkin says. “Any additional actions, like driving them to the company home page or blog, lowers the potential for conversion.” So, make sure you have concise funnels and landing pages, he advises, adding that, “E-commerce brands that are seeing success with Realize include Temu, Wolf and Shepherd, and Avocado Mattress.” ### Challenges for Marketers Profitability on first purchase: Your main goal may be to make a profit on the first purchase, a difficult task in the highly competitive e-commerce space. Aversion to initial losses: Some marketers may be extremely risk-averse and will pause a campaign immediately if it's not profitable on day one. This directly conflicts with the necessary "learning phase" of most ad platforms. ### Recommendations "Make sure you have eye-catching creatives to drive engagement", suggests Dunkin. Great calls to action (CTAs) are essential, too. “Customers actually want you to tell them what to do. Make sure your call to action does just that,” he says. Set a clear test budget: Rather than a fixed monthly spend, establish a sufficient daily budget for a minimum two-week test. This allows the algorithm to learn without prematurely stopping the campaign. Leverage AI-powered bidding strategies: Utilize Max Conversions to drive as many conversions as possible initially, and then Max Value to optimize for the highest-value purchases. Pass back dynamic return on ad spend (ROAS): Implement an S2S integration to pass back the exact revenue from each conversion. This provides the algorithm with the data it needs to identify and target users who make larger purchases, leading to a better overall ROAS. ### Expected Outcomes Profitable scaling: The algorithm will use the dynamic ROAS data to automatically identify profitable segments, allowing the client to confidently scale their ad spend. Improved conversion rates: By using engaging creatives and optimized landing pages, the campaign will achieve a higher conversion rate, directly impacting profitability. Long-term partnership: By demonstrating that the platform can deliver on this strict profitability metric, it’s easier to build a foundation for a long-term, high-spending partnership. ## Tips for Education Marketing Success Performance marketers in the education sector can use Realize to attract new students and drive awareness of their college or university, says Dunkin. “There has been talk of drops in enrollment across the board, but a subset of colleges like SNHU and Strayer are thriving,” he adds. “That’s due to their use of platforms like Realize partnered successfully with other marketing channels.” ### Challenges for Marketers Low budgets, high goals: Budgets for educational campaigns are often unrealistically low, while they are expected to produce a large volume of conversions. Lack of marketing sophistication: In the education sector, clients may not understand marketing concepts like CPA, making it difficult to have a productive conversation about a campaign's return on investment (ROI). ### Recommendations Start with a managed test: Instead of an unmonitored self-service campaign, try a managed service test with a sufficient daily budget. This provides the necessary support and data for a successful launch. Focus on max conversions: Use a "Max Conversions" bidding strategy to prioritize lead volume over efficiency. This will generate a high number of leads, providing tangible results on a small budget. Utilize layered and predictive targeting: Use a combination of layered targeting (geography, interests, and contextual) to create a highly relevant audience. Try the Predictive Audiences tool, which uses AI to find users who behave similarly to your existing leads, helping the campaign scale beyond initial targeting parameters. ### Expected Outcomes Tangible results: The campaign will deliver a high number of leads, providing the client with immediate, quantifiable results. Educational foundation: The data from the successful test will serve as the basis for a more direct, educational conversation with the client about what it truly takes to achieve their long-term enrollment goals. Justified investment: The success of the initial campaign will build a strong case for future, larger investments in the channel. ## Key Takeaways Performance marketers face plenty of challenges, but there are plenty of opportunities, too. AI-powered platforms like Realize offer audience segmentation capabilities that, done well, can help target the right high-intent users to build new audiences and drive conversions within budget constraints. ## Frequently Asked Questions (FAQs) ### How do I use segmentation to create personalized customer experiences? Audience segmentation offers a lot of possibilities for creating personalized customer experiences. Start by gathering all available customer data, then divide customers up by shared characteristics such as demographics or behavior, depending on your marketing goals. Then, create personalized offers, messages, landing pages, and more for each segment, taking the channel and timing into consideration as well. The results can inform your personalization strategy going forward. Realize can help you create high-powered segments using the AI-powered Predictive Audience feature with strategic manual targeting, streamlining the manual work involved. ### What role do AI and machine learning (ML) play in audience segmentation? AI and ML bring powerful capabilities to marketers working on audience segmentation. With its ability to analyze massive amounts of data, AI-powered technology, like Realize’s Predictive Audiences feature, can draw upon much more information to create more precise and predictive customer segments than typical demographic divisions. Marketers can save a lot of time with these features, which help hyper-personalize marketing campaigns by identifying patterns, predicting future behavior, and refining audience segments in real time, using the latest results. Realize uses performance AI to match content with relevant users, aligning ads with audiences by analyzing content and metadata across publisher sites. ### What data do I need to collect for effective segmentation? Consider any data that’s particularly relevant to your industry, and what you already know about your target audience. The data needed for effective segmentation will vary by campaign and industry, but typically, the basics include demographics, geographic location, behavior (such as purchase history or web activity), and any lifestyle or values data from surveys. Realize offers Search Keyword Targeting so you can create audience segments based on user keyword or phrase searches, along with custom mail domain segmenting. Realize provides more than 500 dedicated first-party segments across more than 200 categories, as well as third-party data, with audience segments for 20+ providers available. You can also use interest signals across selected segments to make your ads more relevant. --- ### Q5 Top Advertising Creative Trends and Messaging Themes URL: https://www.taboola.com/marketing-hub/q5-creative-ad-trends/ Last Modified: 2026-03-02 08:49:50 Between December 26 and mid-January, the digital ad landscape makes a dramatic shift. Shoppers move from frantic gift-buying to slower, more introspective browsing, ad costs on the open web drop just as user intent spikes, and across nearly every vertical, people start envisioning a better version of themselves. This period, now known as Q5, isn’t just a bonus month tacked onto Q4: It’s a psychological reset moment for consumers, offering brands the opportunity to reach audiences who suddenly have mental space to think about what they want next. Below, I’ll break down this year’s biggest Q5 creative trends and messaging themes across three high-impact verticals — health and fitness, shopping/style, and personal finance — along with actionable takeaways for advertisers looking to capitalize on this short but high-impact time of year. ## 1. Healthy Living, Fitness and Exercising More than any other category, health, wellness, and fitness dominate Q5. Resolution energy always peaks this time of year, but the creative that performs best is not aspirational perfection: What works well is authenticity, lower barriers to entry, and small wins. Here are some creative trends and messaging themes to help you connect with consumers during this powerful reset window. ### Authenticity > Perfection Consumers aren’t looking for fitness models or staged routines right now: They want someone whose appearance echoes how they feel, i.e., tired and ready to start fresh. Creative that feels like user-generated content builds trust, so think shaky phone video, unfiltered progress clips, and “Expectation vs. Reality” posts. "When it comes to weight loss and wellness imagery, you have to be 'squeaky clean' with premium publishers like Apple News, but that doesn't mean boring. I’ve seen great success with natural imagery—like someone simply holding a GLP-1 pen—rather than overly staged before-and-afters, which often face higher rejection rates in premium environments." — Jeremy Bade, Advertising Sales Manager Strong creative angles include: - The messy beginning (“Day 1…kind of.”). - Micro-transformations (“Woke up with more energy than yesterday.”). - Honest attempts (“Tried a new class. Didn’t die.”). This format translates well into short-format video, making it ideal for Instagram Reels and TikTok. ### Make the First Step Feel Easy In Q5, audiences gravitate toward products or routines that feel accessible, not overwhelming. Instead of highlighting features, top-performing creatives emphasize simplicity. Here are some examples of friction-reducing messaging: - “Start with ten minutes.” - “A routine you can actually stick to.” - “No experience required.” This theme pairs well with formats that support lightweight explanations. This might be a short product walkthrough, a carousel of simple steps, or a vertical video that reassures rather than pressures. "To reduce friction during the New Year reset, we recommend starting with broad targeting on mobile and desktop to keep CPCs down while the algorithm learns. For health brands, targeting based on 'intent'—like the content users are already engaging with in their email inboxes—is a powerful way to find people ready to take that first easy step." — Kelly Basinger, Team Lead, Advertising Sales ### Interactive and Immersive Touchpoints People love to test-drive their future selves. Use engagement-driven content like quizzes, filters, quick polls, and mini-challenges to equip users to mentally “try on” a new routine before committing. "Interactive funnels are huge for wellness. We’ve found that moving 'personal identifiers' like name or date-of-birth just a few slides deeper into a quiz or assessment can significantly skyrocket conversion rates. It’s about building trust through the interaction first." — Jeremy Bade, Advertising Sales Manager Your marketing language can include terms like: - “Find your perfect workout match.” - “Discover your wellness baseline.” - “Join the seven-day mini reset.” This builds stickiness and creates emotional buy-in, especially when paired with gamification such as badges, streaks, or leaderboard milestones. ### Show the Transformation, Not the Product Rather than highlighting ingredients, equipment, or specs, winning Q5 creative frames the offering as a tool for transformation. Outcomes, along with the feelings tied to those outcomes, do the heavy lifting. "In the native environment, users have just finished an article and are looking for what's next. If your creative leads with a transformation story—like 'three common foods to avoid for better brain health'—it feels like a seamless discovery rather than an interruption. This 'innocent' approach to high-intent discovery is our bread and butter for eCom and wellness." — Malik Elijah Ward, Sr. Advertising Sales Manager High-performing transformation-focused messages include: - “Energy that actually lasts through your busiest days.” - “Fall asleep faster. Wake up clearer.” - “Build confidence, one small win at a time.” This approach performs especially well in open-web display, carousel, and native-plus formats, where emotion-forward headlines and imagery drive higher engagement and scroll-stopping power. ### Community as a Selling Point The lone-wolf approach to fitness is fading: Today’s consumer wants a shared workout experience, along with accountability. Show audiences not only the product, but also the support system that comes with it. Effective community-driven messages include: - “Join thousands of people starting over — just like you.” - “Real coaches. Real feedback. Real progress.” - “Train together. Stay consistent together.” In Q5, community is more than a bonus feature, it’s part of the core value proposition that pushes users from intention to action. ### The Rise of Preventive Care and Longevity More consumers now see wellness as long-term self-preservation rather than a short-term aesthetic change. Q5 creative that taps into proactive health outperforms bare-minimum “weight loss” messaging. "We are seeing a shift where wellness is viewed as a long-term investment. To scale these campaigns, we utilize 'Predictive Audiences.' Once a campaign hits a threshold of conversions, our algorithm identifies users with similar long-term behaviors and targets them before they even search for a solution, often improving CPAs by 25-30%." — Kelly Basinger, Team Lead, Advertising Sales Some effective preventive care and longevity messaging includes: - “Future-proof your health one habit at a time.” - “Invest in your next 40 years.” - Build stronger bones for the years ahead.” This type of messaging pairs beautifully with formats that support storytelling. Consider hero imagery, vertical video, and bold, future-focused headlines that help users imagine the healthier self they’re working toward.2. Shopping, Style, and Fashion Retail behaves differently in Q5. Once the last ornaments are packed away, consumers enter a reflective, self-reinvention mindset. They’re eager to refresh wardrobes, redeem gift cards, and shape the version of themselves they want to be in the new year. Below are some current creative trends impacting this category. ### “New Year, New Aesthetic” Video Hauls Users love to watch others edit, clean, and refresh their wardrobes. Q5 haul videos perform especially well because they match the audience’s internal monologue, which asks, “What is my aesthetic going into 2026?” "The mindset of a user in Q5 is uniquely primed for 'the next.' They’ve finished their holiday shopping for others and are now consuming content to define their own new-year identity. Video hauls succeed here because they aren't just ads; they are 'discovery' experiences that provide the visual cues users are already hunting for as they mentally build their 2026 aesthetic." — Abby Burdick, Advertising Sales Manager, Large Enterprise High-performing creative angles include: - “Designing my entire vibe for the new year.” - “From cluttered closet to clean aesthetic in one reset.” - “What I’m keeping vs. what I’m finally letting go of.” As you’re posting these video hauls, keep your tone relatable and real, not overly polished or influencer-perfect. ### Style Resolutions With the reset of a new year, shoppers reassess their wardrobes and shift toward pieces that support the life they’re building. This is where “style resolutions” take hold, turning purchases into intentional identity upgrades. "To hit aggressive ROAS goals for style-conscious brands, we utilize 'Predictive Audiences.' Once we capture enough signal from early Q5 shoppers, we stop looking just for demographics and start targeting users based on their specific behaviors across the open web. This often leads to a significant lift in high-value conversions during the January refresh window." — Abby Burdick, Advertising Sales Manager, Large Enterprise Messaging that resonates: - “Define your 2026 look.” - “Dress for who you’re becoming.” - “Refresh your wardrobe, refresh your mindset.” This narrative works across nearly every open-web creative format, from vertical video to static display to carousel “before and after” outfits. ### Gift Card Redemption Campaigns Q5 often sees consumers holding extra money in the form of gift cards. In fact, one study found that six out of 10 people received gift cards for Christmas in 2024. Nearly half of those consumers redeemed all their gift cards by Valentine’s Day. That gives brands the opportunity to: - Retarget users whose behavior signals gift card ownership. - Run dynamic creative showing remaining balance or suggested pairings. - Use timely headlines like “Your card = Your new look.” It’s important to act quickly, since the gift card spending window closes fast. These consumers are high-intent purchasers who can often be prompted with a reminder that they have extra funds to spend. "For retail partners, the 'cookie pool' of recent Q5 site visitors is where we see the most explosive ROAS—sometimes exceeding 20X on retargeting. By starting with a broad prospecting approach early in the quarter to identify 'refresh' intent, we can then use our DCO (Dynamic Creative Optimization) tools to hit those same users with specific product reminders, effectively capturing those gift card funds while they are top-of-mind." — Abby Burdick, Advertising Sales Manager, Large Enterprise ### The “Treat Yourself” Moment Post-holiday spending is about self-care, not gifting. Once the pressure of finding the perfect gift for everyone else fades, consumers shift toward treating themselves. Lean into that emotional exhale with language that encourages them to click that “buy” button. "We’ve seen that 'Self-Care' and 'Treat Yourself' messaging performs exceptionally well in unique placements like the top of the inbox in Yahoo or MSN Mail. Users are in a 'to-do' mindset when checking mail, and a well-timed, intentional offer for something they actually want—not a gift for someone else—is highly effective." — Kelly Basinger, Team Lead, Advertising Sales High-performing angles include: - “You’ve taken care of everyone else. Now it’s your turn.” - “Start the year with something that feels like you.” - “Have extra holiday cash? Treat yourself.” Q5 is the ideal time to guide shoppers toward purchases that feel restorative, intentional, and easy to justify. ### Reframing Clearance as a Fresh Start Consumers aren’t alone in seeking a restart at the end of the year. Q5 is traditionally one of the biggest clearance periods of the year, as retailers aim to rid themselves of extra inventory. But, clearance messaging doesn’t have to read like a bargain bin: When framed as a reset, it becomes aspirational instead of transactional. Some examples of strong clearance messaging: - “Build your new capsule wardrobe from our year-end edit.” - “Start the year with a closet that feels intentional.” - “Refresh your daily habits with our final picks of the season.” Instead of green tags and percentages, pair this messaging with lifestyle imagery to elevate your creative. "Don't just run transactional clearance ads. Use DCO (Dynamic Creative Optimization) to show suggested pairings or capsule wardrobe edits based on what the user has previously viewed. This reframes 'clearance' as 'intentional selection,' which drives much higher engagement from retail audiences looking for a fresh start." — Abby Burdick, Advertising Sales Manager, Large Enterprise ### Returns and Exchanges as an Upsell Opportunity Retailers expect 15.8% of annual sales to be returned in 2025, and a large share of those returns happen in the final week of the year. If you haven’t already optimized your return and exchange experience, Q5 is the perfect time to reduce friction and encourage upsells. This turns a loss into a conversion opportunity. Here are a few messaging examples to help with upsells: - “Didn’t like it? Let’s find something you’ll love.” - “Not the perfect match? Your better option is waiting.” - “Trade it in for something you’ll reach for every day.” Best of all, turning a return into the right fit can boost customer satisfaction, which may turn a gift recipient into a lifetime customer. "We can take the high-intent keywords users are already searching for—like specific products or return policies—and use those to find them in a less competitive environment. By showing them a 'better match' or a complimentary, display item on the sites they visit after searching, we aren't just bidding against everyone else on a search page; we’re using that intent to lead them toward an upsell when they’re most open to an alternative." — Kelly Basinger, Team Lead, Advertising Sales ## 3. Personal Finance (Fintech, Banking, and Budgeting Apps) Q5 is to finance what November is for retail. It’s the moment when intent, emotion, and urgency come together. Nearly two thirds of Americans are planning a finance-based resolution in 2026, and 71% say they have a plan to reach that goal. In the weeks following Christmas, consumers are actively seeking a simple but clear approach to reaching their financial goals, combined with the opportunity to start the year with a clean slate. That gives fintech brands a prime opportunity to position themselves as the go-to partner for a fresh financial start. Below are the creative trends and messaging themes that resonate most with Q5 audiences in the personal finance space. ### Year-in-Review Storytelling Consumers respond strongly to creative that reflects where they are financially and where they could be if they reach their goals. “Your year at a glance” storytelling, whether hypothetical or personalized, makes New Year's resolutions feel concrete. Strong examples of year-in-review messaging include: - “Here’s what you saved this year.” - “Imagine what you could build by summer.” - “You’re closer than you think to a down payment.” For best results, pair this messaging with visuals that have clean, minimal design. "To maximize impact, don't just show general savings stats. We use 'Predictive Audiences' to identify users based on their specific financial behaviors across the open web. Once we've gathered enough conversion signals, the algorithm finds users with similar intent—like those looking to build a down payment—improving CPAs by 25-30% compared to broad targeting." — Kelly Basinger, Team Lead, Advertising Sales ### Zero-Stress Saving and Budgeting In Q5, shoppers are looking for relief after a month of heavy spending. Consumers don’t want restrictions, though: Instead, they gravitate toward tools that support them in their financial goals. The most effective creative removes pressure and replaces it with ease, showing that better money habits can start with small, automated steps rather than major lifestyle overhauls. Some effective messaging for this pain point includes: - “Set it up once. Watch it grow.” - “Budgeting without spreadsheets.” - “Automation that works while you live your life.” This approach pairs naturally with solutions offering automated rules, flexible goal-setting, or quick, intuitive onboarding. ### Problem/Solution Short Videos Problem/solution videos can be especially effective in Q5, particularly for small, direct-to-consumer businesses with audiences looking to learn about a tool before trying out a new solution. With consumers already in a resolution-driven mindset, this format cuts through uncertainty by showing why they’re struggling and how they can fix it. "Vertical video is the ideal format for this vertical because it mirrors how users consume 'quick-fix' content. We’ve found that leading with a relatable pain point—like 'Ever wonder where your holiday budget went?'—immediately reduces the friction often associated with banking apps and increases CTR on our mobile-first placements." — Malik Elijah Ward, Sr. Advertising Sales Manager The most effective problem/solution videos follow a three-step format: - Call out a reliable pain point: Use messaging like, “Ever wonder where your money went?” This immediately signals empathy and relevance, reducing the friction that often comes with financial topics. - Present a quick, achievable fix: Highlight a single feature or action step that solves the identified problem. No deep dive required. - Show the immediate reward: Demonstrate how the tool delivers clarity, control, or quick savings with messaging such as, “Track every dollar in seconds.” This helps viewers visualize how the solution could impact their day-to-day lives. Vertical video is ideal here because it mirrors the native environments where consumers already consume quick, problem-solving content. ### Positioning the Tool as an Accountability Partner In Q5, motivation is high, but so is the risk of going off track. People looking for financial change want more than dashboards — they want built-in support that helps them stick to their goals. Tools that act like an accountability partner resonate far more than those that simply display numbers. Some effective angles to consider include: - 90-day action plans. - Debt repayment journeys. - Nest egg kickstarters. - Weekly progress check-ins. Additionally, ensure any headlines emphasize achievable goals to make progress feel realistic and within reach. ### Authority and Education Build Trust Finance is a sensitive category, and Q5 audiences are especially cautious as they reevaluate their money habits for the year ahead. They’re looking for guidance that feels credible, calm, and grounded, rather than sales-driven. Educational, authority-driven creative performs well here because it reduces friction and positions the brand as a trustworthy partner for those working toward a financial reset. "Trust is the currency of the fintech space. We recommend using 'Advertorial-style' content rather than just a landing page. If you can provide an expert-backed article on why a user should choose your tool before they hit the signup page, you’re not just buying a click; you’re building the authority needed to convert high-value leads." — Kelly Basinger, Team Lead, Advertising Sales Strong authority-building angles include: - Clear, minimalist layouts that make complex topics feel simple. - Expert insights that validate a user’s financial decisions. - Straightforward charts that show a user’s progress at a glance. When consumers feel informed rather than overwhelmed, they’re far more likely to explore deeper features and take meaningful action. ### Tie Savings to a Big Life Goal Generalizations rarely motivate. This is especially true in Q5, when people are actively imagining the version of themselves they want to become in the year ahead. This is the moment to anchor financial tools to specific, meaningful milestones that feel both exciting and actionable. "In the personal finance and home services space, the offers that perform best are the ones that feel like 'slam dunks' because they are services people naturally use and need in their everyday lives. By narrowing the focus to these evergreen, real-world needs, the tool stops being a generic service and becomes a practical solution for the larger goals they're already trying to fund." — Malik Elijah Ward, Sr. Advertising Sales Manager Some strong examples of this type of messaging are: - “Save for your dream trip.” - “Begin your homeownership journey today.” - “Start your summer vacation fund now.” This approach increases emotional resonance and perceived value, turning everyday saving into a path toward something tangible and inspiring. ## Key Takeaway: Q5 Is the Reset Moment Brands Can’t Afford to Miss While Q4 gets plenty of hype, Q5 is where savvy advertisers quietly gain ground. Consumer behavior is uniquely introspective this time of year as audiences actively seek products that help them build better daily routines. The most effective Q5 creative acknowledges this shift and meets users where they are — hopeful, reflective, and open to change. ## Frequently Asked Questions (FAQs) ### How can advertisers effectively use the drop in Q5 CPMs on the open web to acquire new high-value customers, instead of just retargeting? Q5 brings lower cost per milles (CPMs), allowing brands to expand beyond warm audiences and reach new shoppers who are exploring post-holiday options. This is the ideal moment to broaden targeting, test new creative variations, and run prospecting campaigns framed around long-term goals rather than holiday urgency. With more inventory and less competition, brands can cost-effectively introduce themselves to audiences who are finally ready to engage. Realize is built for precisely the moments in the funnel when lower CPMs and higher intent intersect. Its predictive audience models use more than 17 years of behavioral data to identify users showing emerging intent. By combining Realize’s matchmaking artificial intelligence (AI) with high-visibility creative formats, advertisers can efficiently scale prospecting to reach new, high-value customers. As Q5 unfolds, Realize automatically reallocates spend toward the segments that show the strongest potential, allowing brands to capitalize on lower costs while building richer, more future-ready audiences. ### Beyond standard native ads, what creative formats on the open web are best for capturing the "New Year, New Me" mindset? Vertical video, carousels, and interactive formats like quizzes perform exceptionally well because they let users test-drive a new habit. Display units that highlight emotional outcomes like better sleep and clearer budgets also help motivate users during Q5. Realize supports a wide range of conversion-focused creative formats beyond traditional native, making it ideal for Q5’s self-improvement mindset. Advertisers can deploy short-form vertical video, carousels, rich display units, and other immersive placements that help users visualize new habits or goals. Realize’s Social Importer tool also lets brands instantly repurpose top-performing social content for open-web placements, accelerating creative testing during a period when freshness and agility matter most. Together, these formats let marketers deliver emotionally resonant, high-impact creative to audiences ready for change. ### How does advertising on the open web in Q5 allow brands to strategically test and prepare for sustained Q1 momentum? Q5 is the proving ground for messaging, creative, and audience segmentation. Brands that test early enter Q1 with validated concepts, refined calls to action, and an understanding of what resonates emotionally with their audiences. This gives them a substantial performance advantage once competition increases again in late January and February. Realize turns Q5 into a strategic testing ground, allowing advertisers to refine creative, messaging, and audience signals before Q1’s competition spikes. Its AI-driven optimization and automated bidding strategies reveal which themes, formats, and user segments respond best, while the landing page builder supports rapid iteration. By the time CPMs rise again in late January, Realize has already trained on weeks of high-quality engagement data, giving brands a validated roadmap and the ability to scale confidently into Q1. --- ### Top Products for Q5 Advertising: What’s Connecting With Consumers URL: https://www.taboola.com/marketing-hub/q5-advertising-top-products/ Last Modified: 2025-12-17 06:33:17 You may be thinking that the end of the year means time to slow down and regroup for the coming 12 months. But, ignoring Q5 as a product-based business means you’re missing out on a crucial shopping window over the holiday season. Q5 refers to the post-holiday advertising window that runs from December 26 through mid-January, a period where consumer behavior fundamentally shifts. While traditional holiday marketing typically focuses on gifting, with urgency and external motivation, Q5 is driven by self-focused purchases, replenishment, deals, and a “new year, new you” mentality. For performance marketers, Q5 represents a golden opportunity and a second peak, one less defined by urgency and more by decisive, self-motivated buying behavior. ## 6 Top Performing Products for Q5 Marketing Q5 is a distinct period in retail, with 87% of Q5 shopping ending in a purchase, making it one of the most high-intent periods of the year. It overlaps several moments — the post-Christmas sales, New Year’s Eve celebrations, and the early January reset period where consumers are reassessing routines, budgets, wardrobes, and their homes. This is far from a cool-down period: Research suggests that advertisers who remain active in late December and January see 70% more visits per advertiser and 25% higher conversion rates compared to non-Q5 periods. With all that in mind, let’s take a look at which business types can best take advantage of Q5. ### Health, Fitness, and Wellness Products Health and wellness products consistently dominate Q5 because they directly align with New Year’s Resolution behavior. Consumers are actively looking for products to help them start the new year well and better maintain a healthier routine over the coming months. At-home fitness equipment, supplements, wellness devices, and entry-level fitness programs all perform particularly well because they lower the friction between intent and action. A YouGov survey for 2025 stated that over half of all polled intended to improve their physical health this year. Products that typically convert well during Q5 include compact workout gear such as adjustable dumbbells, resistance bands, yoga mats, and beginner kettlebell sets for those looking to work out at home. Smart scales, fitness trackers, and wearable health monitors are all options that allow users to measure progress, particularly if starting a new health routine. Supplements tied to daily routines like protein powders, hydration mixes, magnesium, and collagen are all strong performers when framed as tools for consistency, rather than quick fixes. The strongest Q5 creative for health, fitness, and wellness focuses both on immediacy and simplicity — starting small, tracking progress, and building long-term habits without feeling overwhelmed. ### Home Goods, Organization, and Home Upgrades Home-related products perform well in Q5 because consumers shift from holiday hosting to reclaiming and freshening up their space. The post-holiday period often brings clutter, returns, and a desire for resetting the home, creating a natural demand for organization tools, storage solutions, and functional home upgrades. Products that resonate in Q5 in this category include closet and pantry organizers, storage bins, shelving systems, label makers, small furniture pieces, and kitchen upgrades such as food storage systems or countertop appliances. With ongoing consumer boycotts of large brands such as Starbucks, appliances like at-home espresso machines are seeing increasing appeal for consumers looking to effect change with their wallets. Decorative homeware tied to seasonal hosting such as tableware, serving dishes, or glassware may continue to perform well as consumers look to purchase in the sales and prepare for hosting smaller gatherings throughout the year. MNTN’s Resolution Season research indicates that home brands were among those seeing improved cost per acquisition (CPA) during this period, suggesting stronger efficiency rather than simply higher volume. Post-holiday returns often trigger replacement purchases, and NRF projects $849.9 billion in total returns for 2025, with much of this re-spent in January. For advertisers in this space, Q5’s messaging should focus on reclaiming time, simplifying routines, and making everyday life easier. ### Self-Improvement, Learning, and Productivity Tools Beyond physical health, Q5 is where consumers invest in personal growth across finances, productivity, learning, and mental well-being. This category includes planners, journaling systems, digital productivity tools, online courses, language learning platforms, and financial planning software. These products work well in Q5 as they symbolize a fresh start. A planner or budgeting tool is not just a purchase, but a commitment to making changes in the coming year. Research shows that over half of U.S. adults plan to make New Year’s resolutions, and many of those center around finances, hobbies, and personal development. Products that reduce the complexity of these life changes perform well, such as guided planners, starter financial tools, short-form courses, and subscriptions that offer immediate structure. Entry-level offers and bundled kits tend to outperform premium commitments because they align with a cautiously optimistic consumer, rather than over-ambition. ### Beauty and Personal Care Products As with fitness and wellness content, beauty products occupy a unique position in Q5 advertising, bridging the gap between rest and celebration. In late December, products tied to New Year’s Eve party preparations perform well, but a shift toward more routine-based self-care occurs in January. Products that work well at this time of the year include moisturisers (particularly in the Northern Hemisphere where winter is in full swing), masks, serums, grooming essentials, and entry-level skincare sets. This is also the time where bundling is more common, giving consumers an opportunity to try products they may not normally purchase. Anything positioned as “winter recovery” or resetting tends to perform best. Beauty products also work well because they offer instant gratification without long-term commitments — an appealing trait in a period of seasonal transition. Industry commentary consistently groups beauty with wellness during Q5 as part of resolution season. Messaging in Q5 around beauty and self-care should emphasize comfort, recovery, and maintenance in the winter months, rather than transformation. While this may work well in other industries like productivity, beauty should align with a more relaxed post-holiday mindset. ### Apparel, Accessories, and Comfort-Driven Wardrobe Refreshes While apparel is not traditionally associated with resolutions, it performs strongly in Q5 due to gift card redemption, seasonal necessity, and the desire for comfort after the holidays. Nearly 60% of shoppers plan to purchase gift cards for loved ones during the holiday season, so Q5 is the natural point where those gift-receivers look for new items with their gift cards. In many cases, shoppers will spend beyond the amount gifted to them, increasing revenue for retail businesses. Consumers often use January to replace or upgrade core wardrobe pieces like winter coats or workwear, rather than purchasing spontaneous extras. Products that resonate in Q5 include hoodies and sweatshirts, loungewear, relaxed workwear, layering basics, winter accessories, and versatile pieces that can transition from holiday to everyday wear. These purchases are often driven by practicality and value rather than trend-chasing, increased by gift card sales in Q4. 66% of consumers report shopping in post-holiday sales, with apparel among the most common category for purchases. Creative for Q5 advertising in this category emphasizes comfort and versatility, rather than urgency. ### Subscription Products and Starter-Level Offers Q5 is an ideal entry point for subscription and repeat-purchase brands, since consumers are motivated to build habits, but are hesitant to overcommit themselves. Starter kits, trial bundles, and limited-time introductory offers perform well because they lower perceived risk while aligning with resolution behavior. This applies across categories such as meal kits, wellness boxes, skincare subscriptions, digital tools, and learning platforms. Behavioral research suggests that consumers are more open to structured habit support in January, particularly when commitment feels more manageable. Gift card redemption also supports this category, with NRF reporting $29 billion in holiday gift card spending this year, much of which will be redeemed in Q5. ## Key Takeaways Q5 is not a passive extension of the holiday season, but a high-intent, behavior-driven window where consumers shift from gifting to self-investment. With 87% of Q5 shopping occasions resulting in a purchase, advertisers who remain active can capture demand that is often more decisive than pre-holiday browsing. The most successful Q5 products share common traits — they reduce friction, support routines, and offer immediate value. Whether through starter kits, bundles, or entry-level offers, Q5 winners meet consumers where motivation is high but attention is selective. For brands advertising on the open web, Q5 represents a strategic opportunity to extend performance beyond traditional holiday peaks and to build momentum that carries through the rest of the year. ## Frequently Asked Questions (FAQs) ### Which product categories naturally see a massive surge in demand in Q5, and how can they be identified for the open web? Product categories that show significant momentum in Q5 are health/fitness/wellness, home goods and decor, self-improvement, beauty, and apparel. The common thread in all of these is that they align with post-holiday motivations like self-investment, renewal, comfort, and utility. Health and wellness categories are particularly lucrative in Q5 as consumers look to activate their New Year’s resolutions. Behavioral data suggests that searches on the open web for “home organization ideas,” “wellness subscriptions,” and “January fitness deals” are indicators of this seasonal trend. ### How can brands use Q5 to effectively sell non-resolution products like apparel and electronics, and what are the best product bundles to promote? Not all Q5 purchases are driven by resolutions, with many also occurring as a result of holiday gift card-giving and post-holiday cash windfalls that make value- and practicality-based purchases more likely. Consumers often spend beyond the initial value of products in this time, especially when promotions and bundles enhance perceived value. Compelling bundles and add-ons are a good move for brands in these categories, such as a winter coat with a glove or hat add-on. Electronic bundles with free cases or a trial subscription to software also work well in Q5. ### What role do “starter kits” and high-entry-level products play in Q5 for subscription services and repeat-purchase brands? Starter kits and entry-level products are effective in Q5 because many consumers are exploring lifestyle shifts at this time, without committing to high-tier purchases. These offer a lower barrier to entry, enabling consumers to try before they fully invest. Curious consumers are an ideal target market for these options, as they’re more cautious about spending but still committed to making small changes. --- ### The Smart Buyer's Bidding Playbook: Max Conversion vs. Target CPA for Growth URL: https://www.taboola.com/marketing-hub/max-conversion-vs-target-cpa/ Last Modified: 2026-03-15 13:46:37 When choosing the right automated bid strategy for native advertising at scale, the decision almost always falls between max conversions vs. target cost per acquisition (CPA). Which one drives better results, scales, or protects margin? To answer some of these questions, we’re joined by Nadim Batista-Kuttab of Xevio, one of the largest native advertisers in the world. Nadim’s team regularly manages campaigns spending between six and nine figures across global markets, testing automated bidding strategies through thousands of experiments. The Xevio team’s philosophy is grounded in practical, scalable dynamics, with the understanding that you can never out-optimize insufficient data. Max conversion is the strategy that unlocks growth and fuels platform learning, while target CPA is a precision tool to be used once campaigns have matured or hit a plateau. Together, these strategies form a two-phase system that supports aggressive scaling without compromising long-term profitability. https://www.youtube.com/watch?v=Ur-bKKz1s0M ## Bidding for Maximum Data Acquisition: Why Max Conversion Wins at Scale When launching a new campaign or attempting to scale an existing one into a higher spend bracket, your goal isn’t to protect every penny of margin. Instead, you need to think about accelerating data acquisition. Native campaigns operate in large-scale auction environments with billions of daily impression opportunities. The only way to train the algorithm effectively is by giving it the freedom to explore, which is exactly what max conversion bidding does. Max conversions tells the system to collect the highest possible volume of conversions for your budget, regardless of short-term CPA. In reality, this means that the algorithm can test new publishers, enter broader auctions, bid more competitively in high-value environments, and gain deeper insight into which placements, audiences, and formats work best for you. As Nadim puts it: “Bidding strategy, super easy: Max conversion, that’s it. That’s currently what’s working. Everything else is not ideal for scaling.” This approach gives the platform the ability to explore during a campaign’s early run, giving it wide exposure so that it can identify patterns across placements, geographies, and creative assets. Restrictive bidding at this early stage prevents the algorithm from gathering a crucial baseline that you’ll need for long-term effectiveness. A constrained strategy at this stage forces the system to make precision decisions before it’s had the opportunity to learn the underlying distribution patterns on why conversions are occurring, which results in unstable delivery, lower reach, and erratic CPA as the campaign progresses. Xevio data supports this, and they recommend that every campaign start with max conversions until the scale naturally increases to a threshold dictated by market size. In the U.S., this may occur at budgets above $20,000-$30,000 per day, while smaller markets like France or Spain could see a saturation point around $5,000-$10,000 per day. Max conversions is the only reliable way to reach these levels because it continually drives the discovery and momentum that feeds machine learning tools like Realize. ## Bidding for Consistency: When to Use Target CPA Target CPA is also an effective bidding strategy, but only when used in the right moments. Nadim’s guidance here is clear: This system is not a starting point and only works once a campaign has reached maturity, meaning that the tool has collected sufficient data to predict conversion probability with a reasonable level of stability. When used too early, target CPA can limit campaigns. When a campaign is transitioning from expansion to consistency, this is the point to assess moving to a target CPA strategy. Nadim says that, “Once you’ve hit the plateau on your scale, you can switch to target CPA for more consistent performance.” A plateau usually occurs when adding budget stops generating significant performance increases, when the system has exhausted all the available high-quality traffic for your creative set and targeting goals. At this point, the algorithm doesn’t need a wider net, but instead should be maintaining efficiency on the volume of traffic you’ve already captured. With target CPA, you’re essentially telling the platform to return conversions at a predictable cost, locking in your profitability once you’ve already gained hundreds, or even thousands, of conversions that the platform can analyze and work from. When you reach this point, target CPA becomes an effective stabilizer without restricting performance. ### Scaling Plateaus A scaling plateau occurs when your campaign reaches its natural limit within a given set of conditions. This may be due to market saturation, ad fatigue, audience recycling, or limited publisher inventory in a specific location. Under max conversion strategies, this plateau typically happens when increases in daily budget stop producing net-new conversions. At this point, the algorithm has learned enough about your environment that further exploratory traffic becomes less useful. Target CPA is an effective tool for navigating this, as it stabilizes delivery while you refresh creative, adjust your funnel, or prepare new tests. Because the system has already mapped your performance boundaries, it can operate with more precision. ### Profit Protection Target CPA also performs well when your priorities for your campaign shift from expansion to profit protection. Brands that have stable funnels or consistent lead goals often need predictable acquisition costs to manage budgets. When your focus is protecting ROAS or overall margin, target CPA becomes an effective way to ensure that you’re not overpaying for incremental conversions. But, it must be enabled only after the system has reliable performance history. Implemented too early, it limits reach, prevents learning, and often results in under-delivery. ## Your Safety Net: Implementing the “Parachute Rule” While automated bidding strategies like max conversions and target CPA manage the majority of auction decisions, human oversight is still essential. Nadim is clear that automated rules used for daily optimization are generally not helpful, and tools that automatically pause publishers for high CPC or low CTR, along with micro-managing placements, can also interfere with machine-learning bidding. The one exception to this is the Parachute Rule. This rule is not designed for optimization; rather it’s designed for disaster prevention. The Parachute Rule acts as an automated failsafe that stops a campaign only when something has gone objectively wrong. A common trigger might be CPA rising to 3x the expected amount after a threshold of at least $5,000 in spend. The purpose is to catch situations like broken tracking, incorrectly loaded landing pages, conversion-rate anomalies, or unexpected creative disapprovals. Without this safeguard, a malfunctioning campaign could spend tens of thousands of dollars before anyone manually detects the issue. The Parachute Rule ensures that large-scale advertisers running high-velocity budgets remain protected from technical failure while still preserving the flexibility that automated bidding strategies need. ## Beyond Bids: Where Marketers Find True Leverage The goal of automated bidding is to free marketers from having to spend extensive time manually controlling campaigns. By eliminating thousands of manual CPA adjustments, max conversions and target CPA gives marketers time to spend on activities that create more impact. As Nadim says, “All you have to do is really focus on the ad experience and the CTR of the ad itself, as well as the post-click performance.” These two areas, creative and conversion funnel optimization, are where human judgment, brand expertise, and experimentation outperform automation every time. Tools like Realize mean that bidding has become a more hands-off process, with marketers now able to focus on refinement instead, such as testing new angles, tightening value propositions, reworking lead workflows, and improving landing pages. These adjustments all impact customer lifetime value, conversion rate, and profit in a way that no bidding strategy can do alone. ### Creative Refresh Creative is still the strongest performance driver in performance advertising. The feed environment in native ads, for example, rewards content that blends into editorial ecosystems while still sparking curiosity. Even campaigns spending $10,000 to $50,000 per day can run for months before hitting true creative fatigue, especially in large markets with billions of daily impressions. Maintaining longevity requires constant iteration. Xevio’s philosophy is simple: never stop testing. Even when a creative appears to be a winner, introduce small-percentage traffic tests for new thumbnails, alternative headlines, or variations. A 0.1% CTR improvement at scale can open millions of additional impressions or lower your CPA. With Realize, the system automatically understands which creative-engagement patterns correlate with conversion likelihood. ### Conversion Rate Optimization The second major lever is post-click optimization. Regardless of how efficiently the platform acquires traffic, the landing page experience determines whether that traffic converts. Brands that optimize their forms, product pages, lead flows, and checkout sequences consistently outperform those that rely on algorithmic bidding alone. Xevio’s team monitors lifetime value, customer-quality metrics, and post-click engagement to adjust their pages accordingly. Improving conversion efficiency reduces CPA and improves scalability, because every improvement in post-click performance compounds the value of the traffic that Realize delivers. ## Key Takeaways Max conversion and target CPA are not competing strategies, but sequential strategies. Max conversion should be your default choice when launching, learning, or scaling. It accelerates data acquisition, feeds the Realize engine the best possible signal set, and unlocks rapid expansion across large markets. Target CPA only becomes valuable once you’ve reached a scaling plateau and need more predictable, stabilized performance. The Parachute Rule functions as your safety mechanism, protecting budgets from rare technical anomalies while leaving room for the algorithm to optimize. Ultimately, the greatest performance gains come not from bid adjustments, but from what you do with the time that all this automation frees up. Creative iteration, CTR optimization, landing page refinement, and funnel performance are where marketers excel. When used correctly, the combination of max conversions and target CPA forms a framework for scaling native spend from five to six to nine figures. If you’re unsure whether your campaign is ready to scale, or which bidding strategy fits your current growth stage, consult your Taboola account manager. They can assess your scale potential, interpret predictive analytics, and ensure your strategy aligns with campaign maturity. --- ### Scaling Your Media Budget: The $300/Day Minimum and Finding Your Saturation Sweet Spot URL: https://www.taboola.com/marketing-hub/scaling-media-budget/ Last Modified: 2026-03-02 08:40:44 Scaling media spend is often portrayed as a simple matter of increasing budgets once a campaign demonstrates early promise. In reality, meaningful scaling that takes a brand from modest acquisitions to sustainable, high-volume profitability requires a more structured approach that can interpret market signals and have a clear sense of when to increase and when to hold steady. A disciplined understanding of how performance shifts as budget grows, rooted in data, is essential. We’ve turned to Nadim Batista-Kuttab of Xevio, a performance marketer known for managing some of the highest-spend native advertising programs online, to break down the essentials of budget allocation and scaling across a complex platform, and provide expert insights into how to maximize profitability and avoid wasted spend. https://www.youtube.com/watch?v=Ur-bKKz1s0M ## The Mandatory Minimum: Why Your Campaign Needs a Budget Floor One of the most frequent mistakes advertisers make when entering the native performance space is starting too conservatively. A campaign launched without sufficient budget does not just underperform, it fails to generate the data density that machine learning (ML) systems require in order to optimize most effectively. Nadim emphasizes that most campaigns, whether focused on lead generation or e-commerce, need at least $300 per day to exit the learning phase of ML and create a stable environment for campaign optimization. This number applies across verticals, placements, and geo locations. A mobile-only iOS campaign needs the same minimum as a large-scale lead gen initiative, because the platform must be able to observe enough impressions, clicks, and conversions to understand signal patterns and target high-intent users. Campaigns launched below this threshold rarely reach their potential because the algorithm doesn’t receive enough information to distinguish meaningful behavior from noise. At $50 to $100 per day, artificial intelligence (AI) systems often end up competing for low-quality inventory or delivering to inconsistent placement groups. The result is an uneven cost-per-acquisition (CPA) trajectory and a misleading impression of poor channel fit. Nadim stresses that advertisers often evaluate the wrong variable when judging campaign viability: The issue isn’t usually creative or the vertical, but rather the lack of budget required to gather high-quality performance signals. For performance advertising, especially with the Realize engine powering bidding decisions, this minimum matters even more. Realize evaluates impressions based on predicted engagement, contextual alignment, user behavior patterns, and post-click conversion likelihood. These predictions depend on rapid and reliable feedback loops. When spend is too low, Realize can’t build an accurate performance model, and the campaign risks becoming trapped in a low-quality auction environment. Setting a budget floor ensures that learning accelerates instead of stalling, allowing the campaign to reach a viable optimization state more quickly. ## The Scale-Up Indicator: Accelerate Based on CPA, Not Hype Scaling should never be driven by guesswork, excitement, or pressure to meet quarterly spending goals. Nadim explains that the only reliable indicator that a campaign is ready for increased spend is performance relative to your target CPA. When a campaign begins to approach, or consistently hits that target, it signals that the algorithm understands where to find high-quality users and can replicate that success at a larger budget. This is why campaign spending around $500 per day and achieving strong performance can often scale to $5,000 per day with minimal disruption. The system already knows what user patterns correlate with conversion and which placements produce the most stable results. Increasing spending is not giving the model a new job, but simply allowing it to execute the same logic at a larger scale. Nadim points out that this phenomenon is especially true on Run of Network, where a campaign is distributed across a network of websites, or the public marketplaces of broad exchange inventory. When a campaign has access to a wider publisher set, the algorithm has enough diversity to expand without running into immediate auction constraints. Not all situations support fast scale, though. Hyper-targeted campaigns tend to saturate early because they restrict the algorithm’s ability to maneuver. If you’re targeting a very small list of publishers, a single demographic, or a limited geography, increasing spending too quickly forces bids upwards and fragments performance. In these situations, Nadim advises that it’s best to work with your Realize account manager before scaling too aggressively. The team can estimate how much inventory a campaign is capable of absorbing based on current cost-per-clicks (CPCs), click-through rates (CTR), and click-to-message (CTM) levels, along with broader market behavior. Scaling based on CPA stability keeps the whole process more disciplined. It prevents advertisers from interpreting short-term wins as long-term capability and ensures that scale only happens when the data demonstrates readiness. ## The Saturation Plateau: Recognizing Diminishing Returns No matter how strong a campaign is, every market eventually reaches a point where adding more budget no longer produces proportional returns. Nadim refers to this moment as the saturation plateau. Understanding when this occurs is essential for long-term profitability. The plateau appears when the campaign absorbs most of the available high-quality inventory and cannot expand further without pushing into less efficient placements. When this happens, each incremental dollar begins to produce fewer conversions and CPA may begin to rise. This plateau is not a sign of campaign failure, though: It’s a natural state of growth in markets with finite impression pools. In the United States, which has one of the largest and most diverse publisher ecosystems in the world, Nadim notes that saturation often occurs somewhere between $20,000 and $30,000 per day. Larger European markets like Germany show similar behavior, but smaller markets like Spain or France may see saturation earlier, sometimes around $5,000 to $10,000 per day, simply because the available audience size does not support deeper scale. Once the plateau becomes visible, the best response is to stabilize rather than push harder. Reducing budget slightly can help restore equilibrium, allowing the campaign to settle into its Maximum Efficient Daily Spend (MEDS), which is the highest spend level at which CPA stability and conversion volume remains optimal. After stabilizing, advertisers can focus instead on creative rotation, funnel refinement, and post-click optimization. These improvements can unlock additional scaling opportunities, since small boosts in CTR or conversion rate often result in disproportionately large performance gains at higher spend levels. Experienced media buyers understand that scale is not linear: It accelerates early, slows as markets absorb budget, and eventually reaches a sustainable plateau. Recognizing this pattern prevents wasted spend and supports long-term growth. ## Bidding for Growth: Max Conversions vs. Target CPA The bidding strategy you use must align with the phase that your performance marketing campaign is currently in. Whether you're scaling or stabilizing, advertisers must understand what the campaign’s current goal is, to best align placements and strategy with this overall outcome. ### Max Conversion (Scaling Phase) In the early and mid stages of scaling, advertisers need the algorithm to explore more aggressively, test new placements, and uncover patterns that lead to more efficient conversions. Nadim explains that max conversion bidding is the strategy best suited for this phase, because it gives the system freedom to discover opportunity pockets without artificially limiting CPA. With max conversions as the goal, Realize can assess impression value dynamically, allocating spend to the placements that demonstrate the highest likelihood of producing conversions at scale. This flexibility is crucial during growth. You should opt for a performance advertising platform which supports vast open-web inventory, and max conversions ensure that the model can rapidly evaluate which contexts and user behaviors correlate with conversion intent. Nadim notes that Realize has become sophisticated enough to eliminate the need for manual bid adjustments across individual sites, replacing thousands of micro-decisions with real-time intelligence. During scale, max conversions accelerate learning and ensure that campaigns build a strong performance foundation before any strict cost controls are introduced. ### Target CPA (Stabilization Phase) Once a campaign reaches its saturation plateau and the MEDS becomes clear, the priority should shift from exploration to consistency. At this stage, target CPA becomes the preferred strategy because it reinforces predictable performance. Target CPA tells the algorithm to maintain the CPA stability, even if it means restricting further exploration. Because the system already understands where conversions are coming from, this constraint helps lock in profitability and ensures the campaign operates efficiently at higher volumes. For many advertisers, switching to target CPA too early is a common mistake. Nadim stresses that target CPA should only be activated once the campaign has already reached stable performance at higher spend levels. When applied prematurely, target CPA limits the system before it fully understands the advertising environment, slowing learning and causing the campaign to stagnate. ### The Only Manual Rule: The “Parachute” Safety Net Although Realize eliminates the need for most manual intervention, Nadim supports using one specific rule: the “parachute” safety net. This rule caps spend when CPA rises dramatically above your normal range. It isn’t meant for optimization, but rather prevents runaway spending during unexpected issues, such as broken tracking, malfunctioning lead forms, or landing page outages. If CPA suddenly triples after a certain spend threshold, the parachute rule pauses the campaign until you can diagnose the issue. It’s the only manual safeguard that Nadim consistently recommends because it catches operational failures, rather than actively influencing your budget optimization strategy. ## Key Takeaways Scaling native advertising is not a matter of intuition or aggressive budget. Instead, it’s a measured, data-driven progression that unfolds in defined stages. Nadim’s approach makes it clear that campaigns must begin with sufficient budget to accelerate learning, scaling only when CPA performance proves that the model is ready and can stabilize once the saturation plateau is reached. Growth is fastest when the algorithm has room to explore and is the most sustainable when it transitions to disciplined controls at the right moment. Max conversions drive intelligent expansion, while target CPA secures long-term stability, with the parachute safety net ensuring operational reliability when unexpected issues arise. Ultimately, scaling is both a science and a partnership. Even the most experienced advertisers will encounter differences in scale potential across markets, verticals, and inventory conditions. Because those dynamics vary significantly by geography, the most effective step is to consult your Taboola account manager when you’re thinking about scaling budget, testing a new country, or evaluating a campaign’s ceiling. They can help interpret the true scale potential of any market, provide predictive insight into inventory depth, and guide you toward the most profitable path forward as your campaigns grow. --- ### Identify and Improve Low Conversion Creative Elements with Realize URL: https://www.taboola.com/marketing-hub/improve-creative-conversions-realize/ Last Modified: 2026-06-10 10:21:51 Native advertising is often misunderstood — it’s not simply display, social, or search marketing, but a different channel entirely that requires a tailored approach. When advertisers understand this, they begin to see significant improvements in CTR, conversion rate, and ultimately ROAS. We’re joined by Nadim Batista-Kuttab of Xevio, a leading expert in scaled native performance, creative testing, and open-web optimization. Nadim has managed millions in monthly spend across global markets, with the Xevio team known for diagnosing funnel problems with clarity and creative innovation. Here, Nadim breaks down how advertisers can use Realize to identify why creative isn’t converting and how to fix these issues through structured diagnostics, smart testing, and platform-specific best practices. https://www.youtube.com/watch?v=ncvZ766nWt4 ## The Native Creative Philosophy: Storytelling and Filtering Native advertising operates on a creative philosophy that’s fundamentally different from platforms like Meta or TikTok. On social, users are scrolling for entertainment and are used to seeing polished, aggressive sales messaging. On the open web, users are consuming content for information — think reading articles, researching solutions, and comparing options when looking to buy something. As a result, native creative needs to look different. Clicks here are high value, and Xevio routinely sees two minutes longer engagement on pages when using Realize — that’s a significant signal that users are deeply engaging with content. But, to earn a click, advertisers must build curiosity and filter for the right users, rather than trying to sell to everyone. Nadim puts it simply: “The image is a stop sign to stop people from scrolling, and the headline is the reason they click.” This shift in thinking reframes how marketers should be building, testing, and optimizing creative. The goal isn’t to push every benefit into a single image: Instead, it’s to attract high-quality users who will convert after the click. ### The Filter Filtering is the foundation for any native creative strategy. On the open web, you can’t rely on tight audience targeting like you can with a platform such as Meta, and you can’t assume that a publisher will deliver your ad to a highly refined segment, so your creative must do this work for you. Headlines aren’t clickbait, but rather, segmentation tools — they speak directly to your audience and the problem they’re trying to solve. Users know whether the message is relevant to them, and those who aren’t interested will simply scroll by, which saves on budget. Those with genuine interest will click. Some examples include: - “Seniors do this to ease back pain” - “Parents are shocked by this new sleep solution” - “Men over 40 are using this simple morning routine” On Realize, headline variations contribute to ad-level policy labels and targeting signals that the algorithm uses to optimize delivery. That’s why it’s essential to get your headlines right and use them as an audience-refining tool. ### The Hook While filtering addresses who should click, your hook is persuading them why they should click. Aggressive claims like “click here to cure back pain” are often flagged. Instead, neutral and educational framing works well, as it mirrors why people are browsing online in the first place — to find a solution to their problem, or to learn something new. Nadim explains that the best native ads “aren’t particularly clear on what they’re selling,” and instead present an invitation that hooks the curiosity of the reader. Remember, the conversion happens after the click, not within the ad itself. Hooks should be intriguing but always truthful, with a problem-focused base that’s broad enough to appeal while remaining focused on the target audience. This is where Realize works well, with the creative dashboard making it easy to test dozens of compliant headlines and automatically identifying the ones that are working best. CREATIVE TESTING AND ITERATION VIDEO HERE ## The 4-Pillar Diagnostic: Isolating Funnel Breakdowns When a new campaign is underperforming, many advertisers default to scrapping their plan and starting again with new landing pages, new ads, and new audiences. Nadim warns against this approach, though: Instead, remember that native funnels are a complex web of ad signals, publisher behaviors, site contexts, and more. If you change everything at once, you lose the ability to truly isolate a single failing element. Instead, try using a 4-pillar diagnostic method, with each step representing a stage of the funnel where a conversion could fall off. The goal is to identify where the problem is before making changes to campaigns. Pillar Symptom Diagnosis Solution 1. Ad Performance - High CPC - Low CTR - Inconsistent engagement The creative isn’t doing its job at filtering or hooking. Users aren’t stopping or clicking. Test stronger headlines, with curiosity-driven frameworks and clearer targeting filters — Realize’s creative variants, live performance labels, and policy tagging help here. Native-style images also outperform social images as they blend more effectively into the editorial environment. 2. Content/Landing Page - Low landing page CTR - High bounce rate The ad message and landing page message don’t align. Users don’t find what they’re expecting once they click. Adjust the content narrative on your landing page to directly address what the headline promised. Improve flow and readability with consistent copy. Realize offers content-level insights and click-to-scroll indicators to help you identify where exactly on the landing page the dropoff is occurring. 3. Conversion Rate - Strong CTR but few form submissions or checkouts - Dropoff at specific funnel points - High engagement with low action Something downstream is broken. This can be a technical element, like slow loading or missing payment information, or informational problems such as unclear benefits. Fix friction and strengthen emotional reassurance to encourage a conversion. Simplify completion steps and add additional trust signals. Realize can help identify event-level issues that highlight whether the problem is back-end technical or a creative misalignment. 4. Product/Offer - Good traffic, engagement, and user behavior, but poor sales - Low LTV The issue is no longer creative, but likely with the product, price, or post-purchase experience. Revise pricing, offer pricing tiers and bundles. Longer risk-free trials can also be helpful. Advertisers often blame the product too early in the process, which is why this is the final pillar. Nadim insists that you must “understand what element of your funnel is causing issues before you write off the platform or product.” ## Innovation vs. Imitation: The Cost of Stealing Creatives One of the most common and damaging mistakes in native advertising is copying someone else’s creative. Nadim sees this constantly — a successful ad going viral in a competitive category and dozens of companies rushing to copy it. These replicas almost never perform well because the original has already established itself. The algorithm knows them best, and users are accustomed to seeing that version. Their position is stronger and their placement prioritized, so engagement is higher. The copycats only get the leftovers after this, the low-intent clicks and expensive CPCs that waste budget. As Nadim says, “If somebody’s running something and you take that something to run yourself, you will be left with whatever’s left in the pot.” Innovation wins here, not imitation. Realize amplifies that advantage with the freedom to generate and test an unlimited number of creative concepts. With AI support, advertisers can create original, attention-grabbing images that outperform competitors. "We can, as advertisers, really throw a lot at the wall and see what sticks,” says Nadim. “One of our top solar ads in Germany was floating solar panels in space." This is native creativity at its best: unexpected, striking, and story-driven. While not realistic, it was visually compelling and earned the curiosity of users. ## Beyond Native: The Power of New Formats While classic native ads with a headline and thumbnail remain the highest volume format, expanding into additional formats can unlock new channels of ROI. Realize supports these formats with unified tracking, cross-format creative insight, and asset reuse features. ### Motion Ads Motion ads introduce movement to otherwise static creative through GIFs, micro animations, or short, silent video clips. Movement is one of the best ways to stand out against static content, capturing attention without feeling intrusive. Nadim notes that motion ads often outperform static creatives placed directly next to them. Realize supports this with the ability to convert imagery into lightweight motion variants. ### Display Inventory Display advertising within Realize's network surprised even an experienced marketer like Nadim. When Realize expanded its inventory to more display publishers, Xevio initially expected limited results due to Google’s dominance in this space. Instead, performance exceeded expectations. For one large e-commerce brand, he says, display spend on Realize now outperforms native ROI, all because the advertiser expanded from native to display with proven creative themes, filtering, and storytelling. Display shouldn’t replace native, but be used as a building block to expand into multiformat advertising at scale. ## Key Takeaways Native advertising requires a fundamentally different creative strategy to social or search. Filtering, storytelling, and curiosity are key, rather than aggressive product-pushing. To identify low-converting elements of campaigns, advertisers must use the 4-pillar approach rather than guessing, or rebuilding entire funnels from the ground up each time there’s a problem. Realize accelerates this process by providing structured creative management, automated variant testing, policy-safe headline generation, and multiformat scaling across native, display, and motion ads. For advertisers looking to scale, using tools like Realize makes this process simple and effective. --- ### Innovation vs. Imitation: Why Stealing Creatives Guarantees Failure URL: https://www.taboola.com/marketing-hub/innovation-vs-imitation-for-ad-creatives/ Last Modified: 2025-12-16 15:16:33 In performance marketing, everyone wants the fastest route to a winning ad. Imitating successful campaigns can seem like a shortcut to success, but Xevio CEO Nadim Batista-Kuttab strongly cautions against it, explaining that copying a creative will not only fail to recreate an ad’s original success, but also place advertisers at a disadvantage. That’s partly because, in performance marketing, true engagement happens on your landing page, not in the ad itself. For that reason, creative originality matters even more here than in social ads. “You don't have a 60-second video like you do on other channels, like Meta or Instagram, where you can really explain and sell,” reasons Nadim. “That'll be done later, in your content piece, after the click.” Below, we break down the dangers of copying creatives, along with Nadim’s recommendations for scaling profitably. https://www.youtube.com/watch?v=ncvZ766nWt4&feature=youtu.be ## The Imitation Trap: You Only Get Residual Clicks Everywhere you look, you see a competitor’s ad. It’s clearly getting clicks and sales, and you understandably want in on that action. It’s only natural to assume that the creative will work the same magic for you. Unfortunately, that doesn’t usually happen in native advertising. With native, as soon as a creative enters the auction, performance history starts to compound. When the ad is a duplicate, that history can work against you. “If somebody's running something, and you take that something to run yourself, you’re not going to be as successful as the first person to run that creative,” Nadim says. “The reason is quite simple: They’re more established than you, because they launched the creative before you did. They'll be higher up in the feed, they'll get a more relevant click, and you’ll be left with whatever's left.” This caution comes down to how the ad auction works. Algorithms reward relevance, quality, and performance history: When you copy an existing ad, you’re competing against an ad that is both earlier in the feed and cheaper to deliver to the platform. Below are two benefits the original advertiser gains that you won’t be able to duplicate. ### First-Mover Advantage When you steal an existing creative, you’re copying from an advertiser who has already built strong engagement signals. This includes: - Click-through rate (CTR). - Relevance scores. - Consistent positive conversion signals. - Audience fit over time. These signals tell the platform, “This ad performs. Show it to more people at a lower cost.” You can copy the image, the headline, and even the landing page, but you can’t copy the performance history that gives the original ad momentum. Without that, your cost per acquisition (CPA) will be higher. ### Dominant Position When you run a copycat creative, the original advertiser stays ahead of you. That advertiser has already: - Trained the algorithm with strong engagement and conversion data. - Accumulated positive performance signals across placements. - Captured the highest-intent audiences at the lowest click costs. Your version competes with that original creative but carries a weaker expected CTR. As a result, the platform pushes it lower in the feed, leaving you with more expensive and lower-intent traffic. With rising competition and higher native ad costs, being deprioritized in the feed makes scaling significantly more costly. ## The Rule of Continuous Testing: Outperform Your Top Winners In marketing, even a top-performing ad has a shelf life. Successful advertisers understand that sustainability is the goal, and sustainability comes from continuously challenging your best performers. “My testing philosophy is, ‘Always be testing,’” Nadim says. “Always try to outperform your top performers. It's a good position to be in: If you found something that performs, throw things against it at low scale and see if it works better than what's scaling.” Even a 0.1% CTR improvement can be meaningful at volume, especially if it’s supported by A/B testing, as it can mean the difference between your campaign plateauing and scaling. Consistent testing also helps combat creative fatigue, which often causes engagement on a high-performing ad to drop 20-30% week over week toward the end of its run. Put simply, your best-performing ad today won’t stay on top unless you keep it competitive. ## Creative Freedom and Out-of-the-Box Thinking Imitation limits growth, and stagnation stalls performance, so what actually does work in 2025? Bold, scroll-stopping creativity. Display advertising offers far more creative freedom than you may realize: While compliance restrictions protect against misleading content, the creative range is still endless. “One of our top solar ads in Germany was floating solar panels in space,” says Nadim. “It didn't make a lot of sense — we aren’t going to be selling solar panels in space! — but it was a very catchy image. It looked super nice: It was AI-generated and touched up by our design team. That is creative freedom 101.” Was it literal? No. Did it stop the scroll? Absolutely, and that’s the point: Users on the open web aren’t interested in being sold to, they’re looking for compelling content. Your ad should feel like the gateway to the story, not a product pitch. This extreme example illustrates how the most effective ads often defy conventional expectations. The key is to: ### Innovate Visually Before your headline can earn the click, the visual must earn the pause. Display ads live among editorial content, so your image needs to work hard to break scanning patterns. This is where creativity — and, increasingly, artificial intelligence (AI)-assisted concepts — offer a competitive edge. Highly effective native visuals tend to: - Break visual patterns. - Trigger curiosity. - Connect abstract ideas to familiar experiences. - Use AI-generated concepts to explore ideas that would otherwise be expensive or time-consuming to create. AI makes it easier to generate unique content that stands out in user feeds. When used intentionally, the technology becomes a creative accelerator, allowing you to test more conceptual visuals and learn from the results. ### Test Volume: Don’t Bet on One Big Idea Once you’ve found a winning creative, it can be tempting to lean into it, but that’s where many performance marketers stall. Nadim emphasizes that variety, not perfection, is what unlocks scalability. “One of the superpowers of open-web display ads and Realize is that we can, as advertisers, really throw a lot at the wall and see what sticks,” he says. Instead of investing all your effort into a single creative, launch multiple small, inexpensive variations. Performance platforms distribute creatives across hundreds of sites, contexts, and audiences. A creative may underperform in one environment, yet deliver surprisingly strong results in another. Volume testing reduces your risk and helps you learn more about what works. A healthy group of creatives might include: - 4–6 visual concepts per theme, including some that are AI-assisted. - 6–12 headline hooks to go with them. - Micro-tests to get a feel for the market before scaling. This approach keeps feeds fresh, reduces creative fatigue, and continuously generates new winners to compete with your top performers. ## Key Takeaways In today’s competitive advertising marketplace, the only way to sustain results is through originality and continuous optimization. Yes, copying a competitor’s high-performing creative can seem like a shortcut, but it places you at an immediate disadvantage: Your ad enters the auction with weaker performance signals, higher CPCs, and lower-intent traffic. Instead, focus on creating original content that stands out in feeds while still being relevant to your brand message. It’s also important to keep your ads fresh. Even a top-performing creative will degrade over time. Test and experiment, keeping a close eye on data and adjusting your campaigns accordingly. The brands that succeed are creating unique concepts that other marketers want to copy, giving them that first-mover advantage. --- ### 5 Must-Use Creative Ad Formats for Total Performance URL: https://www.taboola.com/marketing-hub/creative-ad-formats-for-performance-advertising/ Last Modified: 2026-03-24 08:31:46 As typical social and search channels become overloaded and user fatigue grows, performance marketers have to focus on the best ad format for each channel and for the campaign overall. The current challenge for performance marketing teams is how to achieve total performance: the combination of scale, efficiency, and return on ad spend (ROAS). To succeed at this total performance measurement, it’s important to consider the range of tools available. A painter doesn’t use only one brush, and a modern marketer can’t use only one ad format to fill in the full picture — no single creative format fits every stage of the funnel or every digital channel. With the right creative mix, ad campaigns can achieve high performance on the open web, meeting conversion goals and staying under budget. ## The Performance Creative Mandate: Move Beyond Saturated Channels The solution to marketing success isn’t the same as a decade ago: It’s too expensive and unpredictable to run ads on traditional channels now, and modern ads and better technology platforms offer many new ways to reach and engage audiences. Instead, performance marketers have to audit their current strategy and push toward accessing premium inventory in trusted environments, focusing on creative formats that capture attention and blend into existing user experiences. Performance success depends on matching the right creative format to the right moment across a diverse publisher network. As performance marketing matures to meet new needs, platforms like Realize have emerged to tackle new challenges so that it’s possible to use multiple creative formats without starting from scratch or going over budget. Here are five creative formats to attract new audiences at scale and improve ROAS. ## The 5 Creative Formats Driving Total Performance These creative formats each have their own uses and benefits as part of a broader total performance strategy. Get to know each of them and where they fit best in the funnel. ### 1. Motion Ads: Capturing Attention and Driving CVR The unique benefit of motion ads is that they can bring static images to life and capture attention with short, dynamic creative. For example, the Realize platform allows users to transform static creative into GIFs or short looping videos, running without sound to subtly capture attention. In a busy digital environment, motion ads are an important tool for marketers to deploy to meet conversion goals. They’re also useful in optimizing ROAS, with clients seeing 124% increase in ROAS over time with motion ads. ### 2. Vertical Ads: Scaling Mobile-First Social Assets Vertical ads go hand-in-hand with the popularity of mobile devices for browsing and shopping. Using vertical ads, performance marketers can extend successful mobile campaigns beyond the walled gardens of social channels. They’re designed for maximal visual engagement. Use vertical ads when you’re aiming for both scale and reach, targeting incremental audiences in premium environments. They also have the benefit of allowing you to reuse existing assets while maintaining brand consistency. When you’re using Realize, you get support for both vertical video and static vertical creative. ### 3. Carousel Ads: Showcasing Products and Driving Urgency Carousel ads are a popular way for advertisers to feature multiple products, benefits, and offers in a single ad unit, helping to drive urgency. These multi-frame ads ensure high-impact creative, and they’re best used to drive conversions. Generally, and within Realize, carousels include individual cards that each have their own image, title, and landing page. They’re also interactive, asking users to swipe through images, such as to show off products or tell a brand story. ### 4. Display Ads: Providing High Visibility and Direct Access Display ads are useful for extending campaign reach beyond native formats. They perform consistently across premium environments, offering a high-visibility, efficient option for performance marketers. Use these ads for both reach and consideration, keeping in mind that they use direct publisher supply that ensures better supply path efficiencies. Display ads are also a creative type within Realize, designed to extend reach beyond the feed. They offer flexibility in creative options, running on standard IAB formats with cost per click (CPC) pricing. Realize users can also upload or import existing top-performing social creatives for repurposing into display ads, saving a lot of time and resources. ### 5. Native Ads: Offering Seamless Engagement and High CTR Native ads are ideal for a seamless user experience, as they use images and titles styled to blend in naturally with publisher content. Native ads are best used for driving authentic engagement and high click-through rates. Use native ads when your goal is consideration as well as traffic quality, and when you’re looking to strengthen audience connections. They work well in an always-on approach to achieve both reach and engagement. ## Key Takeaways Each business and marketing team has to choose the right combination of formats for their audience and goals, keeping in mind that total performance success these days is multi-format. Make sure your creative playbook is flexible and diverse, and choose a platform that supports these formats. Performance marketing platforms like Realize are specifically designed to cut through the challenges of user fatigue and overused channels, and they incorporate AI so marketers can optimize for conversions and get the most out of ad spend. Try an initial audit of your creative strategy to see where you can embrace format diversity. Get started here with help from Realize. ## Frequently Asked Questions (FAQs) ### How can I manage and test five different creative formats efficiently without overwhelming my team? Marketing teams can learn a lot and build better campaigns from testing these five creative ad formats, with results like better ROAS and improved conversions. At a minimum, performance marketers should standardize testing processes, use templates, and establish data-driven workflows. Start with hypotheses about your audience, then uplevel or repurpose existing assets into the five creative formats. From there, see what works and pivot quickly to optimize the ad format mix. Getting help from a technology platform is the easiest way to manage and test formats without overwhelming your marketing team, especially a lean team that has to stay within strict budget limits and may not be able to rely on design support. Technology platforms streamline asset transformation and deployment and offer internal tools, plus AI capabilities that can save a ton of time. For example, Realize’s AdMaker feature uses specifically trained gen AI to create and optimize static and motion ads within the platform. ### Since these formats run on the open web, how is brand safety guaranteed across diverse publisher sites? Open-web formats run across diverse publisher sites, and manual management is incredibly time-consuming and often risky. Modern performance advertising platforms need to understand requirements like brand safety and suitability. This requires direct publisher integrations, not just exchanges, along with robust control tools. Look for platforms like Realize, which offer verified inventory free of Made for Advertising (MFA) sites, which can damage brand reputation with their low-quality, money-first approach. Within Realize, integrations with third-party verification partners also ensure that marketers are complying with safety, privacy, and ad quality standards with features like pre-bid filtering, keyword blocking, and customizable allow/block site lists. There are also built-in fraud and privacy protections, along with GDPR support. In addition, contextual targeting analyzes content and metadata on publisher sites to ensure safe placements. ### Can my performance AI effectively optimize campaigns when running native, motion, and display ads simultaneously? Yes, it’s possible for performance AI to optimize campaigns while running native, motion, and display ads simultaneously. The performance AI has to be format-agnostic, solely focused on advertiser outcomes. Successful capabilities and platforms use real-time conversion data captured by implementing a pixel on the site. They can then adjust bids across formats and environments so spend is directed toward the highest value impressions, no matter the ad type. With these abilities, you can optimize campaigns as you go, avoiding wasted budget and getting impactful results. Performance AI, like that in Realize, also helps to qualify prospective audiences through engagement signals. --- ### Why Q5 Is a Golden Opportunity for Advertisers on the Open Web URL: https://www.taboola.com/marketing-hub/why-advertise-on-open-web-in-q5/ Last Modified: 2025-12-16 17:20:09 After the rush of the Thanksgiving to Christmas period, it’s easy for advertisers to pause efforts and turn their attention to the new year. But Q5 — the span of time between December 26 and mid-January — can be a treasure trove for prospecting and converting users, as many consumers are still shopping and often have a different intent than they did during Cyber Week. ## The 6 Reasons Q5 Is a Golden Opportunity on the Open Web In 2024, the period between December 25 and January 1 saw the highest sales growth compared with any other period, according to Mastercard data, and 75% of TikTok users planned to spend the same amount of time or more shopping in Q5 compared to the rest of the year. Self-gifting, New Year’s resolutions, and other trends are all driving the value of Q5, and performance marketers can create a strategy to capture all this activity. As costs rise and the typical channels become oversaturated and less dependable, advertisers are expanding their playbook. Creating new channels for performance includes experimentation and rethinking the typical ways of operating. Taking advantage of Q5 can help marketers finish the year strong, building momentum for Q1 and using cheaper rates to conduct A/B testing. Below, I’ll break down all the reasons you should approach Q5 with intent. ### 1. Ad Costs Drop Because so many brands are accustomed to reducing budgets after the holiday rush, Q5 has less competition for ad space, even as user engagement remains high. With fewer advertisers, inventory becomes cheaper, lowering CPM. Cost per click (CPC) dropped 10% across devices during Q5 of 2024, while impressions surged 94% in the U.S., according to Search Engine Land. As timing is key for performance marketers, consider that Q5 may even extend until late January or up until Super Bowl advertising starts. ### 2. Gift Cards Entice Shoppers Popular holiday gifts include gift cards to favorite stores or websites, with the market for these estimated to be worth $1.24 trillion in 2025. The post-Christmas period is ripe for targeted advertising and deals targeted toward those looking to make the most out of their gifts. Over half of respondents to one survey said they use cash or gift cards they received in the weeks immediately after the holidays, so consider targeting audiences who might be new to a particular brand, or add urgency to ads to use gift cards. ### 3. Consumers Set New Year’s Goals Post-holiday mindsets often shift quickly to self-care and setting resolutions, whether exercise, diet, career goals, or more. According to TikTok data, 94% of users have at least one personal goal for the new year. These shifts can be prime time for advertisers, particularly in the wellness space, to attract and convert new, focused users. ### 4. It’s Time to Treat Yourself Self-gifting becomes popular in Q5 as consumers turn their attention inward after a busy season buying gifts for others. A Google survey found that 45% of Americans enjoy shopping after Christmas, and the percentage of those shopping for themselves was significantly higher during Q5 than the rest of the season. ### 5. Shoppers Are Laser Focused In Q5, consumers tend to have focused intent on what they’re shopping for. Post-holidays, shoppers are very likely to buy, with 87% of shopping occasions resulting in a purchase. It’s an opportunity for performance marketers to engage and re-engage users and prospects to drive conversions and sales. ### 6. There’s Lots of Appetite for Bargains On top of self-gifting and an appetite for New Year’s goals, there can be excellent post-holiday sales for consumers to enjoy in Q5. One survey found that 77% of consumers said they could find better deals after the holidays versus during them, with 51% planning to take advantage of these deals to make big purchases. These discounts and offers may be the best of the year, depending on the industry, so advertisers can get creative with images and copy to entice and convert users. ## Key Takeaways While the post-holiday period in late December and early January may typically be quiet, savvy advertisers can take advantage of this Q5 period to attract and convert high-intent users. In Q5, ad costs are usually lower, consumers are focused, and performance marketers can tailor offers accordingly to finish out the year on a high note. ## Frequently Asked Questions (FAQs) ### Why do advertising costs drop significantly during Q5, and how does this create a budget advantage? From late December to mid-January, many brands have typically paused ad spending, leading to reduced competition for ad space and lower CPM and other metrics. This creates an opportunity for performance marketers to extend their reach, perform testing affordably, and acquire new prospects and customers. With cheaper ad space, advertisers can maximize their reach and ROI, nurture leads and expand lists, build brand presence, and increase visibility and share of voice when other businesses might not be present. This is a time to take full advantage of Realize features that can help you move quickly, like automated bidding and automated A/B testing, dynamic retargeting to reach users with product-specific ads, and automated budget protection. ### How does the consumer's mindset during Q5 differ from Q4, and what kind of messaging is most effective? The Q4 consumer mindset is typically urgent and external, with a focus on gifts for others. Q5 brings a period of self-gifting and self-improvement, along with bargain-hunting. Performance marketers and advertisers can use these trends to craft messaging accordingly. You may focus on consumers looking for gifts for themselves, as well as New Year goal setters who want to change habits or improve other aspects of their lives. Messaging around value and big discounts can also be very successful, along with focusing on home items or winter gear that consumers might not have purchased in the lead-up to Christmas. Q5 can also be a great time to build brand loyalty with newer customers from Q4. Use platforms like Realize to easily create and test various display options like vertical, carousel, and video to see where and how messaging breaks through during this period of time. Consider the automated landing page builder to save time, too. ### What is the long-term strategic value of investing in Q5, rather than pausing advertising until Q1? Pausing advertising for a week or two might not seem significant, but it can be a missed opportunity during the Q5 period. Performance marketers can use this time of lower competition and reduced ad costs to build brand awareness and momentum to strengthen the pipeline for the new year. Restarting campaigns after pausing likely requires higher costs, and the decline in sales and market share can be hard to rebound from. Investing in Q5 can mean a lower-cost time to test creative and engage new audiences, and with platforms like Realize, you can use AI and automation to stay cost-effective in your bidding and A/B testing strategies. --- ### How to Optimize Campaigns Across Different Funnels URL: https://www.taboola.com/marketing-hub/optimize-campaigns-across-different-funnels/ Last Modified: 2026-06-24 12:05:43 Still chasing vanity metrics, tracking click-through rates (CTR) and basic conversion numbers? That may have worked in the past, but the modern performance marketing game requires more than tallying up easy wins. True campaign optimization requires a deeper, more sophisticated understanding of the customer journey, and that journey is rarely linear — it involves multiple touchpoints across various platforms and devices. To succeed in 2025 and beyond, marketers must implement strategies that accurately map this complex journey, leveraging the power of first-party data, artificial intelligence (AI) models, and creative automation to maximize efficiency and return on investment (ROI). ## Accurate Funnel Mapping Effective optimization begins with a precise understanding of how users move from initial exposure to final purchase. If your map is flawed, your entire optimization strategy will be misdirected. ## Key Metrics for Each Page A mature optimization strategy does more than count final conversions — it assigns specific key performance indicators (KPIs) to each stage of the funnel, reflecting the unique goals of that phase. Measuring non-transactional metrics in the early stages enables you to optimize campaigns that drive long-term value, not just short-term transactions. For example, if your time-on-page KPI for the awareness stage drops, you know you have a content relevance problem long before it impacts sales. Funnel stage Primary goal Recommended KPIs Optimization focus Awareness (top of funnel ) Maximize qualified reach and initial engagement. Share of voice, qualified clicks, time-on-page (above benchmark), video view rate (VCR). Creative appeal, placement quality, audience expansion. Consideration (middle of funnel ) Prove intent and generate qualified leads. Lead form submissions, content downloads, high-value page views, cart adds (without purchase). Messaging relevance, offer quality, landing page user experience. Conversion (bottom of funnel ) Finalize transaction and maximize value. Sales, revenue, customer lifetime value (CLV), return on ad spend (ROAS). Checkout flow, pricing, shipping costs, retargeting effectiveness. ## ## ## Data Capture and Tracking ### Data Capture Accurate mapping relies on data capture. The foundation for this is a robust, well-implemented tracking mechanism that captures data across your entire digital property. The core function of a tracking pixel is linked to closed-loop attribution, which involves connecting an anonymous ad impression (the first touchpoint) to the final, identified attribution. Without the granular data captured by a pixel, an advertiser is unable to connect a specific action to the initial marketing investment. Pixels provide the data architecture to construct a seamless, end-to-end customer journey. These tracking pixels are small snippets of code that capture all user activity on a site, from low-level engagement metrics to high-value actions. Pixels collect data points, like IP addresses, browser types, device information, and page views. This step establishes the starting point and the journey path. For closed-loop attribution to work, the system needs the full, high-fidelity log of every step a user takes after the initial ad impression. The pixel provides this detailed log, allowing the attribution model to trace the non-linear path from awareness (TOFU) to the conversion (BOFU) event. ### Conversion Tracking Pixels can track desired outcomes (a purchase, lead form submission, or content download). These measurable, high-value events determine campaign success, and a campaign’s ultimate success is measured by the conversion event. The pixel fires precisely at this point, signaling to the advertising platform that the desired action has occurred. The platform closes the loop and immediately calculates the campaign’s true ROAS or CLV, linking the final dollar amount to the ad spend. ### Cross-device and Cross-platform Linkage Pixels facilitate the use of cookies and unique identifiers to recognize a user even as they pop back and forth between different websites, applications, and devices (e.g., viewing an ad on a mobile phone and completing a purchase later on a laptop). This feature is essential for retargeting and provides the bridge for accurate multi-touch attribution. Without the pixel’s ability to link disparate touchpoints to the same user ID, the closed loop would break — and that’s a problem, considering that customer journeys are rarely confined to one session or one device. The pixel ensures proper crediting of the original top-of-funnel ad impression, even if that conversion happens days later on a different device. When you can track the entire customer journey, from initial marketing touchpoints to final sales outcomes, and use data to directly link marketing activities to revenue, imagine the insights you gain! ## Identifying Bottlenecks Friction along the user journey creates major stumbling blocks. Finding those bottlenecks is necessary for maximizing conversion rates. The first step? Understanding the drop-off points. Common culprits include: - The landing page: High bounce rates often indicate a poor messaging match, slow load times, or confusing layouts. - The cart/checkout process: Complex forms, unexpected shipping costs, or mandatory account creation can cause a significant drop-off. According to a report from Baymard Institute, other reasons customers abandon carts include not being ready to buy (43%), a lack of trust with their payment information (19%), an unsatisfactory return policy (15%), and too few payment methods (10%). Tools like session recordings, heatmaps, and funnel visualization reports offer specific insights into user behavior at each of these moments. ## Prioritizing Fixes Prioritize your fixes based on their potential ROAS lift. Generally, addressing friction in the conversion or high-intent consideration stages yields the fastest and most significant returns, since these users are closest to making that purchase. Fixing a leaky checkout flow or clarifying shipping costs up front, for example, often provides a quicker performance boost than making minor tweaks to top-of-funnel creative. ## A/B Testing and a Multivariate Testing Framework Scientific testing is the backbone of continuous performance improvement. When you plan your A/B tests, be disciplined, and don’t forget the fundamental rule — isolate your variables. Run true A/B tests to compare two versions (Headline A vs. Headline B) while keeping all other elements consistent. Once you’ve established the winners, it’s time for multivariate testing. Automated platforms make it easy to efficiently test various combinations of proven assets (headline, image, description) to quickly identify the highest-performing combination. Every test, regardless of outcome, contributes to internal knowledge. Analyze test results not only by the overall conversion rate, but also by the audience segment. Understanding which creative approaches resonate with specific demographics, psychographics, or device users informs smarter, more targeted campaign planning in the future. ## Scale Creative with High-Visibility Options and Repurposing Refresh creative assets frequently to minimize ad fatigue and keep engagement high. Enterprise platforms facilitate rapid creative scaling through automated tools, which allow advertisers to quickly repurpose high-performing social media assets into native and display formats for the open web. This automation drastically reduces production time. Focus on deploying visuals and messaging across those high-visibility formats, like native placements and vertical video, to maximize impact and blend into the publisher environment. Remember to: - Prioritize storytelling over hard selling: Narrative will hook your audience and provide value. Make the product or service a natural part of the story, not its sole focus. - Design for mobile-first consumption: Since most users consume content on their smartphones or tablets, be sure to optimize visuals and messaging for small screens, keep text legible, and use thumb-friendly calls-to-action (CTAs). - Incorporate text overlays and captions: Since many people watch videos on mute (especially on social media feeds), adding those captions or bold on-screen text communicates your message without sound. - A/B test and optimize constantly: Continue even after conducting the initial tests, as trends and preferences change. Consistent testing helps refine your strategy and avoid ad fatigue. ## Leverage Predictive Signals to Move Beyond Retargeting Taking that next step beyond retargeting, which relies on past behavior, requires using predictive AI. This technology uses historical and real-time data to forecast future customer behavior and campaign outcomes. Predictive AI uses machine learning (ML) and statistical algorithms to analyze huge quantities of past customer data. AI finds patterns and trends humans might miss, and based on these patterns, predicts future events, like which customers are most likely to buy a product, the content customers are most likely to engage with, and the best time to send a message. These tools operate 24/7, continuously monitoring performance and making intelligent, real-time adjustments, and include key applications like: - Tailoring marketing messages, product recommendations, and offers to individual customers based on their predicted behavior. - Creating more accurate audience segments to deliver campaigns to the right people across different platforms and funnels. - Identifying the leads most likely to convert, helping sales teams prioritize (and maximize) their efforts. - Forecasting which customers are at risk of bouncing or leaving, allowing marketers (or AI chatbots) to intervene, troubleshoot, and try to retain them. - Determining which marketing channels, messaging, and timing will be most effective for a given campaign. - Optimizing ad spend by forecasting which campaigns and channels will provide the best ROI. ## Key Takeaways It’s vital to implement full-funnel KPIs, measuring the value of every stage — not just the final sale — to understand true performance drivers. Prioritize fixing leaks in the consideration and conversion stages, and maintain a rigorous test schedule, using A/B testing to build audience-specific intelligence and identify winning creative combinations. Use tools that allow you to quickly repurpose successful assets across different platforms, which helps fight creative fatigue on your team. Finally, embrace predictive AI so you can scale campaigns by targeting new prospects whose browsing behavior predicts high conversion value. ## Frequently Asked Questions (FAQs) ### How does Realize's Matchmaking AI help me scale beyond my current audience and accurately forecast high-value users? Realize’s Matchmaking AI uses sophisticated ML models trained on proprietary data to analyze the engagement and conversion patterns of your existing high-value customers. Then, it scans the web to identify new prospects whose real-time browsing signals — the topics they read, content they engage with — closely match the profiles of your best converters. This capability allows you to go beyond your current audience lists and prospect efficiently at scale by targeting users with a higher likelihood of conversion. ### How can the Social Importer tool streamline creative optimization and testing when running high-volume campaigns? Realize’s Social Importer tool allows advertisers to quickly and easily repurpose their best-performing social media assets from platforms like Facebook and Instagram into display ads for the open web. - It saves time by eliminating the need to design new creative assets from scratch for different platforms, streamlining workflows. - It helps advertisers extend the reach of their social media campaigns beyond the “walled gardens” of social media platforms to a vast network of premium publisher websites (NBC News, Yahoo, and others). - It facilitates the use of various high-visibility ad formats (display, vertical, carousel) on the open web, using repurposed social assets. ### What is the benefit of having the Abby AI assistant integrated directly into the platform for real-time campaign optimization? By integrating Abby AI directly into a marketing platform for real-time campaign optimization, you gain a significant boost in efficiency, speed, and campaign management. This assistant manages your campaigns, using a conversational approach and simple questions to identify marketing objectives and build tailored media plans. Abby AI can translate simple conversational prompts into a fully constructed media plan, including budget allocations between device types (desktop/mobile), freeing advertisers to concentrate on other business goals. By automating the build process based on established best practices, campaigns initiated with Abby can go live faster than those set up manually, enabling you to enter the market more quickly. Abby’s generative AI feature facilitates instant content creation using the same conversational approach. You can ask Abby to “change the background to a sunny beach” or “write a more urgent CTA” without needing specialized image editing software or creative expertise. Because Abby has democratized the testing process, you can quickly generate and test dozens of creative variations to avoid ad fatigue in real-time. --- ### Abandoned Carts: Why They Happen and How to Prevent It URL: https://www.taboola.com/marketing-hub/abandoned-cart/ Last Modified: 2025-11-25 13:12:54 Ideally, the sales funnel for e-commerce companies would be straightforward: A shopper finds your product, perhaps on a search engine results page (SERP), in a social media ad, or in an ad on their favorite digital publication. They click to learn more about it. They read a few articles and reviews, liking what they see. They click to your site and purchase the product. It’s more likely, though, that at some point in the process, the consumer says “forget it,” leaves the product in the online shopping cart, and goes on with their day. In fact, statistics from the Baymard Institute reveal that more than 70% of shoppers, on average, abandon their shopping carts before making a purchase. Let’s take a closer look at why that is, and what you can do about it. ## What Is Cart Abandonment? Cart abandonment, as the name implies, happens when an online shopper halts a transaction and leaves their product in the shopping cart. They might intend to return to make the purchase later, but a majority of the time, those carts are just left to languish. It’s lost revenue and a lost opportunity to create loyal customers. An “acceptable” abandoned cart recovery percentage ranges from 10% to 20%, according to Convertcart. That means, for most businesses, 80% to 90% of lost potential customers are gone for good. ## What Does it Mean for a Business? Cart abandonment costs e-commerce retailers $18 billion per year, with around $4.6 trillion worth of merchandise left in online carts. Cart abandonment doesn’t just lead to lost revenue, though (even if $18 billion is, admittedly, a pretty big chunk of cash to leave on the table): Companies also miss out on the chance to build loyal, repeat customers, especially if shoppers purchase from a competitor instead. Cart abandonment makes it more challenging to boost sales, pull market share from competitors, and scale your e-commerce business. There are ways to win back customers after they’ve abandoned their cart. A report from SellersCommerce shared that cart abandonment emails can help recapture that revenue, citing that shoppers open 45% of abandoned cart emails, 21% click through, and of those, 50% ultimately make the purchase. SMS strategies can also lead to a 58% recovery rate for abandoned carts, since so many people today shop on their mobile devices. Having a solid strategy to recapture customers is good, of course, but it’s still a better business practice to avoid cart abandonment whenever you can, by optimizing and streamlining your check-out process. ## Why Does Cart Abandonment Happen? Cart abandonment happens for several reasons: Personal finance experts, e.g., often recommend cart abandonment as a budgeting and money-saving tactic, suggesting shoppers leave something in their cart overnight to avoid impulse purchases. While this type of intentional spending may account for some cases of cart abandonment, though, a lot of times, cart abandonment is the fault of the seller. Here are some of the most common reasons for cart abandonment, based on data from Statista collected from online shoppers: - Additional charges too high (shipping/tax): 39%. - Slow delivery: 21%. - Must create an account to check out: 19%. - Payment security concerns: 19%. - Complicated check-out process: 18%. These are the top reasons for cart abandonment. Other reasons included unclear pricing before purchase, an unsatisfactory return policy, website crashes, not enough payment methods, or the user’s credit card being declined. ## How to Reduce Shopping Cart Abandonment Knowing the primary reasons for shopping cart abandonment can help you tackle this problem, especially during key parts of the year, like the lead-up to Black Friday/Cyber-Monday, or during hot summer sales times, like Memorial Day and Fourth of July. ### Be Upfront About Pricing, Including Fees It’s easy for brands and e-commerce sites to address the first shopper concern: hidden fees. Follow the standard of companies like Expedia and publish the full price of your product, including any required taxes and fees, right upfront. “Drip pricing,” the practice where you hide fees until the very end, leads to cart abandonment, distrust, and bad will. It can also lead to lawsuits, so being clear about pricing upfront doesn’t just reduce cart abandonment — it may also prevent you from facing compliance issues. For a real world example, look no further than event ticket seller Ticketmaster (and parent company Live Nation), along with re-seller StubHub, who faced legal action for unclear pricing on their websites. Most recently, StubHub agreed to pay out $2.5 million in cash and $20 million in credit for not revealing fees that substantially drove up the cost of their tickets. Most retailers aren’t intentionally deceiving customers: It could be that you’ve adopted a drip pricing model because that’s the way companies have done it in the past. Now that we, as an industry, know better, though, it’s time to change things up. ### Offer Free, Fast Shipping Of course, it can be difficult to disclose the total cost without knowing the buyer’s shipping address. State and local taxes, along with shipping costs, differ. One way to avoid losing customers due to high shipping costs (or slow shipping times) is by offering free shipping across the board, or at least with a minimum purchase. Data from SellersCommerce shows that 80% of customers are willing to meet free shipping requirements. Tiktokers might call it “girl math” when they’re willing to pay an extra $20 to save $10 in shipping, but successful e-commerce companies call it smart business to reduce cart abandonment. It’s also smart to be specific about shipping times. Rather than noting that orders arrive “within three business days,” provide an exact date or, at least, a range of dates. Amazon excels at this, noting below every product (even before you reach the checkout stage) exactly when the product will arrive, based on when it’s purchased. ### Allow a “Guest Checkout” Option Taking advantage of impulse purchases requires reducing opportunities for shoppers to really think about what they’re buying. Every extra step they must take is another chance to ask themselves, “Do I really need this?” Since almost a fifth of shoppers abandon their cart if they have to create an account, why not remove this roadblock and allow guest checkout? The fewer steps customers have to take to make a purchase, the more likely they are to follow through. ### Offer Multiple Payment Choices Offering many different payment options addresses a variety of concerns. If a shopper is hesitant to enter their credit or debit card information, perhaps they’d be more comfortable using PayPal or a secure digital wallet. Alternatively, offering buy now, pay later programs to spread out costs can reduce the number of cart abandonments due to the shopper’s credit card being declined, which occurs in 8% of cart abandonment cases. Another 10% of cart abandonments happen due to the lack of payment methods available, so make sure to set up your site so you can accept all major credit cards, PayPal, various digital wallets, and at least one BNPL service. ## Key Takeaways E-commerce sellers lose an average of 70% of sales because of cart abandonment, representing $18 billion in revenue. Some key reasons for cart abandonment include a lack of upfront pricing, slow or expensive shipping, and a complicated checkout process. While there are tactics to recapture those abandoned carts, it’s better to avoid them in the first place. ## Frequently Asked Questions (FAQs) ### How do you deal with abandoned carts? Smart online retailers deal with abandoned carts by reducing their prevalence with upfront pricing and a seamless checkout process. After a customer has abandoned a cart, retailers can recapture the sale through SMS or email marketing. ### What is abandoned cart recovery? Abandoned cart recovery is a tactic e-commerce sellers use to entice a shopper who left an item in their cart to return and purchase the product. ### How do you write abandoned cart emails? Abandoned cart emails can help you recover carts after a consumer has decided not to make a purchase. The best abandoned cart emails start with a provocative subject line, like, “Did you forget something?” or “ is selling out quick.” An abandoned cart email may also offer a discount or free shipping to get the reader’s attention. The body of the email should address the prospective buyer’s pain points and make it easy for them to click through and make the purchase. --- ### Creating Storytelling Performance Ads That Filter Your Audience URL: https://www.taboola.com/marketing-hub/filter-high-interest-users-with-creatives/ Last Modified: 2026-04-12 07:48:55 To unpack what truly drives return on investment with performance advertising on the open web, we spoke with Nadim Batista-Kuttab, CEO of Xevio, whose team runs high-volume campaigns across multiple verticals and generative experience optimizations (geos). In our conversation, Nadim broke down his creative philosophy for advertising, what makes an ad “clicky without being clickbait,” and how advertisers can use storytelling to grab high-intent users even before the click. ## The Creative Philosophy: A Stop Sign and Reason To Click Advertising on the open web thrives because users have a different user mindset than they do on social media. On social platforms, users scroll for entertainment and personal updates, but on the open web, they’re actively consuming information, from news articles and reviews to educational content. That shift in intent changes everything about how your ads need to be built. “You're reaching people at a time when they're browsing to consume content,” Nadim says. “The native feeds around articles are incredibly powerful at bringing people in to consume your content.” For this reason, the open web can deliver exceptionally high post-click engagement: Xevio sees an average time on page of more than two minutes on quality content pieces — but only if the ad is built correctly. “The ad is really just a stop sign to stop people from scrolling in the feed,” Nadim says. “The headline is the reason they click.” The creative’s job is simple: ### Image: The Stop Sign The visual must be captivating or intriguing enough to interrupt the user’s natural scanning behavior. Humans can process an image in milliseconds, so your creative either stops their scroll or you’ve lost them instantly. An image should: - Stand out visually without looking like a traditional ad. - Spark curiosity or resonate emotionally. - Align with the theme, audience or problem without giving away the “punchline.” ### Headline: The Reason To Click First and foremost, a headline should immediately tell a user why this content is for them. This is where the filter begins. A headline works when it: - Identifies the audience. - Connects to that audience’s problem, ignites their curiosity, and/or helps them reach a desired outcome. - Gives a clear reason to click and read more. Unlike paid social, performance creative is not the place to explain your offer. The ad exists solely to attract the right reader to your brand, product, or content. “You don't have a 60-second video like you do on other channels like Meta or Instagram, where you can really explain and sell,” Nadim says. “That'll be done later, in your content piece, after the click.” When you satisfy both image and headline expectations, you earn high-intent clicks from users who stay longer, scroll further, and convert at a higher rate. ## Storytelling: Creating High-CTR Ads That Filter Your Audience One of the biggest misconceptions about display ads is that they need to “sell harder” to win the click. In reality, this creative should repel the wrong audience just as intentionally as it attracts the right one. The secret to a high-converting ad is not clear sales language, but effective audience filtering that will lead to an increased click-through rate (CTR). You want to attract the person who needs your solution while repelling everyone else. "You need to find a middle ground,” Nadim urges. “Oftentimes, it's storytelling: Why are people buying your product? You start with that. What is the problem that you're trying to solve? Then, how do you get people to click on an ad, while filtering them for potentially having the problem that you're trying to solve?" The more precisely you filter before the click, the lower your blended cost per acquisition (CPA) will be. ### The Filtering Formula Effective ads don’t try to appeal to everyone — they strategically filter out users who aren’t a fit so that only the highest-intent consumers click through. Nadim explains that the most effective headlines clearly define the intended audience and why that audience should care. This method ensures you’re paying only for those who are already looking for your product or service, making them far more likely to engage and convert. Here’s how: - Define the audience: Call out your target audience directly to ensure the ad only appeals to users who fit your ideal customer profile. For example: women over 45, or men with knee pain. - Present a problem: Immediately introduce a relatable frustration, pain point, or unmet need. Users click to solve, understand, or fix something, not to be sold to. The hook: Your Audience + Your Problem = Your Desired Audience. Say your headline is, “Seniors: Do This Simple Stretch to Ease Lower Back Pain.” Anyone who isn’t a senior or doesn’t have back pain will automatically scroll past. That’s the goal. This formula: - Increases CTR without aggressive tactics. - Pre-qualifies users on autopilot. - Improves time on page. - Boosts post-click engagement and conversion rates. “You'll see a lot of headlines where they're not particularly clear on what they're selling,” Nadim says.“Those are very generic headlines that work because they define the audience and the problem, which essentially filters out the people that aren't interested in that topic.” This is how ads become both “clicky” and compliant. https://youtu.be/ncvZ766nWt4 ## Avoiding the Traps: Aggression vs. Generics When it comes to performance ads, you’ll need to avoid being either too aggressive, or too generic. Stray too far in either direction and your campaign collapses, either through compliance rejection or poor CTR. ### The Aggressive Trap It can be tempting to oversell your offerings, but that’s one quick way to alienate consumers and risk non-compliance. “You don't want to say, ‘Click here and cure your disease,’” Nadim cautions. “That's a terrible approach. Compliance will butcher your ads and take them offline, and you get in trouble for it.” To stay on the safe side, avoid: - Promises of outcomes you can’t prove. - Unsubstantiated medical claims. - Fear-based hooks. - “Click here”-style commands. Performance advertisers speak to consumers who want trusted, educational content, rather than marketing hype. ### The Generic Trap On the other end of the spectrum, some advertisers default to bland, brand-centric headlines that don’t compel clicks. The lack of value means they don’t convert. “You can't just be like, ‘Buy my product,’ because people won't click,” Nadim says. “There's too much going on, on the open web for them to really be attracted to a generic ad.” Generic ads fail because: - They don’t speak to the user’s problem. - They don’t identify the intended target user. - They focus on selling rather than informing. Open web users are in “content consumption” mode: They want to be educated, enlightened, or entertained. The best ads use storytelling to engage users and lead them to a landing page that satisfies their curiosity. “Remember, everybody on the open web is looking for content to consume and is looking to be educated,” Nadim says. “So, use that to generate high-CTR creatives that aren't aggressive.” ## Key Takeaways Performance advertising rewards marketers who understand that the ad is a hook, not a pitch. Lead with an image that stops the scroll, pair it with a headline that filters your target audience, and focus on educating rather than selling. When you avoid tipping too far into being either too aggressive or too generic, your ad earns attention without triggering compliance issues, or being ignored. In the end, the brands scaling their display advertising profitably aren’t necessarily the ones with the prettiest creatives: They’re the ones sending high-intent users into their sales funnel. As Nadim notes, the right creative attracts users who are already looking for a solution like yours, making every click more likely to convert. Build your ads to filter before the click, and you’ll scale faster, spend smarter, and turn the open web into one of your most profitable channels. --- ### Carousel ads vs single-image ads: Which Drives Better Conversion Rate and Urgency? URL: https://www.taboola.com/marketing-hub/single-vs-multiple-cta/ Last Modified: 2026-01-15 10:23:34 As an advertising copywriter, so much of my creative energy goes into headlines, pre-headers, subject lines, and body copy. After all, that’s what people are going to see first — it’s what draws them in, gives them the quick sales pitch, and tells them how this product or service could be life-changing. That said, the final piece is what does a lot of the heavy lifting, and shouldn’t be overlooked: the call to action (CTA). The CTA might just be the single most important element of any ad — it’s the moment of truth where engagement becomes conversion. Even so, marketers often struggle with a fundamental strategic question: Is it better to focus a user on a single, clear button, or offer multiple routes to explore? CTAs may seem simple, but the answer isn't always clear — it depends entirely on your campaign goal, the user's intent, and the ad format you choose. ## The Case for a Single, Focused CTA: Clarity and Conversion A singular call to action is the traditional, go-to best practice in certain contexts, capitalizing on those core psychological principles to maximize conversion likelihood. This strategy works by minimizing distraction and focusing the user's cognitive energy on one clear desired next step. ### Minimizing Decision Fatigue A single CTA reduces options, eliminating user confusion about the desired next step and often leading to higher conversion rates on dedicated pages. Users can suffer from "decision fatigue" when presented with too many options, but by offering just one path, you remove a barrier to conversion. In fact, reducing a page to one CTA can boost conversions by up to 266%. ### Achieving Specific Campaign Goals A single CTA offers more control and is ideal for highly focused goals like sign-ups, lead capture, or direct sales on a landing page where the user has already demonstrated high intent. Keep it clear: When the user knows exactly what they want (e.g., download a specific whitepaper), the single CTA confirms they’re in the right place and speeds up the action. ### High Performance in Specific Channels While your creative instinct may be to dazzle the reader and make your campaign stand out, keep in mind that certain channels thrive on simplicity. Being linear and goal-oriented, emails perform significantly better with a single CTA, driving 371% more clicks and up to 1,617% more sales compared to emails with multiple CTAs. ## The Case for Multiple CTAs: Depth, Exploration, and Storytelling Multiple CTAs can still work, especially within a visually engaging format, to keep people from leaving your page. They can excel at capturing attention, providing a deeper user experience, and facilitating mid-funnel exploration. ### Reducing Bounce Rate (Exploration) Multiple, related CTAs on a longer page (like a homepage or a sequential ad format) give users options, lowering the chances of them leaving the site entirely if they aren't ready for the primary conversion. If a user isn't ready to "Buy Now," they might be willing to "Browse Styles" or "Read Reviews," keeping them within your marketing ecosystem. ### Showcasing Product Variety and Complexity For brands with numerous product lines or complex services, multiple CTAs allow marketers to feature a range of offerings in one ad unit, increasing the chance of a user finding something relevant. A user looking for boots might see options for "Hiking," "Outdoor Gear," and "Lifestyle," with a CTA directing them to the relevant category page. ### Sequential Storytelling Leveraging a sequence to build tension or walk a user from a problem (Card 1) to a solution (Card 3) to the final action (Card 5) guides them to a commitment. This format naturally eases the user toward the final conversion by educating and warming them up across the ad experience. ## Realize Carousel Ads: Leveraging Multiple CTAs for Optimal Performance The Realize Carousel Ad format on the Taboola network acts as a strategic middle ground, intelligently using multiple ad cards to provide choice and storytelling depth, while guiding the user toward a specific conversion goal. This format takes the strengths of both single and multiple CTA strategies and combines them for superior performance. ### Interactive Design for Higher Conversions Carousel Ads aren’t just more interactive — they’re proven to be extremely effective, frequently outperforming single-image ads in conversion rate (CR) and return on ad spend (ROAS). The swipe mechanism encourages prolonged engagement, allowing your message more time to resonate. ### Card-Level Customization Each card within the Realize Carousel can be treated as a unique mini-ad, featuring its own headline, description, and link to a unique landing page. This allows you to match the CTA/landing page to the specific product or benefit shown on that card. You can present four distinct products with four specific "Shop Now" CTAs, all within a single ad unit. ### Multiple Link Destinations The format allows you to drive traffic to multiple product pages, landing pages, or offers from a single ad unit. This solves the "single vs. multiple" debate by accommodating both: A primary goal can be placed on the first card, while secondary goals or supporting content can live on subsequent cards. ### Per-Card Optimization You can optimize the sequencing of cards to put the highest-performing visuals and offers first, to maximize interaction. Realize’s artificial intelligence (AI) uses data from user interactions to learn which card order leads to the best result — e.g., if it notices that presenting the "limited stock" message before the "final discount" leads to more purchases, it automatically serves that optimized sequence to future users — often helping marketers improve their funnel flow faster than manual testing. ### Generating Urgency via Narrative Flow Use the multiple cards to showcase a limited-time offer across the slides, or build narrative tension (e.g., a "Before & After" sequence) that culminates in a strong, urgent CTA on the final card. For instance, Card 1 could show the problem, Card 2 could show the solution, and Card 3 could hit the user with a time-sensitive CTA like "Claim Your Discount Now — Offer Ends Soon!" ## Key Takeaways A single CTA is best for high-intent users, landing pages, and channels like email, where the goal is direct, immediate conversion with minimal friction. Multiple CTAs, meanwhile, are best for discovery, showcasing variety and complex products where the user is in the mid-funnel exploration stage. Realize Carousel Ads offer a hybrid approach, using sequential cards for storytelling and product variety while still achieving high CR and ROAS due to their interactive nature and card-level customization. ## Frequently Asked Questions (FAQs) ### How many cards should I use in a carousel ad? The best strategy is to keep your story concise — an engaging swipe experience that generally uses three to five cards. For campaigns with a single focus, like pushing a direct sale, you can still build a sequence in just three cards: show the problem on the first, reveal your solution on the second, and end with a strong CTA on the third. If your brand offers a variety of products or services, maximizing your ad to five cards is a go-to way to showcase them all. This lets you use the format's flexibility to feature different items (like five unique features or products) with each card linking to a specific landing page, making it much easier for the user to discover exactly what suits their needs. With unique URLs and flexible creative options including images, GIFs, and video assets, Realize makes Carousel Ads easy, ensuring optimized rendering and performance tracking across devices. ### Should every card have a different CTA button? Not necessarily. You should only use different CTA buttons if they lead to different, relevant landing pages (e.g., linking to five different product categories). If all cards promote the same offer, use the same strong CTA on each card to reinforce the action. Realize allows you to utilize this multiple link destination flexibility for advanced testing and product promotion. ### How does a carousel ad lead to urgency? A carousel ad generates urgency by using sequential narrative flow. You can dedicate the first card to introducing a limited-time sale and use the subsequent cards to highlight the best deals, culminating in the final card, which carries the urgent CTA button ("Act now/Offer ends soon,” etc.). This guided sequence is more compelling than simply placing an urgent message on a single static image. Realize handles the complex task of finding the perfect sequence and the perfect audience, creating the right amount of drive to get users to click CTAs, and allowing you to focus on making the best content. You’ll find that you get better results when the AI is constantly testing and improving the narrative's flow, driving that sense of urgency effectively. The AI also helps with targeting, ensuring the entire urgent story is only shown to people who are predicted to be high-intent, meaning they’re already close to buying. This way, you aren't wasting your budget showing a "Final Hours" ad to someone who has never heard of your brand. ### Can I use my existing creatives in a carousel ad? Yes! The beauty of the carousel format, particularly within the Realize platform, is that you can repurpose your high-performing static images, videos, or headlines across the different card slots. This allows you to quickly A/B test existing assets in a new, high-engagement format without starting from scratch. --- ### Lookalike Audiences: How to Expand Your Reach URL: https://www.taboola.com/marketing-hub/lookalike-audiences/ Last Modified: 2025-11-13 13:10:02 Digital marketers have many ways to find their ideal customers. Cookies, for instance, can help marketers track user behavior to find people searching for their products or services. Contextual targeting places ads and content based on what the user is doing, reading, or viewing at that moment. Lookalike audiences, meanwhile, can help marketers expand their audience by finding people in a similar demographic as their current, loyal customers. ## What Is a Lookalike Audience? A lookalike audience is a group of people whose behaviors and traits match those who have previously purchased from you, or otherwise engaged with your brand. Lookalike targeting operates under the assumption that people in a similar demographic, who exhibit similar online behaviors, will be interested in the same content and brands. A lookalike audience can be created from an email list for your brand or website visitors, for example. ## Why Are Lookalike Audiences Important for Marketers? Lookalike audiences offer some advantages for marketers who are looking to expand their audience: ### Find Your Target Audience Quickly Lookalike audiences help you locate your target audience quickly and double down once you dial in on the most effective creative for that audience. You can also optimize ad spend by minimizing the cost to find your audience. ### Eliminate Guesswork Lookalike audiences aren’t a sure thing, but they’re one reliable method to find people likely to respond to your ads. By targeting people with similar behaviors and attributes as your current customer base, you increase the likelihood of conversions. ### Expand Your Audience Once you’ve exhausted your warm market, your sales funnel needs fresh leads. Lookalike audiences help you find them with a lower cost of acquisition than many other types of targeting. ## How Do Lookalike Audiences Work on Social Media? ### TikTok When you set up an ad campaign on TikTok, you first choose one of your custom audiences, which consists of users who have already engaged with your brand. Then, you create a lookalike audience based on that custom audience. You can choose from a narrow, balanced, or broad audience, depending on the reach you want to achieve. A narrow audience will be the closest match to existing customers or your source audience. Make sure to “exclude custom audience” from the campaign if you don’t want to retarget those users. ### Meta Creating a lookalike audience on Meta’s Facebook or Instagram is similar to TikTok. First, upload an audience, which may include existing clients or website visitors. Then, choose where you want to find the lookalike audience, and use the slider to select the size. The larger the audience, the more people you will reach, but your overall campaign cost may also rise. ## How Do Lookalike Audiences Differ From Custom Audiences? Lookalike audiences are all about expanding your reach. Custom audiences are made up of people who have already engaged with your brand on a specific platform; lookalike audiences are people who share characteristics with your custom audience, but it should exclude them. ## When to Use Lookalike Audiences vs. Interest-based Targeting Lookalike audiences can be effective when you already have a large group of customers. Meta, e.g., recommends building your lookalike audience from a source audience of 1,000 to 5,000 people, although you can start with an audience as small as 100 people. If you’re a start-up with no data on your customers, you might be better off using interest-based targeting to find people your brand resonates with, until you have a wider audience to build from. ## Benefits of Using Lookalike Audiences ### Reach High-Quality Prospects Because you’re reaching people with similar characteristics to loyal customers, lookalike audience campaigns allow you to get better results faster, and with a smaller ad spend. ### Reduce Cost-per-Acquisition (CPA) Click-through and conversion rates rise when you show ads to a highly qualified audience. ### Scale at Will It’s not always easy finding new leads. Lookalike audiences can optimize your funnel, taking people from “prospect” to “loyal customer” faster, because you’re already targeting people who are more likely to convert. ## How to Create a Lookalike Audience ### From Website Visitors To create a lookalike audience from website visitors, you can install tracking pixels on your website. Then, you can build a lookalike audience based on specific actions those visitors took, whether that’s opting in to an email list or making a purchase. ### From an Email List You can upload data directly from your CRM to create a lookalike audience. ### From People Who Watched Your Video Depending on the platform you’re using for advertising, you can choose a custom audience based on engagement. Choose “video” and then choose the length of time people watched for. You can then use that custom audience to create a lookalike audience on the platform. ## Where to Source Your Audience You can source custom audiences for lookalike audiences from a variety of places: - Your CRM. - Social media interactions. - Articles they’ve read. - Their email. - Their mobile ID. - Pixels that track what they did on your website. ## Location and Size Targeting The success of your lookalike campaign depends not just on the accuracy of the lookalike matching, but the location and size of your audience. Location targeting matters more for local businesses, like restaurants or retail locations, but national advertisers are typically better off aiming for a wider reach and narrowing the audience size instead. ### Location Targeting Again, national brands may want to start with a broader reach. However, you might want to narrow campaigns based on geography under certain circumstances. For instance, a national clothing retailer wouldn’t advertise winter coats to customers in Florida. That would be an example of a national company narrowing a campaign to a specific region. Even if you have a national brand, you can narrow your targeting by location to reduce ad spend and find closer matches at higher percentages. ### Size Targeting Size targeting is crucial and enables advertisers to control costs while attracting new leads. You can adjust the size of your campaign by percentages: A campaign might be shown to the top 1%, 2%, or 5% of platform users who match your criteria for a lookalike audience. As you increase the percentage, the users may not be as close of a match, which could reduce conversions. A 1% audience is likely to be a very close match to your ideal customers. As you increase the percentage to expand reach, your return on investment (ROI) might drop. However, if you’re in a niche market, you’ll need a larger audience to reach enough people. Experiment with different-sized audiences to find enough conversions at an acceptable customer acquisition cost (CAC). ## How to Test and Optimize Lookalike Audiences for the Best Performance Results A/B testing is one of the best ways to optimize lookalike audiences. When you’re A/B testing, remember to change only one parameter at a time for the best results. ### Change the Audience Size Changing the audience size can give you a broad or narrow reach. An audience size of just 1% will get you a closer match — the more you broaden your reach, the less “alike” your lookalike audience will be to your source audience. ### Experiment With Different Source Audiences Where you get your source audience from also matters — you might find better results compiling an audience from your CRM or tracking pixels on your website, based on content they’ve viewed. ### Use A/B Testing for Creative Once you’ve homed in on an effective lookalike audience, switch up an element of your creative to refine your results even further. This might involve ad copy, graphics, or even the advertising format, such as a carousel ad, a short video, or a static ad. ### Be Aware of Overlap If you’re sourcing your lookalikes from different audiences, there might be some overlap. This can drive up campaign costs and also lead to ad fatigue if the same people keep seeing the same ad. ### Be Patient Don’t rush to switch up your creative, reach, or your source audience. Give a new campaign time before you gauge its effectiveness. ## How to Scale Campaigns Using Lookalike Audiences ### Increase Your Lookalike Percentage Having success with a narrow lookalike audience of 1% to 2%? It’s time to scale up and expand your reach. You should use A/B testing to find the best return on ad spend (ROAS) and find out when your reach is so broad that the lookalike set isn’t a good match. ### Update the Seed Audience as It Grows The right seed, or source, audience can make or break a lookalike advertising campaign. To successfully scale and find new leads, refresh your seed audience every few months as your loyal customer base grows. As successful lookalike audiences become customers in your CRM or visitors to your website, update the seed audience to include others who look like them. Your lookalike audience should evolve and grow as your customer base grows, to maintain accurate targeting. Be aware of overlap, however, especially if you’re seeding your lookalike audience from different sources, including your CRM, videos, emails, or your website. ### Increase Your Budget Increase your budget slowly and gradually, making sure to track key performance indicators (KPIs) so you can slow down or shift gears if you reach a point of diminishing returns. Once you’ve dialed in a successful audience, increase the budget by 10% to 30% every few days. Those with smaller budgets can increase by 30% to see a change, while those with a larger budget might see a lift with a 10% increase. ### Add More Regions It’s smart to begin your lookalike campaign in a single state or region. Create additional campaigns using the same seed audience and percentage reach, but targeting different states or regions. Alternatively, you can reduce percentages and broaden geographic reach. However, treating each region separately allows you to tailor ad creative and even language for a specific geographic audience. ## More Lookalike audiences Tactics Finding the right audience using lookalike tactics isn’t always easy: Marketers can approach and scale their lookalike campaign in many ways, and the data you can use to create your lookalike audience is virtually endless. The key is finding what works best for your brand. ### Value-based vs. Standard When algorithms use your source audience to choose the standard lookalike audience, users in the first percentile have more in common with your source audience than those in the tenth percentile. Other factors, such as cost per acquisition, how recently the customer made a purchase, the customer’s average order value, or customer lifetime value (LTV), don’t come into play. However, value-based lookalike audiences consider these other factors. While decreasing the lookalike customer pool, using a value-based lookalike audience can increase your ROAS. By considering factors such as average order value, most-recent order, or customer lifetime value, you can find the lookalikes most likely to convert. ### Tap Into CRM Data Performance marketing platforms allow marketers to upload real-time, first-party data from their CRM, eliminating a need for cookies and capturing the most reliable data related to customer interactions. ### Use Offline Conversions to Build Your Source Audience You can use APIs to track real-world actions — like in-store purchases and phone orders influenced by online or SMS ads — to expand your source audience. As a result, you can find more accurate lookalikes that represent a wider cross-section of your customer base. ## Key Takeaways Lookalike audiences share characteristics with audiences who have previously purchased from or otherwise engaged with your brand. Using lookalike audiences is a cost-effective way to increase your customer base and build brand visibility, but the success of a lookalike campaign depends on effective targeting. ## Frequently Asked questions (FAQs) ### What does a 1% lookalike audience mean? A 1% lookalike audience encompasses users who are the most similar to your existing, or source, audience, based on characteristics that may include demographics, interests, or behaviors. ### What are Facebook lookalike audience terms? Facebook lookalike audience terms relate to the ways you will identify and find your lookalike users. You can create a lookalike audience from your source audience based on demographics, behaviors, or interests. ### How do lookalike audiences work on LinkedIn vs Facebook? You can create a lookalike audience for a Facebook ad campaign by uploading a custom audience from your own data, or choosing leads through the Meta Business Suite. LinkedIn, meanwhile, has shifted from traditional lookalike audiences to “predictive audiences” based on the marketer’s data from lead gen forms or contact lists. ### Can I create a lookalike audience in Google Ads (Customer Match)? Yes, you can create a lookalike segment in a new Demand Gen campaign in Google Ads. Choose a seed audience based on your existing CRM or email, website, or app activity, or YouTube content engagement. Then, determine whether you want a narrow, balanced, or broad reach to find users who are a similar match. ### Can you help me set up an A/B test for 1% vs 5% lookalikes? You can set up an A/B test for 1% vs. 5% lookalikes by initiating two campaigns, both from the same source list and using the same creative, set at different reach percentages. ### How and when should you layer lookalikes with interest targeting? Layering lookalike audiences with interest targeting can help marketers scale a successful campaign. However, be wary of narrowing the audience so much through interest targeting that you can’t reach enough prospects effectively. ### Should I use purchases or add-to-carts as my source audience? Source audiences for lookalike audiences can vary. If you’re running a value-based campaign, using analytics like “purchases” can be an effective way to build a lookalike audience. If you’re just starting out or don’t have a large customer base, using customers who added an item to their cart but haven’t converted to a sale yet, might make sense for a broader reach. --- ### Eight Best Performance Marketing Channels URL: https://www.taboola.com/marketing-hub/best-performance-channels/ Last Modified: 2026-06-16 11:19:09 Performance marketing seeks to reach audiences in the action segments of the sales funnel: consideration and purchase. Have your recent marketing campaigns delivered the results you expected? Are they leading to sales? If not, you could be trying to reach your audience through the wrong channels. Determining where your audience looks for solutions to the problem your brand solves is the first step. Marketers have more choices than ever for performance marketing campaigns. Let’s explore some of the top channels in 2026. What’s changed in our 2026 update: - Updated and current information added to all channels. - 3 new channels added: - Email marketing - Generative AI - Paid Social - Pros and cons of performance marketing added. - New FAQs added. ## 8 Marketing Channels to Achieve Campaign Success For years, email marketing, paid social, and paid search (namely, Google AdWords) stood out as the top performance marketing channels. But, that’s shifting as affiliate marketing, native advertising, CTV, and Generative AI reach audiences in new ways with compelling creative that drives sales. ### Pros and Cons of Performance Marketing Whatever channel you choose, performance marketing often provides a straight and direct route to profitability and growth. In general, there are many advantages and a few caveats to this method of lead generation and sales. Pros - Cost-efficiency. - Scalability. - High ROI. “Since payment is based on specific actions, businesses can better manage their marketing budget and ensure they are getting a good return on investment,” per Quora. Cons - Less effective for brand awareness. - Requires real-time optimization for best results. “The intent of the campaign is to drive consumer action, as opposed to raise awareness,” per Quora. With these benefits and drawbacks in mind, some performance marketing channels deliver better results than others. Let’s explore some of the best. Channel Pros Cons 1. Affiliate Marketing - Only pay for results. - Wide reach. - Scalable results. - Trusted voices promote your brand. - Limited control. - Variable quality of content. - Challenging to track ROI. - Difficult to reach the “messy middle”. 2. Display Ads - Personalization. - Scalable reach. - Advanced targeting. - Data-driven insights. - Tools for automation and retargeting. - Ad blindness. - Lower conversion rates compared to search or native. 3. Email Marketing - Proven, time-tested channel. - High conversion potential. - Easy retargeting. - Easy to track. - Limited reach. - Heavy creative lift. 4. Generative AI (ChatGPT, Claude, Perplexity) - Automatic trust. - Personalized responses. - Highly targeted by context. - No ad costs involved. - Short funnel. - No marketer control over content. - No demographic targeting. - Difficult and time-consuming to rank in Gen AI search - 5. Google - Extensive reach. - Clear metrics. - Reaches the bottom-of-the-funnel audiences. - Diminishing returns. - Highly competitive. - Constant monitoring required. 6. Mobile Advertising - Massive reach. - High conversion potential (BOFU reach). - Difficulty reaching the “messy middle”. - Ad format limits. - Ad fatigue. 7. Native Advertising - Reaches people in the decision stage. - Blends with content. - Builds trust. - High ROAS. - AI helps match campaigns to audiences. - Highly personalized based on context. - Disclosure required. - Requires platform and creative alignment. 8. Paid social - Large audience. - Advanced retargeting. - Low entry cost. - Reach customers where they are. - Highly targeted by demographic. - Diminishing returns. - Costs can add up quickly as you scale. - High cost per lead - Ad fatigue. - Strong creative demand (i.e. “scroll-stopping” ads). ### 1. Affiliate Marketing As consumers experience information overload, more and more people make buying decisions based on suggestions from people they trust. Not quite “family and friends,” affiliate marketers stand out as knowledgeable, trusted sources. Roughly 80% of brands today use affiliate marketing, and the industry is expected to exceed $17 billion in revenue in 2026. Affiliate marketing can take place across any of the performance marketing platforms listed above. In affiliate marketing, advertisers don’t pay unless the placements convert to measurable goals — typically, sales or downloads. Affiliate marketing shines in the “messy middle,” that stage where brands want to nudge consumers to take action. “Retail media struggles in mid-funnel because there isn’t a lot there in a lot of ways in terms of moving somebody along the purchase funnel,” Mike Mallazzo, writer of the Zero Clicks newsletter, opined in this blog post. ### 2. Display Ads Display ads can appear on any website, typically in a banner format or as a vertical ad down the side of the page. Display ads may include video, graphics, text, motion, and, increasingly, interactive formats to attract and engage viewers. Online display advertising is estimated at $242.36 billion, according to Mordor Intelligence, up from $212.10 billion in 2025. Thanks to advancements in AI-driven targeting, real-time bidding, and cross-platform integration, display advertising offers significant advantages for performance marketing at scale. “Marketing teams have long collected vast amounts of data on customer behaviors, content engagement, and campaign performance, but have struggled to fully utilize it,” Holly Fee, VP of Marketing at Infragistics, told Forbes, adding that she expects to see more companies invest in creating data-driven cultures and “preparing their data for AI tools.” Modern display ads employ machine learning to deliver hyper-personalized content that’s based on user behavior, rather than just traditional demographics, which hugely improves both click-through and conversion rates. With the rise of cookieless tracking technologies and privacy-first ad solutions, marketers can also feel confident about reaching their desired audience without falling foul of data compliance regulations. ### 3. Email Marketing Email marketing is far from dead. Three quarters of marketers said they plan to maintain or increase their email marketing spend in 2026, according to HubSpot data. Meanwhile, one in five marketers say it remains their top ROI-driving channel. As effective as it is, one drawback to email marketing is that you need to use other channels to build your email list. Unless you’re retargeting visitors to your website who’ve already opted in, you’ll need to attract customers via social, search, native advertising, or other tactics. Email marketing also relies on a hefty investment in quality content, as well as the software needed to build and maintain your list and create easily scannable, enticing emails. Increasingly, marketers are turning to AI to level up their campaigns with enhanced personalization, optimization, and segmenting. “The value here spans across all campaigns, audiences, and industries, making it a true game-changer for email marketing efficiency,” Bernard May, of National Positions, told Forbes. ### 4. Generative AI AI is increasingly playing a role in most performance marketing channels. LLM and machine learning help marketers create ads and find their audiences at scale, but generative AI stands out as a performance marketing channel of its own. That’s because people are turning to their generative AI chatbot of choice — typically ChatGPT, who held the greatest market share at more than 60% in February 2026 — to find brand recommendations they can trust. Along with ChatGPT, people also ask Google Gemini, Claude, Grok, and others for tips on travel planning, major purchases, recipes, and more. Results that turn up in Google’s AI Overviews also fall into this category. Garnering AI mentions can dramatically shorten the lead funnel, pushing people from the research stage to buying in the era of “no-click” search. However, optimizing for AEO (Answer Engine Optimization) or GEO (Generative Engine Optimization) takes time, just like traditional SEO. Costs add up when you consider the vast amounts of content creation required. No one has the “secret formula” for getting mentioned by AI yet, so it’s still a bit of a gamble. ### 5. Google Google’s ad revenue continues to grow, driven by the search platform and YouTube — in Q2 2025, parent company Alphabet brought in $54.2 billion through “Google Search and Other.” As organic search traffic declined across major sectors between 2025 and 2026, text ads captured between 7 to 13% of what was previously organic traffic, Alm Corp reported. Acquisitions come from web user searches that may bring them to AI-generated recommendations, company websites, or paid ads. Google still holds roughly 90% of the Web’s search traffic. People use the platform every day, seeking solutions to a problem, specific products, or general information, such as, “Where’s the best pizza in New York?” Brands can rank for the answers to these searches through search engine optimization, answer engine optimization (AEO, or AI summaries), or through ads placed on the Google Adwords network. Paid search is often called search engine marketing (SEM). Ads can appear as text at the top of the search engine results page (SERP), or as text or display ads on various websites that use Adwords as a revenue stream. ### 6. Mobile Advertising Mobile advertising can span any of the platforms above, with roughly 98% of global users accessing the web through mobile phones in 2024. Mobile advertising encompasses SMS ads, in-app ads, ads on social media platforms, and paid search through mobile devices. Mobile is an effective way to reach bottom-of-the-funnel users, since 76% of adults in the U.S. have made purchases through their smartphone. ### 7. Native Advertising Native advertising looks like editorial content on a website, providing valuable and relevant information to readers. While native advertising will have a disclaimer that it’s a paid ad, it tends to blend in with other content on the platform. Ad blockers eliminate visibility of display ads for many users, but native advertising often slips through to reach browsers. If it delivers the right information to a well-targeted audience, users will even welcome it. Native advertising is expected to grow 13.1% in 2026 to reach nearly $148 billion in revenue. Native advertising formats vary based on the platform where the ad appears, but the common thread is that the best native advertising offers readers or viewers the content they need to make an informed decision. Powerful AI algorithms use contextual data to place native ads where they make sense, putting content in front of audiences ready to make a buying decision. For instance, someone reading an article about the benefits of light trucks for family road trips may see an article titled “Top 10 Best Light Trucks for Families” as part of a native advertising campaign. ### 8. Paid social Meta leads paid social in 2026, with a combined 3.58 billion users across Facebook and Instagram. With low start-up costs of $5 per day for ads, many performance marketers view Facebook as a safe bet, but that’s not always the case. Facebook’s cost per click averages 0.54, but the cost per lead (CPL), a much more relevant analytic, ranges from pocket change ($3.16 for restaurants and food) up to double figures as the average for most service industries — if you’re in the home improvement industry, the CPL is $41.26 on average, and dentists can expect to pay more than $71 per lead. Ad fatigue, targeting the wrong audience, and creative that doesn’t “stop the scroll” can all be factors in a failed campaign. “Facebook is still a good channel for building an audience, but if you’re a small business looking for fast growth and results, $5 a day on Facebook will get you nowhere,” Rima Mattok, demand generation director at Taboola, previously stated when discussing Facebook Ads. Other social channels create similar challenges. Whichever platform you choose, a successful social media campaign requires identifying where your target audience spends their time online, determining which ad formats will cut through the noise, and creating content that resonates. ## How to Determine Which Performance Marketing Channels Work Best As with any campaign, finding the performance marketing channel that works best for your audience requires some advance research and trial and error. Performance marketing allows you to test platforms affordably, at least, and once you dial in on what works, you’ll want to double down on those efforts. Follow these steps to start your quest for choosing the right channels: - Determine your KPIs: Before you enter any campaign, determine your key performance indicators. In most cases, performance marketing KPIs will be purchases or clicks, and how much it costs to achieve those conversions. Determine what you’re measuring before you begin so you’ll know when a campaign is successful. - Find your target audience: Next, do diligent research to determine where you’re most likely to find your target audience. - Start testing platforms and creative: At some point, you have to dive in with your best creative on the platforms you think will convert. - Double down: Every good performance marketing campaign gets to a point where something is working, which is the point you should increase the budget until you either saturate the market, or cease to see results. ## Key Takeaways Performance marketing campaigns can reach prospects through a wide range of channels, spanning from social media and search to display ads. Campaigns that focus on the action stages of the funnel (consideration and purchase) are likely to deliver results faster and at a lower cost. However, it’s crucial to choose the right channels to reach your target audience when they’re ready to buy. Platforms that use AI-powered real-time analytics can deliver results faster. ## Frequently Asked Questions (FAQs) ### Is performance marketing the same as SEO? Performance marketing is not the same as SEO (search engine optimization). Performance marketing relies on paid advertising to deliver messages to users in the action stage of the sales funnel, while SEO is an organic marketing strategy designed to help websites rank first on search engine results pages (SERPs). ### Is email a performance marketing channel? Email is a performance marketing channel because it allows marketers to deliver messages to their target audience and measure results based on clicks, downloads, or sales. However, it’s different from other performance marketing channels because it does not rely on a third-party platform like social media or a publishing network. ### What is a marketing channel strategy? A marketing channel strategy uses analytics to determine the most effective places to reach customers both online and offline. A solid marketing channel strategy should evaluate who you are trying to reach, where you’ll find them, and what resources to allocate to each channel. ### What are performance marketing channels? Performance marketing channels are the best places to reach your audience in the action stages of the sales funnel, when they are focused on consideration and action or purchase. ### Which performance marketing channels scale best? To scale your performance marketing, you need to choose the right channels. Performance marketing channels that scale the best include cross-platform display advertising and native advertising. These deliver a low customer acquisition cost and the ability to expand by doubling down on your most effective content. ### How do you track cross-channel performance of marketing campaigns? You’ll need a platform that tracks campaign performance across channels and devices, ideally without third-party cookies. Realize, e.g., uses first-party data to track important metrics like cost per action (CPA) and return on ad spend (RoAS), while visual dashboards make it easy to analyze performance and optimize targeting. --- ### Advertising Network URL: https://www.taboola.com/marketing-hub/ad-network/ Last Modified: 2025-11-13 12:47:05 With thousands of publishers and millions of potential consumers, it can feel overwhelming when trying to build an audience, sell a product or service, and scale your business. An advertising network enables you to make connections from a single centralized platform, like a bridge connecting advertisers and publishers. The right advertising network can help you save time and money, whether you’re on the supply or demand side. ## What Is an Advertising Network? Advertising networks connect advertisers with publishers whose websites or mobile apps have available ad space, drawing from a wide range of publishers’ websites, and aggregates all the available ad space inventory. These platforms can also provide tools to gather consumer data, enabling advertisers to target new audiences, create and edit campaigns, and offer a wealth of metrics and data beneficial to advertisers looking to expand their reach. ## Why Are Ad Networks Important? Ad networks can help determine the best websites and apps for a particular product or service, thereby increasing a brand’s reach through exposure, impressions, and sales. “As an advertiser, you do not need to contact each site yourself: You set your budget, choose who you want to reach and what kind of ad you want to run,” says Alex Smith, manager and co-owner of Render3DQuick in Toronto. “The network places your ads where they fit best. It saves time, cuts down on manual work, and gives your ads wider exposure. Think of it like this: You are a small furniture company launching a new line of office chairs. You want to reach interior designers, architects, and small business owners. You sign up with an ad network, set a daily budget of $50, and pick your audience. The network automatically shows your ad on architecture blogs, home design sites, and business tools apps where your audience already spends time.” ## Pros and Cons For Advertisers Pros Cons Provides you with a wide reach. Can be expensive. Lets you scale campaigns. May not let you control ad placements. Offers targeting tools (by country, platform, and operating system). Risk of fraud. ## Pros and Cons For Publishers Pros Cons Tremendous access to advertisers. May have to share any revenue with your ad network. Easy to set up and sell space. Potential quality control issues. Builds revenue through targeted users. May run into issues with compliance. ## Benefits of Ad Networks ### Provides Cost-Effective Access “For small companies, ad networks are essential because they level the playing field,” says Smith. “Without deep pockets for massive ad campaigns, small businesses need cost-effective ways to reach the right people. Because ad networks provide access to a wide pool of publishers, you can skip direct negotiations, saving you time and resources. On top of that, they offer precise targeting options, so you’re not wasting money on irrelevant audiences.” ### Allows You to Scale Your Campaigns You can scale your campaigns up or down according to your budget, marketing goals, and other targets on an ad network, and most advertising networks offer you access to editing tools to help you manage all your campaigns in one place. ### Easy Setup for Publishers An ad network makes it easy for publishers to set up codes, banners, and other ads without sacrificing real estate in the app or site. By partnering with an advertising network, publishers can benefit from CPV (clicks per view) on their site and generate revenue from the sales and potentially increased traffic. ## Considerations When Advertising on Ad Networks ### Cost Some advertising networks may be too expensive or offer more tools than a small business needs. It can pay off to look for competitive pricing and transparency in billing and fees. ### Lack of Control Using an advertising network, you risk losing control over placements, which means your ads may appear next to non-relevant content and could be subject to false metrics if bots or spam interact with the app or web page. ### Takes Time to Fine-Tune You may need to continuously fine-tune campaigns to get the full return on investment. If your network’s platform doesn’t have robust editing capabilities, it may take longer to implement changes during a campaign. ## What Are the Different Types of Ad Networks? ### Blind Ad Network These have wide exposure, but you can’t pick where your ads appear. While compatibility with sites and brands is offered, you have no control over the placement. This is beneficial for businesses that prioritize page views over placement. ### Vertical Ad Networks These networks allow you to advertise your products on apps and websites with a compatible audience. “For example, if you sell software for architects, a vertical network that includes design or construction sites would be a smart fit,” says Smith. ### Premium Ad Networks These give you access to reputable websites, like high-quality news media, lifestyle publications, high-traffic blogs, and other respected publications. They typically cost more, but in return, you get to advertise on a trusted site. ### Programmatic Ad Networks These networks utilize software and data to create ad placements in real-time, using a set budget and targeting a specific audience. They offer less control than other networks, but you may have a larger reach using this type. ## Ad Targeting Capabilities A network’s ad targeting capabilities allow advertisers to reach a specific target audience. They include, but are not limited to: - Demographics. - Age. - Income. - Online shopping behaviors. - Browsing habits. ## How to Monetize Your Website With an Ad Network There are many ways to monetize your website with an ad network. First, make sure your website is user-friendly, provide consistently updated content, and make sure it can support a volume of high traffic. An ad network can provide you with the right type of advertising content, from affiliate marketing articles, headers, banner ads, and other eye-catching advertisements from brands that align with the content or overall brand of your website or app. ## How to Measure Ad Network Performance The key metrics to measure ad network performance are: Click-Through-Rate (CTR), Cost Per Click (CPC), Cost Per Mille (CPM), and Return on Ad Spend (ROAS). Many ad network platforms provide this data to their advertisers to help them measure cost, effectiveness at driving traffic, purchases, and clicks (page views). ## How to Optimize Ad Network Performance “Monetizing and optimizing an ad network comes down to two things: knowing what kind of inventory you’re offering, and understanding how to match it with the right demand,” says Smith. Here are some ways to optimize ad network performance: - Understand your campaign visibility goals and bidding strategy: Do you want clicks, leads, or conversions? - Provide the data and metrics for each campaign that the ad network needs to perform at its best. - Make a scaling-up plan. ## How to Choose the Right Ad Network ### Budget Paying for an ad network can run from a few dollars a click to upwards of thousands of dollars a month, depending on your target audience, the type of ads you’re running, and the ad network platform’s capabilities and deliverables, like metric reports and other data. Research and compare the best ad networks for your business before committing to a single platform. Consider both monthly costs and any anticipated revenue share (typically on the publishing side). ### Goals Ad networks should align with your goals, depending on what you want from them, including monetizing your ads, generating leads and sales, or offering partnerships with high-quality sites to connect with a specific audience. ### Audience Some ad networks offer access to a specific target audience, while others are more general, providing limited control over ad placements and who sees them. ### Requirements Some ad networks have specific requirements that must be met, including minimum traffic and visibility standards, content guidelines that must be adhered to, and even geographical restrictions. ### Tracking Tools “If you’re thinking about using an ad network, I would recommend looking into one that allows you to track your campaign metrics,” advises Breanna Hendry, social media marketing director at Minky Couture. “This includes things like CTR, conversions, and CPA. Then, you should use this data to optimize and figure out what is and isn't working. They also make it easier to test different creatives, formats, and placements to see what resonates most.” ### Editing Tools Many ad network platforms allow you to manage, edit, and even pause ad campaigns based on their performance and audience share. Being able to adjust your weekly or monthly campaign budgets, remove ads with low visibility or traffic, or increase ad content on sites with higher traffic may be possible, but these offerings tend to come at a higher cost. ## Ad Network vs. Ad Exchange An ad network compiles and captures a wide range of ad space inventory to sell to advertisers. Ad exchanges are open marketplaces where advertisers and publishers bid on rates in real-time directly from one another. ## DSP vs. Ad Network A Demand Side Platform (DSP) is a platform used by advertisers to bid in real-time auctions for ad space. An ad network gathers up all the available ad space for sale and matches. Ad networks are intermediaries between advertisers and online publishers who want to buy and sell ad space. ## Key Takeaways An advertising network is an intermediary entity that connects advertisers and publishers to facilitate the sale and purchase of ad space. These platforms can also provide advertisers with data and metrics to help them improve their reach and generate revenue. If you are a small business, using an advertising network can improve your visibility and save you time and effort in finding available ad space. There are different ad networks (with varying tools and price points) to choose from — some focus on specific audiences, others specialize in a particular kind of inventory, such as mobile apps, and others partner with high-profile publishing sites. ## Frequently Asked Questions (FAQs) ### What is Google Display Network (GDN)? The Google Display Network (GDN) is a group made up of more than 2 million apps, websites, and videos where Google Ads are posted. ### What are the Big 4 advertising networks? The four largest global advertising holding companies are collectively referred to as the "Big 4," and own numerous major marketing firms and advertising agencies worldwide. The "Big 4" are WPP (UK), the Omnicom Group (USA), the Publicis Groupe (France), and the Interpublic Group, often referred to as IPG (USA). ### What is the biggest ad network? Currently, the largest ad network is Google Ads — often referred to as the Google Display Network (GDN). ### Do ad networks support programmatic advertising? Yes, ad networks support programmatic advertising in many ways, and, in fact, it is a part of an ad network’s system. Programmatic advertising allows advertisers with an ad network to place ads and brand-related content in real-time, among other abilities. ### How do I block low-quality ads in my ad network? You can install ad-blocking software, block specific advertisers via your ad network platform, and make adjustments in your browser extensions and settings to keep low-quality ads from popping up. ### How do I integrate an ad network into my website? You can integrate an ad network into your website by signing up for one. It will connect with your website or app and begin the process of integrating ads onto your spaces. ### Why are my ad network earnings low? There are several reasons for low ad network earnings, including improper ad placement or low visibility on a site or app. Issues with technology on the site, low visibility for ad clickability, and ill-defined ad metrics can all contribute. ### Which ad network pays the highest CPM? There are a lot of variables in play here, making it hard to give a definitive answer, but there are a few contenders, including Google AdSense, Adsterra, and RevContent. ### Which ad network is best for small businesses? Several factors influence the decision to choose an ad network, including budget and target audience. Google Ads is a popular choice for small businesses, as well as AdPushup, Realize, and Facebook Audience Network. --- ### Ad Compliance: Major Challenges and How to Ensure You Meet Content Policies URL: https://www.taboola.com/marketing-hub/ad-compliance/ Last Modified: 2026-06-04 09:09:07 The world of online advertising moves fast, often down to the millisecond. Making this even harder is the fact that running a successful campaign nowadays requires more than just compelling creative and optimized bidding — it demands strict adherence to an ever-changing landscape of rules that can be tough to keep up with. There’s a term for it: ad compliance. Failing to meet these standards doesn't just result in frustrating ad disapprovals (though that’s definitely a part of it); it exposes your brand to legal risks, regulatory fines, and reputational damage. Let’s break down what ad compliance is, outline the major challenges in the digital era, and talk about how to create a structured workflow for keeping your campaigns clean, legal, and profitable. ## What Is Ad Compliance? Starting with the most basic understanding of the term itself, compliance refers to ensuring all marketing activities adhere to legal regulations, industry standards, and internal company policies. It’s the essential guardrail that keeps your brand safe and your campaigns running smoothly. ### Honest Advertising and Truthfulness (FTC/ASA) Ads must be legal, decent, honest, and truthful, with all claims substantiated by evidence. If you’ve ever been scammed or misled by an ad, you know there’s good reasoning behind this. Regulatory bodies like the Federal Trade Commission (FTC) in the U.S. and the Advertising Standards Authority (ASA) in the U.K. impose strict rules on financial claims, health claims, and any content that could be considered misleading. The core principle here is transparency and accuracy. ### Substantiating Claims Say what you can prove, and prove what you say. Every quantifiable claim, such as "99% effective" or "lowest price guaranteed" needs to be backed up by verifiable evidence. As an advertiser, it’s your responsibility to have this evidence readily available in case of an audit or consumer complaint. ### Clear Disclosures Disclosures (like terms and conditions, financial disclaimers, or product risks) must be front-and-center, and close to the claim they qualify. Burying important legal text in tiny font at the bottom of a page or requiring multiple clicks to find it is a common violation, especially in finance and health sectors. ### Data Privacy and Consumer Consent (GDPR/CCPA) Compliance heavily involves adhering to laws governing how consumer data is collected, used, and stored. Major legislation like the General Data Protection Regulation (GDPR) in Europe and the California Consumer Privacy Act (CCPA) mandate that advertisers respect consumer data rights. ### Explicit Opt-In To lawfully collect user data, particularly in regions governed by GDPR, advertisers must obtain explicit consent using un-checked default boxes. Users must actively choose to opt in, and clear, easy-to-use opt-out mechanisms must be provided at all times. ### Data Security This is a major concern nowadays, and you’ve probably seen multiple news stories about data breaches across the board. Ad compliance extends to maintaining the confidentiality, integrity, and availability of customer data. This means using secure systems for data storage, encryption, and ensuring strict access control to prevent these types of breaches. ### Brand Safety and Suitability This focuses on ensuring your ads run in appropriate spaces, avoiding content that could damage brand reputation, such as hate speech, controversial politics, or explicit material. Brand suitability involves setting specific comfort thresholds for content. ### Third-Party Verification To effectively manage safety and suitability on the open web, advertisers should use recognized industry accreditations and tools for pre-bid protection. Vendors accredited by the Media Rating Council (MRC) are essential for validating inventory quality. ## The Challenge: Manual Review vs. Real-Time, Global Compliance As I’ve talked about above, there’s lots of rules and regulations in place for good reason, and following them can be a challenge in itself. Again, though, the speed and complexity of digital advertising brings a whole other set of significant compliance hurdles. The massive scale of modern digital media, particularly programmatic buying, has far outpaced traditional manual review processes, leading to things like delays and — the biggest thing you have control over — increased human error. ### Regulatory Fragmentation and Change Regulations are going to vary widely by industry (such as finance, life sciences, commerce, etc.), region (state, country), and product type. To make it even more complicated, they’re constantly evolving, too: What's compliant in California, e.g., probably won’t be compliant in Spain, creating a complex, fragmented legal landscape for global marketers. ### Creative Rejection Bottlenecks Manual creative review cycles by platforms (like Google or Facebook) are often a bottleneck for campaign launches. If a creative is rejected due to a minor policy infraction, the resulting delay can cause marketers to miss opportunities and slow down campaign momentum, losing out on the big bump that usually comes with a launch. ### Programmatic Risk (Open Web) The open web, particularly through programmatic buying, introduces significant risks that human oversight simply can’t address fast enough. This includes things like Invalid Traffic (IVT), non-viewable impressions, and brand-damaging placements, where an ad can appear next to unsuitable content in the milliseconds between the bid and the display. ## Best Practices for Cross-Platform Ad Compliance Understandably, this is a lot to deal with. In order to achieve sustainable performance, it’s vital to implement a structured, automated compliance workflow that covers campaign setup, creative content, and ongoing monitoring. ### 1. Pre-Launch Creative and Setup Vetting Integrate legal review early: Make compliance review a built-in part of the creative workflow, involving legal, brand, and compliance teams before any assets are finalized. This shifts compliance from a final hurdle to a core design input. Utilize AI/automation for creative vetting: Deploy AI-powered tools (like generative AI scanners) to scan ad copy, imagery, and video for banned language, required disclosures, and prohibited content in real-time before submission.This drastically eliminates delays from rejections by catching errors instantly. Verify targeting parameters: Ensure that all audience targeting, especially for sensitive categories (e.g., health, finance), strictly adheres to regional laws and platform policies, preventing discriminatory practices. ### 2. Real-Time Data and Placement Monitoring Implement consent management: Deploy a robust Consent Management Platform (CMP) to legally collect, manage, and document user consent in compliance with global privacy laws (GDPR, CCPA). This is super important for legally collecting high-quality data used for targeting and measurement. Enforce brand safety pre-bid: Use third-party brand safety vendors (like DoubleVerify or Integral Ad Science) for pre-bid filtering. This automatically blocks ad placements next to unsuitable, high-risk, or malicious content before the ad is served, eliminating programmatic risk. Automate exclusion lists: Manage and update negative keyword lists and exclusion site lists in real-time, leveraging machine learning to gauge the context and sentiment of current open web content. Monitor ad-to-landing page consistency: Ensure the product/service promoted in the ad exactly matches the product/service on the landing page, adhering to platform relevance standards. Any inconsistency is considered misleading and might risk non-approval. ### 3. Internal Governance and Training Foster a compliance culture: Promote accountability and a commitment to ethical marketing across the entire team, making compliance everyone's responsibility, not just the legal department's. Standardize with checklists: Create and regularly update compliance checklists tailored to different types of marketing materials and target regions, simplifying complex regulatory requirements for everyday use. Document everything: Maintain comprehensive records of all compliance efforts, including training records, approval processes for creative assets, and consent logs, to serve as evidence during audits or legal challenges. ## Key Takeaways Ad compliance is mandatory adherence to legal, ethical, and platform rules, covering truthfulness, data privacy, and brand safety. Since the speed of digital advertising and the fragmentation of global laws make manual compliance obsolete, best practices involve shifting to an automated, pre-launch vetting system, using AI for creative checks, and implementing real-time pre-bid brand safety tools. Establishing a culture of compliance and rigorous documentation is essential for sustained campaign success and brand protection. ## Frequently Asked Questions (FAQs) ### How can I prevent creative rejections on new platforms (beyond social/search)? The key is to leverage automation and AI to your advantage. Use AI-powered compliance scanners to review your assets (images, copy, video) before you submit them to the platform. These tools can flag banned words, check for necessary disclosures, and ensure image standards are met instantly, eliminating the back-and-forth of manual rejections. Realize, for example, offers ABBY, a Gen-AI performance tool, which is designed to provide real-time fixes for creative rejections and reduce delays. Beyond compliance checks, ABBY automates campaign setup and media planning with an Onboarding Wizard that applies best practices, helping you launch compliant campaigns faster and more efficiently. ### What is the first step to ensuring my targeting is compliant with data privacy laws? The first and most critical step is to implement a robust Consent Management Platform (CMP) on your website. This is the only way to legally obtain and document the necessary explicit consent from users (especially in GDPR/CCPA regions) to collect their data for advertising purposes, ensuring your targeting data is legally sound. When running campaigns with Taboola, you’ll benefit from built-in measures: All Realize inventory is validated against IAB standards and aligned with the Coalition for Better Ads. This provides critical, integrated support for GDPR and other privacy frameworks, delivering greater brand confidence and transparency in every impression you serve. ### How do I ensure my claims are legally sound without constant lawyer review? You’ll need to adopt a "Substantiation First" policy. Before any claim is written into ad copy, ensure you have clear, written, and verified evidence to support it. For common claims, create pre-approved compliance templates with standardized phrasing and required disclosures that can be used by the marketing team without needing a fresh legal sign-off every time. Realize can assist here by minimizing risk through its core function of optimization: The platform focuses on maximizing performance based on data, and the ABBY assistant provides recommendations on budget allocations and optimizations, which indirectly steers marketers toward successful, sustainable creatives that are less likely to rely on exaggerated or unsubstantiated claims just to win a click. ### What is the best way to handle compliance across multiple countries? The most effective approach is to organize your compliance by the strictest regional standard, often GDPR, as a baseline. Then, use compliance checklists tailored by region and language that specifically flag unique requirements, such as language-specific disclosures or prohibited local targeting categories, before campaign launch. When utilizing a global platform like Realize, your work is instantly enhanced by the system's global integrity standards. Advertisers running campaigns through Realize benefit from built-in protections across fraud prevention, content quality, and data privacy, ensuring that fundamental compliance requirements are met across the entire network's inventory, regardless of the specific country your ad is running in. --- ### Blending Generative AI With Human Expertise for High-Converting Content URL: https://www.taboola.com/marketing-hub/ai-vs-human-expertise-performance-marketing/ Last Modified: 2026-03-19 13:08:22 Artificial intelligence (AI) stands to transform every industry, from customer service to network security. Nowhere is its impact more immediate, or more misunderstood, than in performance marketing. Marketers are using AI tools to generate headlines, thumbnails, and landing pages in seconds, reshaping what it means to produce creative at scale. Yet, one universal truth remains: The best-performing campaigns still rely heavily on human judgment, strategy, and experience. To explore AI’s strengths and weaknesses, we spoke to Nadim Batista-Kuttab, CEO of Xevio, a leading performance marketing agency specializing in native advertising. Nadim’s team manages some of the largest enterprise accounts in the space, combining deep platform expertise with data-driven creative strategy. In our conversation, he explained how human expertise is vital to performance marketing, even as AI is revolutionizing campaign management. ## The Rise of the Prompter and the Decline of the Studio For years, design studios have been a necessary expense for marketing teams. Campaigns relied heavily on experts using expensive hardware and software to generate videos and design the images necessary to connect with consumers. But, as Nadim explains, AI is fundamentally changing the requirements for creative production teams. The need for a dedicated studio will gradually be eroded, he says: Prompters, instead of specialized design teams, are the future of performance marketing. “Specialized, dedicated design teams aren't necessary anymore for performance campaigns,” continues Nadim. “On the flip side, content is still very much human-produced. So, we're using AI regularly in the generation of landing pages, but they're always edited by people.” ### The Human Factor As Nadim points out, AI tools are increasingly capable of generating the raw materials of performance creative. The technology can generate dozens of assets in minutes, eliminating the need for a full photo studio with a lighting rig and multiple designers. The new high-value skill is prompt engineering: Performance marketers need skilled professionals who can instruct the AI to produce assets that meet the technical requirements. The assets must also speak to the emotional triggers and visual hooks that connect with humans. This ease of asset creation can make it easy to lose sight of your strategy, though. Each image, video, and text output should fit within the broader context of your campaigns. This is where a human expert needs to filter assets, discarding some, refining others, and constantly testing and iterating. ### The Shift From Studios to Idea Factories This gradual shift means that performance marketing teams can easily streamline operations. Instead of a full photography shoot for every asset, you’ll now have an “idea factory” where skilled prompt engineers generate multiple versions, a marketing expert chooses the best, a graphic designer or video editor polishes them, and teams test each creative. “Our recommendation right now, especially on the content side, is to use AI as an accelerator, not as a replacement to your internal tools,” Nadim says. “It's not complicated, I think, to create a good creative. The hard part is really just getting the entire value stream right. So, from click to content to product to purchase — all those pieces have to be in sync before you can see success on it. And that's the hard part. The individual pieces aren't that hard.” While AI is an invaluable tool in content production, then, it’s still just that — a tool. Humans need to refine and filter AI output to ensure campaigns convert. When you remove the human strategist and rely purely on AI-generated assets, you risk producing uninspired or misaligned creative, potentially leading to months of wasted marketing efforts. https://youtu.be/SNck87ia2To ## The Content Fallacy: Why AI Drafts Fail to Convert AI excels at generating content quickly, but as impressive as the technology can be, it struggles to compete with skilled, experienced human experts when it comes to that final strategic refinement necessary to drive high-volume, profitable conversions. “Just because you know ChatGPT can write you a landing page in 10 seconds, it doesn’t mean that landing page is going to perform remotely as well as a page that has a couple of hours of several people's time put into it, to really fine-tune those edges and make it perform better,” Nadim says. ### Generative Capability vs. Conversion Capability As already discussed, AI can easily produce something that looks like a landing page, ad copy, or script. It will typically follow the prompt, incorporate brand terms, and be grammatically correct. Conversion goes beyond perfectly crafted sentences and visually appealing images, though: In short, no technology has the in-depth understanding of your business that you’ve cultivated over the weeks, months, and years you’ve been in the trenches. The biggest issue with AI-generated content is that it lacks context. Humans generating the same content, for instance, would be aware of your competition and platform dynamics. They’d understand the key friction points and typical click-through rates of your channel. They’d also recognize subtle performance cues, like when a click doesn’t translate into a conversion or when a headline’s emotional tone clashes with the audience’s intent. That level of insight comes from lived experience, not algorithms. ## The Full Funnel Principle: Where Human Expertise Still Reigns Beyond individual assets lies what Nadim identifies as the value stream, which describes the customer journey from click to content to product to purchase. This is where human strategists shine: From managing interdependencies to spotting friction and ensuring the funnel is cohesive, optimizing content across the entire buyer journey is where human strategists provide indispensable value. "The individual pieces aren't that hard,” Nadim says. “The hard part is getting the entire value stream right. So, from click to content to product to purchase, all those pieces have to be in sync before you can see success." Human expertise is needed to: ### Bridge Disconnects One of the most common failures in performance marketing happens when the ad’s promise (headline, creative, and hook) doesn’t match what the landing page or checkout process delivers. A mismatch can lead to one or more of the following: - Higher bounce rates. - Dissatisfied users. - Poor-quality conversions. - Platform sanctions. For example, an AI-generated ad might read, “Discover the secret to youthful skin in seven days,” but when the user clicks through to the landing page, the messaging offers a generic e-guide. A user who expected immediate answers is likely to bounce. While AI likely wouldn’t spot the gap, a good strategist would catch the disconnect, adjust either the copy or the offer, and monitor results. ### Optimize Conversion Rate (CRO) AI can’t walk through your checkout process the way a human can. It won’t notice that your German-language page loads slowly on a certain publisher’s feed, or that a publisher’s page structure causes your tracking pixel to drop off. Businesses that regularly monitor their marketing efforts will likely identify those issues fairly quickly and repair them before they can become a major issue. Human teams can run multivariate tests, interpret anomalies like sudden drops in publisher performance, and foster iterative improvement. Nadim’s agency, Xevio, prioritizes testing. The Xevio team runs hundreds of landing page variants with large-scale data to refine quality, something a self-service AI approach struggles to replicate. ### Align With Policy Advertising on the open web, particularly via native networks, requires navigating a variety of issues, including: - Platform policies. - Publisher rules. - Regulatory complaints. - Brand safety concerns. Unfortunately, AI-generated copy may inadvertently violate policies — a video might include exaggerated claims, or a blog post could include banned terms. Those are items your human team members would have likely filtered out, considering both compliance and conversion optimization while crafting content. Fully relying on AI seems easy at first: You just need an image and a headline. As Nadim points out, though, self-service often fails because of compliance issues, whitelist/blacklist segmentation, and rapidly shifting platform trends. “Self-service has never worked for big advertisers, mainly because there are so many different pieces to a Realize campaign,” he says. “Without some sort of internal support, you're going to have a really hard time navigating policy issues, creating whitelists or blacklists for sites, and in general just understanding the trends of the platform, which is changing very quickly. So, my recommendation is, if you're serious about performance advertising, get an account manager.” AI-generated copy should be edited aggressively. It can be easy to focus on getting high click-through rates and conversions while editing, but with AI, it’s crucial to make sure every asset is compliant with regulations and publisher requirements. ## Key Takeaways In performance marketing, AI should be seen as an enhancement, not a replacement. It can streamline ideation, drafting, and testing, but the real competitive edge still lies in human expertise. As Nadim sees it, success comes from experience, data interpretation, and the ability to think holistically across the entire funnel, from click to conversion to retention. While AI can produce countless variations at lightning speed, it falls short when it comes to recognizing the messaging that resonates and why an audience converts. The marketers who succeed in this new AI-dominated environment will learn to pair the technology with human strategy while also continuously testing, refining, and aligning every asset to the larger journey. In the end, speed and efficiency mean nothing if they don’t bring results. --- ### How to Optimize Click-Through Rate (CTR) With Realize URL: https://www.taboola.com/marketing-hub/optimize-ctr-realize/ Last Modified: 2026-03-19 07:00:20 Click-through rate (CTR) is one of the most intensely measured metrics in performance marketing. For many advertisers, this data serves as the first indicator of whether a campaign is connecting with a target audience. Strong CTRs suggest that your advertising is engaging, your targeting is correct, and that your ad placements are positioned where users are most likely to take action. Optimizing for CTR goes beyond just tweaking a headline or testing a new visual, though: The process of building and optimizing a strong performance marketing campaign is much more nuanced and layered than you might expect. With the help of robust performance marketing platforms, advertisers can navigate this complex environment, taking into account publisher restrictions, compliance nuances, and the underlying relationship between CTR and return on investment (ROI). Understanding that CTR is part of a multi-dimensional ecosystem is crucial. A high CTR alone doesn’t guarantee conversions, but a consistently optimized CTR will improve algorithmic learning, increase your reach on publisher sites, and ultimately drive more traffic into your sales funnel. ## Three Challenges of CTR Optimization ### The Mismatch Between CTR and Conversions It’s tempting to celebrate a high CTR as the ultimate success metric, but clicks are only the beginning of the process. CTR is a top-of-funnel metric, meaning it measures initial user engagement on your ad, not a guaranteed final result. One of the biggest challenges with CTR, then, is viewing this metric within the big-picture context: A high CTR without subsequent conversions can actually be a signal of wasted spend, since this means you’re paying for traffic that’s not generating tangible business outcomes. This, in turn, highlights a bigger problem, in that performance is going to look different for every business goal. For lead generation, this could mean low-quality form submissions. For app advertisers, it could look like installs without long-term retention. For publishers, engagement may not translate into revenue. With Realize, the goal is to align click volume with business intent. CTR must always be considered alongside other metrics like conversion rate, number of conversions, cost-per-acquisition (CPA) and return-on-ad-spend (ROAS), rather than in isolation. Without this, you run the risk of optimizing for the wrong outcome that doesn’t align with your overall business goals. ### Content Tags Realize’s automated content tagging system is designed to maintain the integrity of the advertising network by ensuring that ads align with publisher standards. While this protects both users and publishers, it can also create roadblocks for advertisers. For example, businesses promoting dating apps may be tagged as having “mature themes,” even if the ad itself is compliant with Taboola policies. As the tagging system errs on the side of caution, it can sometimes mean that ads suitable for general audiences are prevented from running on premium publishers. This can be a significant challenge, as premium publishers typically deliver the highest CTRs and strongest engagement rates; ads flagged with these content tags are missing out on access to a portion of these networks’ highest-quality traffic. This directly impacts a campaign’s potential performance and ability to achieve a higher CTR. With proactive management, content tags can be worked around in order to enhance your CTR performance. Regularly reviewing your campaign dashboards for any creative flagged with tags should become a routine part of your workflow. Realize account managers see whether any creative tagged with this label is justified. If the creative is determined to be compliant with our core policies, they can help you submit a remediation request with their policy team to have the tag removed. ### Campaign Approval and Formatting Issues Small but significant details can make or break your performance marketing campaign. Something as simple as an incorrect capitalization, DKI keyword, or incorrectly cropped image file can cause an ad to be rejected. For instance, languages like German have capitalization rules where nouns are always capitalized. An ad that ignores this rule could be automatically rejected, even if everything else about the ad is correct. Or, consider a creative that includes punctuation at the end of a headline. Ultimately, what may seem like a stylistic choice could violate formatting standards that result in rejection, and what may seem like minor inaccuracy can have a disproportionate impact on your overall campaign performance. Rejected ads can also eat up valuable time for the advertiser: With a time-sensitive launch like seasonal sales, rejected ads mean that you might miss critical sales windows. This highlights the core challenge of optimizing creatives — that success often depends on operational precision as much as on creativity. ## Actionable Recommendations to Optimize CTR Using Realize ### Creative A/B Testing Creative tests are a foundational element of CTR optimization. Without it, advertisers run the risk of stagnation as audiences grow tired of seeing the same ads, otherwise known as ad fatigue. This can result in performance decline as conversion rate and CTR go down, even if all the targeting remains the same. With Realize, advertisers can continuously test headlines, images, and video thumbnails to find better variations. Running short-term tests, around 48-72 hours, can help you see if new creative outperforms previous ads in terms of both impressions and CTR. Applying frameworks like PANDA (Pre-test, Analyze, Narrow, Deploy, Assess) can also be beneficial, so you can strategically validate which creatives have the strongest CTR lift. The key to success with A/B testing is to test systematically, rather than sporadically. That means always having a consistent pipeline of creative assets to run, then making decisions based on the data from the results of these experiments. By testing in a continuous loop, you’re better able to optimize CTR over time. ### Strategic Targeting with Publisher Data One of the most underused tools in CTR optimization is publisher-level data. Many advertisers assume that they need to build rigid whitelists to secure top placements, but this isn’t always the case. With a strong CTR performance and sufficient budget, Realize’s algorithm naturally favors high-performing publishers. In addition, working directly with your account manager to get access to premium publishers can be a significant competitive advantage. The value of this approach is that it keeps your efforts data-driven, rather than assumption-driven. For example, if a campaign has a strong CTR but isn’t winning impressions on a premium publisher site, the issue could be budget constraint or a creative misalignment. However, if the campaign is already appearing on top-tier publisher sites, a whitelist might not be needed. By grounding your decisions in publisher data, you’re better able to refine your CTR optimization strategies without over-restricting your campaign scalability. ## What the Advertiser Can Expect ### Increased Impressions and Reach By resolving content tag and formatting issues, along with continually refreshing creative, you’ll see a significant boost in campaign reach. This doesn’t only mean an increase in overall impressions, but more impressions on high-quality publisher sites where CTR potential is the highest. This lift can be significant. Studies have found that ad placements in premium environments can deliver up to 36% higher engagement compared to non-premium placements. With Realize, unlocking this level of premium inventory can be a campaign-changing strategy. ### Improved Overall ROI While CTR is important, it’s not the ultimate goal — that would be better return on ad spend and increased ROI. Better CTR often leads to lower cost-per-click (CPC), which makes your spending more efficient while generating higher volumes of qualified traffic. Overall, you can expect to see higher conversion volume at the same spend level, lower customer acquisition costs, and higher ROAS as every dollar works harder throughout the sales funnel. Advertisers who choose to invest in creative testing and compliance management can see significant increases in ROI, and Realize can boost that further thanks to algorithmic reinforcement, making CTR optimization one of the most impactful avenues for profitability. ## Key Takeaways CTR optimization on Realize requires a focused strategy, incorporating creative A/B testing to fight ad fatigue, along with actively monitoring and remediating content tags, which can help you unlock premium inventory. With access and opportunities to leverage publisher performance data, you can make more targeted decisions without restricting the scale of your campaigns. Ready to turn your CTR into a true performance driver with more impressions and improved ROI? Start your journey with Realize today. ## Frequently Asked Questions (FAQs) ### What are some practical ways to maintain a consistent pipeline of fresh creative assets for ongoing testing? “The most practical way is by utilizing AI tools, which can really help to create a lot of assets and make them variable,” says Kamilla Tursunova, media account manager, growth, at Taboola. “We recommend using Realize Generative AI — ad creators know they need to produce fresh, high-quality content on a continuous basis to stay ahead of the competition and drive sustainable results.” ### If you have a high CTR but low conversion rate, what would be your first steps to diagnose the problem? The first step is to evaluate the post-click experience. Does the landing page match the content that the ad has promised? Check factors like page load speed, mobile usability, form length, and the clarity of calls-to-action (CTAs). With Realize, you can connect CTR data with conversion tracking to identify where in the funnel a breakdown may be occurring. If the ad attracts clicks but the landing page fails to convert, the issue is likely there, rather than creative-driven. ### What are specific, actionable tactics you would use to improve CTR for a search campaign? Improving CTR search campaigns requires both relevance and strong presentation. Align ad copy directly with user intent by matching headline phrasing to top-performing search queries. Use specific CTAs that promise clear value, such as “schedule a demo,” instead of a more generic “learn more.” You can also use ad extensions like siteclicks or callouts to provide more clickable entry points from your ad. When used with Realize’s creative testing framework, these tactics allow your search campaigns to continually evolve to capture user attention at scale. ### How can advertisers balance using publisher data for optimization without unintentionally limiting campaign reach or scalability? “According to our best practices, we recommend making all optimizations based on the data that you can find on your Realize dashboard,” says Tursunova. “First of all, don’t rush to pause sites: We encourage you to start optimizing only when the campaign's learning phase is over, which is usually two to three days after the campaign is launched. In addition to that, the minimum requirement is to have at least 100 clicks on each publisher and spend that equals 2-3x the expected CPA goal. This threshold provides enough data to assess performance accurately, while preventing unnecessary overspend.” ### How can you use data and analytics to inform your CTR strategy? Data is the foundation of an effective CTR optimization strategy. By reviewing and analyzing performance across creatives, audiences, devices, and publishers, you can discover which factors are consistently driving better performance. With Realize, you’ll gain access to granular reporting that provides publisher and ad level CTR performance data. By comparing campaign data against competitive benchmarks, you can identify gaps in your current strategy and invest more in high-performing creatives and publishers. --- ### 3 Affiliate Marketing Tips to Earn More Revenue URL: https://www.taboola.com/marketing-hub/scaling-affiliate-marketing-business/ Last Modified: 2025-10-26 16:26:42 When it comes to making money through affiliate marketing, plenty of attention and optimization advice is directed toward bloggers and influencers. However, they’re only half of the profit equation, and businesses promoting their products have a lot to gain as well. Here’s how it all works. ## 3 Tips for Affiliate Marketers to Make More Money “Affiliate marketing can be used by companies both to expand quickly and generate greater revenue,” says senior marketing manager Antje Eggersdorfer. “It just must be done the right way.” In fact, strong partnerships with affiliate marketing blogs, newsletters, and blogs can lead to a significant average return on ad spend (ROAS) of 12:1, according to a report by the Performance Marketing Association. The same report found that affiliate marketing investments yielded $71 billion in US e-commerce sales in 2021. But, eMarketer reports that up to 28% of marketing executives don't know the effect that affiliate marketing has on their revenue. Eggersdorfer has worked with businesses navigating their ways through market fluctuations and broad shifts in consumer behavior for 25 years. Throughout this, affiliate marketing has remained an effective and uncomplicated tool for not just navigating these ups and downs, but succeeding despite them. “It's not magic,” she says. “It's planning.” Below, Eggersdorfer and other experts share exactly how to plan your way to more customers and sales with the help of affiliate marketing. ### 1. Choose Affiliates Wisely While it may be tempting for companies to partner with many affiliates, Lucas Lee-Tyson, the chief executive officer at LTT Research, advises against casting a wide net. Instead, he recommends zeroing in on “whale affiliates,” i.e., websites, bloggers, and influencers that could drive more leads and sales. Rather than having a generic affiliate program that anyone can sign up for, and relationships that may draw zero revenue, businesses “would be far better off focusing on fewer affiliates with whom they have a closer, higher-level relationship and incentivizing them as strongly as possible,” he says. When searching for whales, it’s important to look at affiliate engagement metrics like click-through rate and conversion rates. Lee-Tyson recommends using tracking tools like Hyros and Triple Whale to help find the best revenue-driving affiliates for your brand. Other affiliate management platforms like ShareASale, CJ Affiliate, and Rakuten help companies find higher-performing affiliates to reach out to as well. ### 2. Nurture Affiliate Relationships Once you’ve identified the right affiliate partners to promote your products and services, businesses can strengthen these partnerships with a mutually beneficial revenue-sharing structure. On average, affiliates' commission rates range from 5% - 30% for physical products and 20% - 50% for digital products and subscription services. Always be sure to research your industry to ensure your results are equal to or better than the standard, says Stephen Montagne, founder of NetHustler. As much as commissions matter, so do the terms for how often affiliates get paid. “If you pay them two months after the sale, it might be harder to motivate them or sign up new affiliates,” he notes. In addition to a proper payment structure, Eggersdorfer and Montagne agree that it’s crucial to provide affiliate partners with the appropriate banner images, videos, promo codes, email swipes, and other product details and assets. “The easier a company makes it for the affiliate to promote them, the better,” Montagne says. ### 3. Campaign Management After forming quality affiliate partnerships and providing support to nurture these relationships, companies can optimize affiliate marketing channels. While affiliates will have their own resources to maximize sales, tools like Taboola Pixel can help companies obtain data far beyond how many clicks ads receive for their products. For instance, the Chrome extension will track when someone puts your product in their cart, but doesn’t complete the purchase, indicating a need to re-target ads to that potential customer. Other automated tools such as Smart Bid allow companies to maximize their campaigns by adjusting bids based on cost-per-click (CPC) data and the likelihood of converting these clicks into revenue. These resources not only allow companies to drive more sales, but also allocate their ad budgets for more profitable results. ## Key Takeaways Businesses will see the most return from developing a small number of high-quality relationships with affiliate partners that have an engaged audience representing their ideal customer. Once you form these partnerships, you can strengthen them through equitable revenue sharing and reliable support with images, videos, email swipes, and other promotional materials. Companies can utilize various data and tracking tools to optimize their ad spend and target consumers who are more likely to buy their products and services. ## Frequently Asked Questions (FAQs) ### What is the best strategy for affiliate marketing? Affiliate marketing doesn’t have a best or catch-all strategy, Montagne says. But, picking the right niche, and using data and common sense to gain an understanding of your prospective customer base and how your products and services can help them, are all essential for maximizing returns. “If you can do that, then you can succeed, whether it's with free or paid traffic,” he says. ### Can you make $100 a day with affiliate marketing? Experts agree that it’s entirely possible for companies to make $100 a day through affiliate marketing with the right partnerships, quality promotional materials, and comprehensive audience data. Within a few months, a reasonable goal should be to make twice that daily, “if you’re doing it well,” Eggersdorfer says. Lee-Tyson adds that after an initial setup and implementation period, “any marketing channel should be generating at least 10% of the business' total revenue in order for it to be worth it.” ### What is the 80/20 rule in affiliate marketing? Also referred to as Pareto's Principle, the 80/20 rule states that about 80% of consequences result from 20% of causes. With respect to affiliate marketing, the 80/20 rule can mean several things: 20% of your products and services drive 80% of your sales, or 80% of sales come from 20% of affiliate content. Ideally, the 80/20 means that 20% of your efforts will account for 80% of sales. That is why access to detailed data and analytics is essential for knowing which strategies to prioritize to provide the best return. --- ### Tracking Parameters: How to Use Them Effectively in Digital Marketing URL: https://www.taboola.com/marketing-hub/tracking-parameters/ Last Modified: 2026-03-09 08:13:04 In digital marketing, launching a campaign doesn’t mean much if you’re not measuring performance. Businesses spend thousands or millions of dollars on the components of marketing, including ads, content, and promotions, and they need to know what works and what doesn’t. Tracking parameters provide behind-the-scenes data that lets you see where your traffic is coming from, how users behave after clicking, and which campaigns drive conversions. Tracking parameters may seem like just small strings of text attached to the end of a URL, but they play an outsized role in attributing results and shaping future strategy. Without them, marketers would be guessing at performance rather than measuring it. ## Understanding Tracking Parameters ### What Is the Primary Purpose of Using Tracking Parameters in Digital Advertising? The main purpose of tracking parameters is attribution. They allow marketers to connect a click on a link to subsequent actions, such as browsing a website, signing up for a newsletter, or completing a purchase. This connection helps businesses measure the return on investment (ROI) of specific campaigns, platforms, and ad creatives. In essence, tracking parameters bridge the gap between an ad impression and a business outcome. They help answer the question: Which marketing efforts are driving results? ### How Do Tracking Parameters Help in Attributing Traffic and Conversions? When users click on a URL with tracking parameters, those details are passed on to analytics and advertising platforms. The parameters typically specify the source of the click (for example, Facebook or Google Ads), whether they’re from email or paid search, and the campaign name. This data is then stored in analytics tools such as Google Analytics. Tracking parameters are also sometimes called URL parameters or query parameters. They are bits of text added to the end of a web address that usually follow a question mark (?) in the URL and provide structured information to analytics systems. For example: https://example.com/landing-page?utm_source=google&utm_medium=cpc&utm_campaign=spring_sale Everything after the question mark is the tracking parameter. You can customize the form of the tracking parameter after “utm_source” which stands for urchin tracking module. ## Common Types of Tracking Parameters ### What Are UTM Parameters and How Are They Used? UTM parameters are the most common type of tracking parameters, standardized by Google Analytics. They include fields like: - utm_source (where the traffic came from, e.g., Facebook). - utm_medium (the channel, e.g., email, cost per click (CPC), social). - utm_campaign (the campaign name, e.g., holiday_sale). - utm_term (keywords for paid search ads). - utm_content (to differentiate between ad creatives or links). Marketers use UTMs to track and compare performance across different platforms. ### What Are Some Other Common Tracking Parameters Used by Advertising Platforms? - gclid (Google Click Identifier): Used by Google Ads to connect ad clicks with performance data in Google Analytics. - fbclid (Facebook Click Identifier): Used by Facebook and Instagram to track ad clicks across their ecosystem. - msclkid (Microsoft Click ID): Used in Microsoft Ads (Bing). These identifiers are automatically appended by platforms and ensure accurate attribution even without manual tagging. ### How Do Custom Tracking Parameters Work? Custom parameters allow marketers to create their own tags that capture details beyond standard UTM fields. For example, a brand might use a parameter like utm_region=north_america or utm_influencer=johndoe to track performance by geography or partnership. Custom parameters provide flexibility and can deliver insights tailored to business needs. ## How Tracking Parameters Work ### How Do You Append Tracking Parameters to Your URLs? Marketers append parameters by adding a ? after the main URL, followed by key-value pairs separated by =. Multiple parameters are linked using &. For example: https://example.com/?utm_source=linkedin&utm_medium=social&utm_campaign=product_launch ### What Happens When a User Clicks on a URL Containing Tracking Parameters? When someone clicks the link, their browser sends the full URL, which includes the parameters, to the website server. The analytics platform reads these parameters and associates them with the user session. ### How Do Analytics and Advertising Platforms Read and Interpret These Parameters? Analytics tools (like Google Analytics, Adobe Analytics, or Mixpanel) detect the parameters and categorize traffic accordingly. Advertising platforms then use these signals to link ad impressions and clicks to conversions. ### How Is the Data From Tracking Parameters Used in Reporting? Data collected from tracking parameters is aggregated into reports that show metrics like sessions, bounce rates, conversions, and revenue that is broken down by campaign, channel, or creative. ## Benefits of Using Tracking Parameters ### How Are Tracking Parameters Essential for Understanding the Performance of Different Marketing Channels? Tracking parameters make it possible to evaluate how each channel contributes to traffic and conversions. Without them, all traffic from social media, search, or email might appear as “direct” or “unknown” in analytics reports. With parameters, marketers can isolate performance and optimize accordingly. ### How Can Custom Parameters Provide Even More Detailed Insights? Custom tags let marketers measure performance at a level of granularity they choose. For instance, with a custom parameter, you can track which influencer drove the most sales or which region responded best to a promotion. This level of detail supports smarter decision-making and more efficient ad spending. More broadly, tracking parameters allow: - Accurate campaign tracking: Ensuring that every click is tied to the right initiative. - Source/medium identification: Knowing exactly where traffic originates. - Granular analysis: Pinpointing which ads, creatives, or audiences are responsible for results. ## Best Practices for Using Tracking Parameters ### How Should You Consistently and Accurately Tag Your URLs? While developing your campaign, be sure to use clear naming conventions, as too many labels will lead to confusion when you sit down to analyze the data. Other best practices include always using lowercase letters, avoiding spaces (pros use underscores or hyphens to separate terms), and standardizing campaign names. ### How Can You Avoid Common Mistakes When Using Tracking Parameters? Some pitfalls include: - Using too many parameters, which can make URLs unreadable. - Forgetting to encode special characters (like spaces). - Not testing links before publishing, which can result in broken tracking. ### What Tools Can Help You Generate and Manage Tracking URLs? Free tools like Google’s Campaign URL Builder, or paid platforms like HubSpot and Bitly, can streamline the process of tracking URLs. Spreadsheets or dedicated tagging tools also help teams manage consistency across campaigns. ### How Do You Ensure that Tracking Parameters Don't Negatively Impact Your Website Functionality? If parameters ever interfere with page loading or user experience, try to simplify them, and avoid adding parameters that duplicate content unnecessarily. Slow page loading speeds and poor user experience can lead to pages dropping in search rankings. Be sure to implement canonical tags to indicate the preferred version of your pages to search engines. ### How Should You Document Your Tracking Parameter Strategy? To document your strategy, maintain a central document like a spreadsheet or internal wiki that records all campaign tags, naming conventions, and usage rules. This ensures team alignment and prevents mistakes across departments. ## Key Takeaways By appending structured information to URLs, marketers can accurately attribute traffic and conversions, evaluate performance by channel, and make data-driven decisions. Consistent tagging, careful management, and proper documentation are essential to avoid mistakes and ensure reliable reporting. ## Frequently Asked Questions (FAQs) ### What are UTM codes and how do they relate to tracking parameters? UTM codes are a standardized type of tracking parameter used primarily with Google Analytics. They are a subset of tracking parameters. ### How many tracking parameters should I use in a URL? Only use what’s necessary. Typically, three to five parameters (source, medium, campaign, term, content) are enough. ### Can tracking parameters affect my SEO? Tracking parameters can create duplicate content issues if search engines index multiple versions of the same page. Using canonical tags or excluding parameterized URLs from indexing prevents this. ### Are tracking parameters case-sensitive? Yes, most analytics platforms treat uppercase and lowercase values differently. “Email” and “email” would be recorded separately. Stick to lowercase for consistency. ### What are some common reporting dimensions based on tracking parameters? Typical dimensions include source, medium, campaign, ad content, and keyword. These dimensions enable marketers to break down performance and make comparisons across campaigns. --- ### Incrementality: How Do We Measure It? URL: https://www.taboola.com/marketing-hub/incrementality/ Last Modified: 2025-10-27 10:05:50 We’ve all heard of the so-called “butterfly effect,” a theoretical concept wherein a small change in the initial conditions of a complex system can lead to large, unpredictable differences in a later state. It posits that even a seemingly insignificant event, like a butterfly's wing flap, can trigger a chain reaction with huge, unforeseen consequences unfolding later. The butterfly effect is all about the unexpected and unforeseen, which are things marketers don’t waste much time on. That’s because skilled marketers know that with careful study of data and proper planning, advertising campaigns can largely be predictably effective and successful, and that they can be improved upon and optimized even once underway. Both new marketing campaigns, as well as changes made within an existent marketing campaign, can lead to great success, but unlike with the butterfly effect, these successes are anything but random or unexpected. Instead, they are planned for, tracked, and measured. ## What is Incrementality? Incrementality in marketing measures the additional lift a marketing activity provides, showing how much a specific action directly contributes to a desired outcome that wouldn't have happened otherwise — this can be more sales, more sign-ups, more clicks or likes, or other actions taken by targeted users. It can be used to study a campaign as a whole (as compared to sales that would or would not have occurred without the ads) or to changes within a given campaign. Unlike standard attribution, which assigns credit to touchpoints, incrementality uses test-and-control examination to determine the true, causal impact of marketing efforts, helping advertisers eliminate waste, find growth opportunities, and optimize return on ad spend (ROAS). What incrementality is decidedly not is simply spending more money or running the same advertising another time; it refers to conscious, specific changes that can be measured and learned from. ## Why Do Advertisers Need Incrementality? Advertisers need to consider incrementality to understand a campaign's true, additional impact by differentiating actual sales driven by the ad, from those that would have occurred anyway. This allows them to optimize budget allocation, eliminate wasted spend, identify truly effective channels, and improve overall return on investment (ROI) by making data-driven decisions rather than relying on potentially inflated attribution metrics. ### Helps Prove The Value of Marketing At the most basic level, paying attention to incrementality helps prove that marketing and advertising are worth it. Tracking the lift in sales (or other desired actions by users) after a marketing campaign kicks off can help show its value, and can be compared to previous or concurrent efforts, as well. ### Maximizes Return on Ad Spend By seeing what creative is demonstrably effective when you launch or tweak an ad campaign, you can see exactly how well your ad dollars are — or aren’t — being spent. You can use this information to make sure you’re getting the best possible ROAS and are making changes as needed. ### Eliminates Bias and Guesswork Traditional attribution models can't account for conversions that would have happened organically. Incrementality corrects for this bias by effectively comparing a test group, namely people exposed to ads, with a control group of those who were not exposed to them, such as would-be consumers prior to the launch of a campaign. ## Key Incrementality Terms Explained ### Incrementality The term refers to the impact of a marketing activity on a specific advertising goal, such as an app install or a sale, beyond what would have occurred naturally. It answers the question: "What happened because we did this?" ### Attribution In marketing, attribution is the practice of matching marketing “touches” (clicks or views, for example) to specific conversions like sign-ups or sales. Incrementality, meanwhile, focuses on the causal impact rather than just the association. ### Counterfactual Analysis This refers to the hypothetical scenario (or the actual outcome) that would have occurred if a specific marketing action had not been taken. ### Holdout Testing This is a form of A/B testing wherein a portion of a potential target audience is excluded from a campaign to serve as a control group. I’ll break this down below. ## How Can We Test Incrementality? ### Define the Goals Before you try to see how successful an ad campaign or a change to a campaign was, know what you’re hoping to accomplish. Identify the specific marketing element (channel, message, ad, and so on) and the key performance indicators (KPIs) you want to measure, such as sales or leads. ### Implement Ads or Changes to Marketing Materials In Measurable Ways You might release ads with geo-fenced parameters to select demographics, only at certain times, and so on; however you plan to run your advertising, know all the parameters that you’re putting in place so you can have a sense of where they appeared and who potentially saw them — and who almost surely did not. ### Collect and Analyze Data Once you’ve run new or updated ads for a predetermined amount of time, gather performance data for both the test and control groups of people not exposed to the ads. Compare the outcomes of the test and control groups by subtracting the conversion rate of the control group from the test group's conversion rate to find the incremental lift. ## Requirements to Measure a Campaign Incrementally ### Clearly Defined Goals and Metrics Before launching a campaign or an update to advertising materials, you must define the precise objective of your testing and note the key performance indicators you will track. ### Test and Control Groups These facets of the campaign are arguably the most critical requirement for determining true cause and effect of marketing. The test group is exposed to the marketing campaign, while the control group is not. The groups must be as identical as possible in size, demographics, and behavior, to avoid bias. ### Fixed Testing Duration You must run a marketing test long enough to collect sufficient data, but short enough to avoid outside influences like seasonality, holidays, market fluctuations, other promotions, or other factors. A typical test may run anywhere from two to six weeks, depending on your sales cycle and traffic volume. ## How Do We Measure Incrementally? Incrementality in marketing is usually measured using controlled experiments that compare a test group exposed to a given campaign with a control group that is not exposed to the materials. The difference in performance (incremental lift, e.g.) between the groups shows the actual contribution of the campaign to results like sales or conversions. ### Geo Holdout Testing In this common approach to measuring incrementality, advertisers divide the market into geographic regions and withhold the campaign from a specific region (the control) while running it in another (the test). The difference in revenue (or other KPIs) seen between these groups shows the incremental impact of the campaign in that area. ### Audience Split Testing Here, a randomly selected segment of the target audience is withheld from the campaign (the control group), while the rest are exposed (the test group). This is a direct way to see how many users convert only because of the marketing effort. ### User-Level Testing This granular approach involves withholding ads from users at the individual level. The difference in conversions between the exposed and unexposed users demonstrates the campaign's incremental contribution. ## How to Calculate the Campaign Uplift To calculate campaign uplift, you need to compare a test group exposed to a campaign with a control group that was not exposed, then calculate the difference in outcomes (like sales or conversions), and divide it by the control group's performance to find the percentage lift. ### Baseline Performance Establish baseline performance by the performance of the control group, which represents what would have happened without the campaign, to see how effective your efforts actually were. ### Incremental Lift Take note of the additional sales or conversions generated by the campaign, beyond what would have occurred naturally. This is incremental lift, and it’s your main goal in terms of data and marketing success. ## How to Relate Your Campaign Incrementality to Attribution Measurement Marketers can relate campaign incrementality to attribution by using them together. Attribution identifies which marketing touchpoints led to a conversion, while incrementality validates that those touchpoints genuinely drove new business rather than conversions that would have happened anyway. This pairing creates a feedback loop, allowing for more effective budget allocation, future media planning, and a truer understanding of the marketing effort’s impact. ### Attribution This shows the customer journey and credits specific channels, campaigns, or interactions for a conversion. ### Incrementality This measures the true, additional impact of a campaign by comparing conversion rates in a test group exposed to the campaign, versus a control group that wasn't. ## Key Takeaways In marketing, “incrementality” refers to the measurement of the true impact of a marketing activity — the additional value or conversions that can be attributed directly to a specific campaign, channel, or tactic, beyond what would have happened anyway, meaning organically or through other unrelated marketing efforts. Marketing campaigns can appear successful due to factors like seasonality, brand loyalty, or external influences; incrementality helps isolate what actually works, ensuring you spend money on what drives true value instead of mistaking mere correlation. For advertisers, attribution identifies which marketing touchpoints a customer interacts with before a conversion, while incrementality measures the true impact of those marketing efforts by determining if a customer would have converted anyway. Attribution uses data and correlations, such as clicks and impressions, to give credit to various channels, whereas incrementality uses controlled experiments with test and control groups to establish a causal link between a marketing campaign and a conversion. ## Frequently Asked Questions (FAQs) ### Is incrementality measurement difficult? Yes, often incrementality measurement in marketing is difficult. This is due to the complexity of isolating marketing effects from the overall customer journey; the need for specialized expertise in experimental design and data science; the technical challenges of setting up tests; the potential opportunity costs of excluding customers from tests; and the difficulty of measuring across increasingly complex, interconnected marketing channels. ### What questions can be answered with incrementality measurement? Incrementality measurement answers fundamental questions about true marketing effectiveness, such as: - What is the actual revenue or sales impact of our advertising campaigns? - Which media channels or tactics are truly driving new business? - How should I allocate my marketing budget to maximize growth? It helps determine if sales would have occurred without the marketing effort, allowing businesses to optimize spend, avoid cannibalizing existing sales, and prove the genuine value of marketing investments to stakeholders. ### What is campaign incrementality vs. revenue? Campaign incrementality measures the additional impact or uplift a campaign generates beyond what would have occurred naturally, whereas revenue is the total money earned from sales or services over a given period. (All incrementality can be counted as revenue, whereas not all revenue comes from incrementality.) Incrementality answers whether a campaign caused something new, while revenue simply states how much money was brought in. If, for example, a company has sold $1,000 worth of widgets every month for the past year without any new advertising, then it runs ads and sees monthly widget sales jump to $1,500, that $500 increase is the incrementality. That $1,500 is revenue, just like the $1,000 was before it, but $500 of that larger figure can be classified differently. --- ### Ad Fraud: How to Spot Suspicious Stats URL: https://www.taboola.com/marketing-hub/ad-fraud/ Last Modified: 2025-10-27 09:51:12 As a copywriter for over 20 years, I've seen a lot of things. Trends come and go, platforms rise and fall (still miss you, Friendster), and every few years, new tech promises to change everything. But, one unfortunate aspect has stayed constant: ad fraud. It's a shadowy corner of our industry that, for a long time, was treated as a necessary evil — though it doesn’t need to be anymore. Ad fraud affects your budget, your data, and is ultimately a threat to your entire campaign strategy. With some basic knowledge, though, you can navigate this murky space and start fighting back. Because if you’re not actively fighting ad fraud, your campaigns are paying for it. Literally. ## What Is Ad Fraud, and Why Should Performance Marketers Care? Simply put, ad fraud is any deceptive practice designed to generate false impressions, clicks, or conversions for a profit. It's not a new concept, but in the digital world, it’s a multi-billion dollar problem. You might not notice it at first, especially when your campaign metrics still look great, with high clicks, decent CTR, etc., but look closer and you’ll realize that none of those clicks are turning into actual customers. Your high click numbers are a lie, and your ad spend is disappearing into thin air. Ad fraud is, unfortunately, a lucrative business for the bad actors behind it. The perps are often organized criminal groups who operate bot farms (vast networks of computers infected with malware) or create fake websites designed to attract bots instead of real people. The scammers profit directly from the deception: They get paid by the ad networks and exchanges for the impressions and clicks they generate, even though no real person has seen the ads. This means that advertisers are essentially paying for nothing. The money that should have gone towards reaching potential customers instead goes directly into their pockets. For performance marketers, ad fraud is an existential threat. Our entire job revolves around measurable results — CPA, ROAS, conversions. When ad fraud infiltrates a campaign, it completely poisons our data. We’re left making decisions based on false information, which leads to poor optimization and wasted budget. It’s a vicious cycle that makes it impossible to achieve real growth. The first step of fighting it is knowing what to look out for. ## The Most Common Types of Ad Fraud Fraudsters and scammers are a creative bunch, with a whole toolkit of schemes to siphon off your budget. Here are a few of the most common ones I've run into, and you probably will too: ### Bot Traffic This is perhaps the biggest one. Bots are automated scripts that mimic human behavior. They are programmed to visit websites, click on ads, scroll around, and generally look like real users. The traffic looks good on paper, but it never converts because it's not real. ### Click Farms Think of these as a human version of bot traffic. They are groups of low-paid workers in a physical location who are paid to sit in front of screens and click on ads all day to generate impressions and clicks. ### Domain Spoofing This is a more sophisticated scam where fraudsters create a fake website that looks like a high-end, premium publisher (like a well-known news site). Your ads are served on their low-quality, fake site, but the ad tech platforms are fooled into thinking the ad ran on the real publisher's site, so you’re stuck paying premium prices for garbage traffic. ### Ad Stacking This one is just plain sneaky. Multiple ads are placed on top of each other in a single ad slot. Only the top ad is visible to the user, but an impression is counted for all of the ads, and the advertiser pays for each one. ### Pixel Stuffing Similar to ad stacking, this involves placing an ad in a 1x1 pixel on a web page, making it invisible to the human eye. The ad still registers an impression, costing the advertiser money without any chance of being seen. ## Ad Fraud Red Flags: How to Spot It in Your Campaigns Spotting ad fraud isn't always easy, but it leaves telltale evidence in your data. When you know the signs, it’s easier to find. It's all about looking for things that don't make sense, such as: ### Sky-High Click-Through Rates (CTR) This is the reddest of red flags. You see an impossibly high CTR, but your conversions and engagement metrics are flat. Bots and click farms can generate clicks at a rate no regular humans can match. ### Abnormal Traffic Spikes I’ve had this one happen more than once. Did your traffic suddenly spike at 3 AM from a random country you aren't targeting? That's most likely a sign of bot traffic. ### Unusual Geographic Locations or IP Addresses If you see a lot of clicks from a country or a specific IP address that isn't in your target region, dig deeper and see what’s going on. ### High Bounce Rates with High Clicks This is a clear indicator that something’s up. Users are clicking your ad, but then leaving the landing page instantly. That's not a user in a hurry — it's a bot with no interest in your content. ### Lack of Customer-Level Data When you can't get clear user-level data or see a high percentage of anonymized or suspicious user profiles, it's a major red flag, and hard to fight back if you can't see who's actually on your site. This is where having a tool that provides deep insights and granular control over your traffic sources is invaluable. ## How to Fight Back: Actionable Strategies to Prevent Ad Fraud This may all seem overwhelming. After all, how are you supposed to go up against massive foreign bot farms? But, ad fraud is a fight you can win — it just takes vigilance and the right tools. ### Vet Your Partners Only work with reputable ad networks and publishers. Look for transparency and a commitment to fighting fraud. ### Monitor Your Metrics Closely Don't just look at the high-level numbers — dive into your analytics and look for the red flags mentioned above. ### Implement a Blocklist Manually block IP addresses, domains, and suspicious traffic sources that show signs of fraud. ### Use Ad Fraud Detection Tools I can't stress this one enough. Manual checks alone can’t handle the amount of work needed to combat all these (and new/emerging) types of ad fraud. You need a dedicated solution that uses machine learning and AI to identify and block fraudulent activity in real-time. This is where an AI-driven performance ad platform can be a game-changer, since it wouldn’t just find the right audience, but ensure the traffic is real, protecting your ad spend from the get-go. ## Key Takeaways Ad fraud is a massive, costly problem that skews your data and wastes your budget. To spot it, look for common red flags like impossibly high CTRs, abnormal traffic spikes, and high bounce rates. Fighting back requires vigilance and, most importantly, the right tools. ## Frequently Asked Questions (FAQs) ### What is the biggest challenge in preventing ad fraud? The biggest challenge is that ad fraud is constantly evolving, and that’s not going to stop. As advertisers and platforms develop new detection methods, fraudsters develop new, more sophisticated ways to evade them. It’s an eternal online arms race where new threats can appear at any time, requiring constant vigilance and advanced tech to stay ahead. That’s why Realize’s AI is constantly learning and adapting. It doesn’t rely on a static list of known fraudulent IP addresses or domains, since that would be outdated fast: Instead, it analyzes user behavior, signals, and patterns in real-time to identify anomalies that indicate fraudulent activity, even from sources that have never been seen before. It’s this type of proactive approach that ensures your campaigns are protected against both the old and new forms of fraud, and since it’s all done through the platform, you don’t have to worry about manually updating your defenses. ### What are the telltale signs of bot traffic in Google Analytics? When you know what to look for, you can spot bot traffic by keeping an eye out for a few key red flags in your analytics data. These include traffic from suspicious or low-quality referral sources, unnaturally high traffic spikes that occur at unusual times (like the middle of the night), and high bounce rates alongside very low session durations and zero conversions. You might also see traffic from a single geographic location with no corresponding conversions. Realize has the power to address all of this. You should definitely still be vigilant in your analytics, but Realize’s built-in protection is designed to prevent bot traffic from reaching your site in the first place. The platform’s proprietary AI filters expose invalid traffic and suspicious activity before the impression is even served. This means that when you’re looking at your own analytics data, the traffic coming from your Taboola campaigns is already pre-vetted, so you can focus on optimizing for real, human engagement and not waste time trying to filter out the fakes. ### How can I use ad fraud detection tools to optimize my ad spend and increase my ROI? Ad fraud detection tools aren’t just for blocking bad traffic — they’re also powerful optimization resources. By accurately identifying and eliminating fraudulent clicks and impressions, these tools provide you with a clearer picture of your campaign’s true performance. Once you’ve got that, you can then use clean data to make better decisions about where to allocate your budget, which creative deliverables are truly resonating with real users, and which audiences are actually converting. This ensures that every dollar of your ad spend is working toward a positive ROI, not chasing false leads. Again, Realize is more than just a fraud detection tool — it’s a full-stack performance marketing platform. It uses its AI-driven fraud protection to ensure you’re only buying quality traffic. It then uses that same AI to identify and target high-intent audiences on the open web, including from a pool of over 600 million daily active users. By combining fraud prevention with intelligent audience targeting and optimization, Realize helps you not only save money by avoiding fraud, but also makes your ad spend more effective at every stage of the funnel. You get a real picture of performance and a clear path to scalable, profitable growth. --- ### Bid Optimization: Hitting the Sweet Spot In Online Advertising URL: https://www.taboola.com/marketing-hub/bid-optimization/ Last Modified: 2026-04-27 13:12:08 Hitting the sweet spot; being in the zone; being right on target — call it what you will, but when done correctly, proper bid optimization is the best way to ensure that your online ads get served the most number of times, to the right users, and at the best price. Here’s how it works. ## What Is Bidding Optimization? Bidding optimization in online advertising involves adjusting how much you bid for ad placements (and in turn pay for placements when your bid wins). This helps you get the best possible results, such as clicks, conversions, or impressions, all while staying within your predetermined budget. Once set up, most bid optimization happens automatically, without the need for constant human intervention on the part of the marketer. But, it only works effectively if a person or a team of people have carefully put it in place. ## Where Is It Used? ### Programmatic Advertising Programmatic advertising is the automated, real-time buying and selling of online ads. A demand-side platform (DSP) is the software platform advertisers use to manage these automated ad purchases. DSPs connect advertisers with available publisher ad inventory through real-time bidding (RTB) auctions, allowing them to target specific audiences, place ads across various channels, optimize campaigns in real-time, and manage their advertising spend more efficiently. The ads you see on a site you’re visiting that appeal directly to your interests are an example of programmatic advertising. ### Search Engine Marketing You’ve no doubt seen the ads served atop search engine results pages (SERPs). Bid optimization in search engine marketing (SEM) involves strategically adjusting bids for ad placements to maximize return on investment (ROI) and achieve specific marketing goals. Those goals may include generating conversions or increasing brand awareness by using data and algorithms to determine the most effective bid amounts for keywords or other ad targets, ensuring ads appear in prime positions without you overpaying for the placements. ### E-commerce Platforms Placing ads on platforms where people are already primed to spend money can be a great way to use your ad dollars in an impactful way. Bid optimization on e-commerce platforms like Amazon involves strategically adjusting your advertising bids — or letting them be adjusted automatically — to achieve specific goals. This process involves dynamically changing how much you’re willing to pay for clicks or ad placements based on data such as product performance, audience behavior, and real-time competition within ad auctions. ## Pros and Cons of Bidding Optimization As with all aspects of marketing, there are upsides and downsides to the process of trying to squeeze every last bit of value out of your online ad spend. Pros Cons Hands-off real-time optimization. It might take you a while to notice if your optimization tool is not performing as desired. Ideal return on ad spend and budget management. The learning period still costs time and money while you find that sweet spot. Detailed data collection. Harder to change course quickly. ## Advantages ### Hands-Off Real-Time Adjustments The single biggest benefit of automated bid optimization in advertising is arguably the fact that it frees advertisers up to focus on other work. When ads are being placed based on calculations run by machines, humans can be developing new ad campaigns, or they can work on myriad other aspects of their business. ### Ideal Return on Ad Spend When you know exactly what you’re going to spend on a given ad campaign, the element of surprise is removed and you can plan your budget and manage your expenses. If you’re happy with your ROAS (return on ad spend), you can let things ride; if not, you can make changes. ### Detailed Data Collection Savvy advertisers collect much more than just revenue off the ads they place — they also collect information. The more you know about how much ads are costing you, the population to whom they are being served, and what’s working well (or not so well), the better you can plan for future ad campaigns. ## Disadvantages ### Less Immediate Visibility The self-driving car or automated train is the pinnacle of safety and convenience — until there’s a glitch, followed by a crash. While issues in automated bid optimization may not be as dramatic as a multi-car pileup, it also means you may not notice at first that things have gone awry, leading to greater losses. ### Money and Effort Spent During Learning Period One of the most amazing things about our computing machines these days is their capacity to learn about and improve how they handle the tasks we give them. But, peak algorithmic performance doesn’t emerge right away, and as with human testing, there’s still a price to pay during that learning period. ### Difficulty in Changing Course If you detect issues with an automated bidding optimization process, you can, of course, get things back on track, but it may not be as fast and/or straightforward as when doing things manually. ## Bidding Optimization Best Practices ### Setting a Core Strategy Building a core bid optimization strategy involves aligning your bidding approach with specific marketing goals, using data to improve campaign performance, and maximizing ROI. To decide which bidding strategy you should use for a campaign, align your strategy with your primary campaign goal. You can select goals like targeting cost per action (CPA) or maximizing conversions. Deciding which bidding strategy to use for a campaign can seem overwhelming, but if you focus on target CPA, target ROAS, maximized conversions, or manual cost-per-click (CPC), you can get down to the granular work. To determine the right target CPA or target ROAS, first ensure you are working with accurate conversion tracking and data. For determining the best target CPA, set a target that aligns with your overall profit margin and business goals — Google Ads can provide a recommended starting point based on historical data. For target ROAS, which focuses on conversion value (revenue, e.g.), calculate your target by analyzing your historical conversion value and setting a target at (or below) that historical rate to ensure safe profitability. Start with a less aggressive target and gradually adjust based on performance data, especially if you’re starting off without much data. You can take a more aggressive approach later. As for choosing a bid strategy — whether for clicks, impressions, conversions, or something else — you need to think in both the short- and long-term: Will it be more valuable to spend money and effort on building brand awareness now, then creating conversion opportunities later, or do you need to start generating sales right away, even if the initial pool of potential customers will be smaller? Realize analyzes historical campaign data (impressions, clicks, conversions, cost per conversion, etc.) to identify which bidding strategies have performed best in similar contexts, and can help with these processes and decisions. ### Leveraging Data for Optimization Leveraging data is essential for optimizing ad bidding, moving your efforts beyond manual adjustments toward creating more precise, efficient, and profitable campaigns. This approach relies on collecting and analyzing vast amounts of data — both historical and real-time information — to better predict user behavior, refine audience targeting, and inform automated bidding algorithms. To determine the right target CPA or ideal ROAS to grow your revenues (once you have established things are working!) again analyze your historical campaign data to understand your average CPA or ROAS, then set the target slightly above the average, in order to maintain profitability and account for likely fluctuations, and increase cash flows. Choose a target ROAS for campaigns with varied conversion values, focusing on revenue generation. Realize uses both past campaign data and industry benchmarks to suggest realistic and competitive CPA or ROAS targets. To improve your bidding strategy by leveraging historical data, ensure accurate conversion tracking and assign differentiated conversion values to specific actions customers have taken — and can take again. Then, use this data to set bid strategy targets based on past performance, leveraging value-based strategies like target ROAS to guide your bidding and, ideally, to achieve your business goals. To use bid adjustments effectively, first you need to analyze performance data for each factor (device, location, time of day, audience, and more) and identify trends — and spot areas ripe for improvement. Then apply percentage-based bid increases or decreases to improve performance for high-converting segments and reduce investment in underperforming ones. Sometimes, you may need to do deep dives, even studying the performance of specific keywords, changing them slightly, and seeing if your ROI improves. You can analyze a search term report and identify new opportunities for bidding on new or altered keywords, or you can add negative keywords that you don’t want affecting performance. ### Automated Bidding At some point, you have to take your hand off the proverbial tiller and let the algorithms have a go at ad bidding optimization. This can be nerve-racking in the early phases of a campaign: For a new Google Ads campaign with little conversion data, for example, the best bidding strategy is to start with manual cost-per-click inputs to gather initial data and control costs, then transition to a smart bidding strategy only once you have enough conversions to train an algorithm. (Realize often recommends manual CPC for new campaigns until sufficient conversion data is collected.) While “smart” bidding can use even minimal available data, manual (AKA human) CPC provides the necessary control to establish baseline performance and an understanding of your audience before you allow the artificial intelligence (AI) to take over and optimize for conversions. To “feed” your smart bidding strategy with high-quality data, ensure accurate and robust conversion tracking, make sure you have sufficient conversion data and budget, and implement value-based bidding to achieve a complete picture of your business goals. Then, refine your audience targeting, use portfolio bid strategies to group campaigns, and consider adding ad extensions to increase interactions. Finally, allow sufficient time for the algorithms to learn, and always monitor performance and make adjustments as needed. If you did all that and you’re still having issues with automated bidding, that’s not unique to you! Common reasons for poor smart bidding performance include insufficient data or budget, excessive campaign changes during the machine learning phase, unrealistic targets (like a CPA or ROAS set too high for realistic expectations), and inaccurate targeting or settings. To fix these issues, allow sufficient time for the bidding algorithm to learn, provide a larger budget to gather more data or narrow your focus, avoid frequent adjustments to settings, set realistic performance goals, and try to verify that your audience targeting and language settings are authentic to your products, platform, or services. To balance “broad match keywords” (which allows for variations on wording and phrasing) with automated bidding, focus on negative keywords to maintain efficiency and accuracy and monitor search query reports to identify opportunities. Start new broad match campaigns with a smaller budget and leverage signals like audience lists and contextual information. Use automated bidding to optimize bids in real time, seeking out new, higher-performing search terms while you refine your negative keyword list to help ensure campaign relevance. If you’re interested in setting up an ad experiment for testing a new bidding strategy, navigate to the “Experiments” section in your Google Ads account, select “Custom experiment,” and choose your base campaign to create a duplicate. In the Experiment settings, modify the campaign to test your chosen bidding strategy, then schedule the experiment with a traffic split, define your success goals, and set the duration. After the experiment runs, review the results in Google Ads to decide whether to apply the new bidding strategy to a base ad campaign. ### Budget and Scalability Scaling your marketing budget without hurting your CPA or ROAS requires a gradual, data-driven approach. A sudden, large budget increase can force advertising algorithms to target less qualified audiences, which raises costs and diminishes returns. Don’t plan for major scaling until you have enough data to account for outliers. To balance your ad spend for consistent performance, determine your overall monthly budget. Then, calculate your ideal targeted daily spend by dividing the monthly total by the number of days in the month — or, the number of days on which your ads will run. Leverage platform features for overspending on days with higher traffic and monitor key performance metrics (KPIs) like clicks, conversions, and ROI in real time to make informed budget adjustments. Finally, use historical data (and take into consideration holidays and seasonality) to adjust budgets for periods of increased or decreased consumer activity, to ensure steady performance across the month. ## Key Takeaways Bidding optimization in online advertising is an ongoing process of adjusting how much you are willing to spend for ad placements to get the best possible results. Those results come while you stay within the budget you have determined, and largely without any hands-on effort once you dial the ads in via algorithm, thanks to machine learning. ## Frequently Asked Questions (FAQs) ### How do I leverage first-party data, like website visitors, email lists, and CRM data, to create targeted audience segments for real-time bidding? Advertisers best leverage first-party data for real-time bidding (RTB) by segmenting audiences based on behavior and demographics, thereby creating customized, personalized ad campaigns. Marketers use customer relationship management (CRM) data for targeting, website visitor data for remarketing and lookalike audiences, and email lists to build detailed audience profiles. Integrating this data into ad platforms allows for more precise targeting, optimizing bids, and improving overall campaign performance and ROI. Realize can help greatly here thanks to its Predictive Audience Targeting, which was built for performance predicated on behavior rather than just identity. ### What is the best strategy for A/B testing different bidding models and creative assets on a massive scale without compromising campaign performance? To perform larger, effective-scale A/B testing without compromising campaign performance, adopt a tiered experimentation framework that isolates variables, prioritizes incrementality testing, and leverages automation. This approach mitigates risks of inaccurate or inactionable data by testing significant changes on a smaller and then ever-increasing scale before broader implementation, ensuring that optimization is continuous and data-driven, and not a waste of time, effort, and money. Regular check-ins and strategic adjustments will allow you to maintain consistency throughout the month. Realize can detect when performance could be improved with a different bidding strategy and may recommend setting up a test campaign with a different bid type (e.g., manual CPC vs. target CPA) or a different budget or CPA target. ### Beyond standard post-click conversions, how can marketers best measure the incremental impact and true return on ad spend (ROAS) of open web campaigns? To measure the incremental impact and true ROAS of an open web campaign beyond those standard post-click conversions, marketers can use controlled experiments and advanced modeling techniques. These methods isolate the causal effect of advertising by comparing outcomes in exposed and unexposed groups, offering a more accurate view of true ad effectiveness. To scale your advertising budget without negatively impacting your CPA or ROAS, you must take a gradual, data-driven approach: Abruptly and aggressively increasing your spending can disrupt the ad platform's algorithm, forcing it to find lower-quality audiences and causing your CPA to rise and your ROAS to fall. Realize helps marketers use specific and even customized tools for analysis rather than relying on more “generalist” tools, thereby helping produce more granular information and driving better results. --- ### Investing in Ads on Walled Gardens vs. the Open Web: Insights From Xevio URL: https://www.taboola.com/marketing-hub/walled-gardens-vs-the-open-web/ Last Modified: 2025-10-28 16:30:42 Should you double down on Facebook and Google, or explore the open web? It’s the question almost every performance marketer has asked themself at some point. Walled gardens like Meta and Google have become familiar playgrounds for advertisers, thanks to their predictable, controlled, and often effective environments. But, beyond those walls lies the vast, seemingly chaotic, but opportunity-filled territory of the open web. For many marketers, that frontier can feel intimidating. Walled gardens promise convenience, while the open web offers possibility, but also complexity: How do you weigh them against each other? The truth is, though, that the open web isn’t the wild west we sometimes think it is — it’s a strategic landscape that, when understood, can deliver returns that walled gardens simply can’t match. To help make sense of this landscape, we turned to Nadim Kuttab, CEO and co-founder of Xevio, who uses the perfect analogy to describe the differences between these two worlds — one that transforms how advertisers can think about investment strategy. ## The Walled Gardens: A Predictable, Controlled Mall Nadim likens walled gardens like Meta and Google to a mall. Picture walking through a closed space with a limited set of stores — a couple coffee shops, a handful of clothing outlets, some restaurants. Everything is tidy, structured, and controlled to a select number of stores, or placements. The mall analogy works well to describe the environment of a walled garden. These platforms provide a finite amount of ad inventory across their properties. Targeting is straightforward, placement options are standardized, and the environment feels familiar. In turn, this offers important benefits to advertisers. Predictable targeting, for instance on Meta, allows advertisers to segment audiences into granular buckets by demographics, interests, and behavior. The controlled environment means that campaigns are run within an ecosystem where placements and outcomes are easier to model. The ease of optimization that comes with a walled garden is also a significant benefit, as algorithms provide quick feedback loops that make it easy to adjust bid strategies or creative. The “mall” does have drawbacks, though. Limited inventory means that advertisers only have a select number of “stores” to choose from. Once the audience is saturated, diminished returns are possible. There’s also a higher degree of competition when everyone else is also vying for the same limited number of placements, driving up costs. It’s important to remember, too, that these types of environments are controlled by their platforms, meaning there are fewer options for differentiating beyond standard ad formats. Despite their limitations, it’s easy to understand why walled gardens have become the default option for many advertisers. For those newer to performance marketing, this plug-and-play option means that campaigns can be launched quickly, with reliable measurement tools. These walled gardens are also deeply embedded in the customer journey. Google Search still plays a pivotal role in capturing high-intent queries, while Meta remains a cultural hub through Instagram and Facebook, places where users spend a large amount of time. For objectives like brand awareness, reach, or direct-response, walled gardens still provide unmatched efficiency. However, advertisers who over-rely on these environments can quickly hit a ceiling. As Nadim points out, true profitability is rarely uncovered when you’re competing in the same crowded place as everyone else. ## The Open Web: A Vast and Strategic “Bazaar” By contrast, Nadim describes the open web as a “bazaar,” with a neverending stretch of stalls filled with everything from large, mainstream publishers like Yahoo or CNN, to small niche blogs and news sites. Where the mall limits your placement options, the bazaar offers infinite opportunity. There’s scale in the abundance that a bazaar environment provides, with billions of possible impressions across millions of sites. Some stalls are large, polished, and bustling; others are small and with less traffic, but a loyal audience. The key difference here, though, is choice: In the bazaar, you don’t have to buy from every stall. You can be much more selective, curating your own placements that align with your audience and goals. For performance advertisers, the open web’s large reach means that you can reach hundreds of millions of unique users that walled gardens simply can’t cover. Ads can run alongside a wide range of content, from breaking news to lifestyle tips. These placement-level strategy opportunities are enticing to advertisers who can whitelist high-performing sites, blacklist underperforming ones, and optimize with granular data. Ads can often feel disruptive in walled gardens, but on the open web, they can be highly contextual and feel less out of place. A financial services ad running beside a personal finance article on a news site, e.g., is relevant in a way that doesn’t feel forced or unnatural. Studies suggest that ads placed in trusted environments often drive higher attention and brand recall than those in social feeds. Not only does this make the open web a good choice for performance campaigns, then, but also for brand-building efforts. ## ROI Optimization Strategy on the Open Web Choice is powerful, but it requires strategy. To maximize profitability on the open web, advertisers must go beyond audience targeting and focus on placement-specific performance. Nadim recommends several key tactics: ### Whitelisting and Blacklisting By analyzing campaign data, advertisers can identify which publishers, domains, or placements drive the highest ROAS. Building whitelists of these top performers means that more budget flows to quality traffic, while blacklists cut wasted spend on the least effective placements. This approach is particularly important when testing at scale: Without filters in place, budget may be passed toward low-quality sites that generate clicks, but not conversions. With careful placement management, advertisers can turn their cluttered bazaar into a curated marketplace with high-value stalls. ### First-Click vs. Last-Click Impact Realize’s data shows that first-click performance on the open web can generate three to four times more sales than last-click attribution alone. Without tracking first-click contributions, advertisers run the risk of underestimating ROI and undervaluing your campaigns. For instance, a user might discover a skincare brand via a Realize-powered content site, click through, then leave. A week later, they convert via a Google Search ad. If you only credit the last click, Google appears responsible, but the true catalyst was their first interaction. ### Lifetime ROAS (LTV-Based Optimization) Short term CPA targets are useful for daily adjustments, but long-term success heavily depends on lifetime value. Looking at 30-, 60-, 90-, or even 180-day profitability reveals the true impact of open web campaigns. Feeding this data back into Realize’s algorithm helps you scale with confidence. This is especially the case with verticals like subscription services, where profitability may not be apparent until month three or four. By only measuring ROAS, advertisers may undervalue important placements that deliver loyal, high-value customers. ### AI-Powered Optimization With platforms like Realize, AI-driven optimization accelerates performance, thanks to automatically adjusting bids, placement updates, and budget optimization based on granular signals. This leaves performance marketers free to focus on strategy, rather than manual campaign tweaks. AI is one of the best ways to accelerate creative testing, too, looking at the best performing headlines, images, and CTAs faster than manual experimentation. In a landscape as vast as the open web, this speed is critical for maintaining campaign efficiency. ### Intraday Scaling One unconventional tactic that Nadim highlights is intraday scaling. This is where budgets are aggressively increased during historically strong performance windows, such as weekends or paydays. Advertisers can multiply daily spend by 5-10x, unlocking profitable scale while maintaining overall campaign effectiveness. This is only possible on the open web, where the large quantity of inventory supports sudden bursts of spend, without oversaturing a limited audience. ## The Tradeoffs: Open Web vs. Walled Gardens Ultimately, the choice between investing in a walled garden or the open web isn’t binary, coming with tradeoffs that performance marketers must understand. While there’s control and predictability with walled gardens, they limit how and where ads appear. The open web offers greater flexibility and scale, but requires more active management. The walled garden can also feel easier, which is enticing for performance marketers stretched thin. Open web placements require more strategic thinking when it comes to placement analysis, LTV modeling, and creative iteration, but can pay off for those putting in that strategic work. For performance marketers, the most profitable approach is rarely all or nothing: Instead, it’s about diversifying strategy and your investment — use walled gardens for predictable scale, while leveraging the open web for strategic growth and long-term profitability. Diversification also provides a buffer in a world of increasing privacy regulations and algorithmic changes that can plague walled garden platforms. Over-reliance on a single platform can put your revenue at risk, but by spreading spend across both environments, advertisers can somewhat insulate themselves from sudden shifts, while still benefiting from platforms that offer more incremental growth. ## Key Takeaways The landscape of advertising is no longer confined to a handful of players, with the open web offering vast and diverse choice. True profitability requires looking beyond ROAS and measuring lifetime value, first-click impact, and the spillover effects of all types of performance campaigns. Tools like Realize make navigating the bazaar of the open web easier, automating optimizations while allowing advertisers to focus on strategy. With the advertising world at an inflection point, where open web is beginning to look more enticing due to the limits of walled gardens, it’s a strategic advantage if advertisers understand how to harness the power of both environments. As Nadim’s insights suggest, success lies in shifting perspective, treating the open web like a bazaar of opportunity rather than a chaotic risk. For advertisers willing to embrace this choice, measure impact holistically, and optimize campaigns strategically, the ROI potential is immense. ## Frequently Asked Questions (FAQs) ### What are some examples of open web platforms? Major publishers like CNN, Yahoo, MSN, and niche blogs or news outlets all fall under the umbrella of potential open web placements. ### How can I track ROI across so many different open web placements? Using attribution tools and Taboola’s analytics to measure first and last click impact, as well as lifetime ROAS, is one of the best ways to keep track of all the data accumulated by your campaigns. ### Is the inventory on the open web as high quality as walled gardens? Yes, premium publishers provide trusted environments, and with whitelisting/blacklisting, you can easily ensure that your campaigns focus on the highest quality placements. ### What’s the best way to start testing the open web? Begin small with a diversified set of placements, track performance closely, and then scale into the publishers that deliver the best long-term ROI. --- ### Beyond ROAS: The Case for Lifetime Value (LTV) by Xevio URL: https://www.taboola.com/marketing-hub/beyond-roas-ltv/ Last Modified: 2025-10-27 08:15:03 Numbers that tell a story have always been the focal point for performance marketers. Return on ad spend, or ROAS, has long been one of those top-tier metrics that signal your campaigns are working. Spend a dollar, make four — a clear formula for success. What looks good in the short term can hide the real truth about profitability, though. A campaign showing a strong ROAS today could flatline when you try to scale. Another might seem underwhelming at first, yet quietly drive customers who come back again and again. This is because ROAS tells you what happened after the first click, but not what happens after the customer relationship begins. To unlock true growth metrics, marketers must move beyond surface-level metrics and look at what each customer is worth over time: In other words, their lifetime value (LTV). Understanding and optimizing for LTV demonstrates the full financial impact of your campaigns, turning isolated wins into sustainable performance. ## The Short-Sightedness of ROAS: Why the Picture Is Incomplete ROAS is fast to calculate, easy to benchmark, and gives an immediate sense of effectiveness, but it primarily measures revenue from the first purchase, not the entire customer journey. When marketers rely solely on ROAS, they risk optimizing for the wrong outcomes. A campaign that delivers strong initial returns may not generate repeat business, while a campaign that looks unprofitable within the first week might actually attract loyal, high value customers who purchase multiple times in the coming months. This is particularly evident in e-commerce and lead generation campaigns, where conversions don’t always translate into immediate revenue. Some customers take weeks to complete a purchase, while others enter the funnel as leads, then convert months later. If marketers pause those campaigns, you risk cutting off your most valuable audience before results can even be gathered. For example, picture two e-commerce campaigns running side by side: Campaign A drives an average order value of $50 with a ROAS of 4:1 on the first purchase. Campaign B delivers a 2:1 ROAS on initial orders, but these customers have a 60% repurchase rate in the coming months, with many spending more as their brand trust builds. By the end of the first week, Campaign A appears to be the better investment, but at the three-month mark, Campaign B delivers nearly double the total revenue per customer and a stronger overall profit margin. The difference isn’t in creative or targeting, but time: Campaign B’s customers kept buying long after Campaign A’s churn post-purchase. This is where the short-sightedness of ROAS is apparent — it rewards immediacy, not long-term endurance. When you only measure performance in the first few days or week, you risk turning off campaigns that quietly build long-term customer relationships. On platforms like Realize, it’s possible to attribute data to first and last click attribution, taking into account advertising placements where a first click may be generated, but a sale doesn’t take place until later. Nadim Batista-Kuttab, CEO and co-founder of Xevio, notes that his company saw that for every last click sale generated by Realize, there were three to four first click sales that began from that same interaction. Without both first and last click, you miss out on the true impact of your campaigns. ## LTV as the “King of Metrics”: The Key to Sustainable Growth Calculating LTV doesn’t need to be complicated. Use this simple formula: LTV = (Average Purchase Value) x (Purchase Frequency) x (Customer Lifespan) For example, if your average customer spends $75 per order, makes three purchases a year, and stays active for two years, their LTV is $450. Different industries use variations of this framework: - E-commerce: Calculate the average revenue per customer over a six- to 12-month window, factoring in repeated purchases and upsells. - Lead generation: Track the average contract value or recurring revenue per converted lead. - Subscriptions: Use churn rate and average monthly spend to estimate total customer lifespan. The most important part to remember is the overall perspective. Even an approximate LTV benchmark can help you to understand true profitability far better than a single ROAS snapshot. ### Long-Term Growth ROAS and LTV don’t need to compete, and instead should both be used to review the agility of your campaign in the short term, and the long-term strategy moving forward. In the early stages of a campaign, optimizing for immediate metrics like CPA and ROAS helps you quickly identify what’s working. Later, you can go back and review which campaigns actually drove more customers. Measuring at 30, 60, 90, or even 180 days can help you see how each channel contributes to profitability. Campaigns that look excellent after one week could be poor long term, while others that are performing averagely could be the standout winners by two or three months. Integrating data is essential for getting this big picture view. Platforms can only optimize based on the numbers they receive: If you’re not feeding long-term profitability data, such as repeat business or lifetime revenue, into the algorithm, you will continue optimizing only for short-term efforts. When you share information back into a platform like Realize, the algorithm begins to understand what a high value customer looks like for you. It will then begin to prioritize placements, audiences, and creative to attract these types of buyers, rather than always going for the cheapest click. Over time, these shifts compound. Your cost per conversion may be slightly higher, but your total revenue per customer rises. ### Native Advertising and LTV One of the most powerful ways to harness LTV is through native advertising on the open web. This reaches customers when they’re in the right mindset for researching and learning, making it ideal for generating awareness and attracting high-intent customers early in the funnel. In contrast, walled gardens like Meta and Google mean that there’s limited inventory and highly competitive auctions, unlike the great “bazaar” of the open web. Marketers working with Taboola can choose exactly where and how to engage users, refining targets that gather the strongest performance data. This flexibility not only helps capture first-click traffic, but builds stronger and more lasting relationships with audiences across content that these users trust. The result is an ecosystem where LTV can flourish. When measured over time, those early engagements reveal their true worth. This means that in an open web environment, first click information is often just the beginning of a longer, more profitable relationship. ### Data and AI in the LTV Flywheel LTV-focused marketing efforts rely on endless amounts of data. The more you understand about your customers’ behavior, the more effectively you can optimize for acquisition and retention. Today, AI makes that process faster and smarter. Platforms like Realize use AI-driven modeling to identify signals that predict high-value customers, and help advertisers to automatically bid toward placements that are most likely to yield profitable, repeat customers. This is all possible through analyzing conversion quality and engagement patterns based on the updated information input into the platform. This in turn creates a feedback loop — campaigns generate conversion, conversion data is fed into Realize, the algorithm identifies high LTV patterns, campaigns are then reoptimized toward those audiences. Not only does this result in more conversions, but better quality ones, too. This means you can set higher cost per acquisition goals and scale campaigns more aggressively, because your spending is driving meaningful, lasting customer relationships over one-time transactions. For instance, Xevio clients now use intraday scaling, a strategy where budget is temporarily increased to account for best-performing days. Because campaigns are optimized toward long-term LTV, scaling harder without cannibalizing profit is now possible. ### How to Implement LTV-Focused Strategies To make LTV actionable, marketers can follow a simple roadmap: - Define your measurement window: Choose a timeframe that reflects your product’s natural purchase cycle, i.e., 30, 60, 90, or 180 days. - Unify your data sources: Blend information from Realize, Meta, Google, and your CRM or e-commerce system to create a single source of data. - Segment your cohorts: Identify which campaigns and creatives drive the highest value customers over time, not immediate buyers, and segment those out for potential further analysis. - Feed profitability signals back into Realize: Give the algorithm visibility into which customers deliver recurring value, so it can adjust targeting and bidding strategies accordingly. Each step of this process builds on the next, allowing you to refine your creative strategy and investment. Rising acquisition costs, fragmented attribution, and multi-device behavior have now made it very difficult to rely on single-metrics alone. Marketers who optimize for both short-term ROAS and long-term LTV have a much stronger understanding both of what converts and endures, with greater insight into the most loyal customers and which creatives or placements generate these, and how they compound over time. ## Key Takeaways ROAS will always be a useful tool, giving you the pulse of daily performance. But, if you think of LTV as the overall heartbeat, you get a much better read on long-term success. When marketers start thinking beyond immediate returns and focus on the full lifespan of a customer, they’re able to make better decisions, scale smarter, and build more resilient growth models. ## Frequently Asked Questions (FAQs) ### What is a “good” LTV? It depends on your industry and business model, but a good LTV to customer acquisition cost ratio is around 3:1 or higher. This means that every dollar you spent on gaining that customer generates $3 in lifetime revenue. ### How do I track LTV across platforms? Consolidate data from various platforms like Realize, Meta, Google, and your CRM into one dashboard. Attribution tools can help with long-term tracking back to original acquisition source. ### Does LTV matter for lead generation? Yes, for lead-gen marketers, lead value functions as your version of LTV. Track total contract value, renewals, or upsells from leads over time to understand their lifetime profitability. --- ### Server-to-Server (S2S) Tracking: The Future of Conversion Management URL: https://www.taboola.com/marketing-hub/s2s-tracking/ Last Modified: 2026-04-12 05:58:31 For years, marketers have relied on browser-based pixels to track conversions. Pixels are easy to set up, they’re widely supported, and are critical for measuring performance, but the world of digital advertising has changed significantly: We’re living in an age of tightened privacy regulations, ad blockers, and browsers with increased tracking-prevention features. That’s where server-to-server (S2S) tracking comes in. While not new technology, it offers a more accurate and privacy-friendly way to measure performance across channels and devices. Shifting measurement from the browser to the backend allows marketers to recapture lost conversions, improve attribution models, and implement measurement strategies that will outlast cookies. ## What is S2S Tracking? Why Is It Important to Marketers? Server-to-server tracking involves sending conversion data directly from your server to an ad platform’s server, completely bypassing the browser. Instead of relying on tracking pixels in a browser, your website backend sends conversion events through secure server calls, such as https://, often using a postback URL. S2S solves several problems in digital advertising. One of the biggest is data discrepancies between ad platforms and customer relationship management (CRM) systems. Browser-based pixels can fail for a variety of reasons, e.g., the page might not fully load, the user may leave too quickly, the network could become unstable, or the pixel might not fire correctly. These discrepancies make it difficult to trust your performance data and optimize campaigns. Another major challenge is the widespread use of ad blockers and privacy tools. A large number of people on the internet use tools that block tracking pixels, preventing advertisers from accessing conversion data. Because S2S tracking happens outside of browsers, it’s not affected by these blockers. Even the browsers themselves have moved to limit third-party cookies and tracking; Safari’s Intelligent Tracking Prevention (ITP) and Firefox’s Enhanced Tracking Protection are good examples. Again, S2S tracking avoids these browser limitations. ## How Does S2S Work? S2S tracking can sound complicated if you’re unfamiliar with the terminology, but it’s actually fairly straightforward. When a user clicks on an ad, the ad creates a unique click identifier and attaches it to the landing page’s URL. Google’s gclid or Facebook’s fbclid are examples of identifiers. When the user arrives on the website, the identifier is captured via a tag manager or script and stored as either a first-party cookie or server-side session. If the user converts, making a purchase or submitting a form, the event is captured in the backend of the website. Instead of firing a browser pixel, a “postback” is sent to the ad platform’s server. A postback is a message that’s sent from your server to the ad platform's server, and contains the key details required for attribution. Once the ad platform has the postback, it matches the click ID with the original ad click, and attributes the conversion to the correct ad campaign and ad. ## Types of S2S Tracking ### Postback URL Tracking Postback URL tracking is the simplest form of S2S tracking and is commonly used in affiliate marketing. With postback, your server sends a request to a postback URL provided by the ad platform or network. This passes along key data, such as the click ID, conversion value, and transaction details. Postback tracking is lightweight, reliable, and doesn’t require a browser. ### API-based Integrations With an API-based integration, your server communicates directly with an ad platform’s API, using structured data (such as JSON, a text-based format readable by humans) and authentication. This approach is more technical than postback, but it provides more flexibility and more in-depth data transfer. ## Pros and Cons of S2S Tracking Pros Cons More reliable data, as it doesn’t rely on browser pixels. Requires a more technical setup. Complies better with modern privacy standards. Costs more to implement. Gives advertisers more control and flexibility. Risk of data gaps if identifiers aren’t captured correctly. ## Benefits ### More Reliable Data One of the biggest advantages of S2S tracking is that it improves accuracy by capturing conversion events directly from your server. This reduces the chance of data loss that ad blockers or browser limitations can cause. ### Stronger Privacy Compliance S2S tracking allows you to control what data is collected and shared. This makes it easier for advertisers to align their activities with privacy regulations, like General Data Protection Regulation (GDPR) and the California Consumer Privacy Act (CCPA). ### Gives Advertisers More Control S2S tracking provides advertisers with richer data and more advanced analytics. This allows them to build a unified, cross-channel view of performance that you can’t get with pixel-based tracking. ## Considerations ### S2S Requires a More Technical Setup Whether you’re outsourcing or doing it yourself, S2S tracking requires some backend development skills to get it set up and running. This includes capturing identifiers, secure data storage, and triggering server-to-server events. The more ad platforms you’re working with, the more complex the integration will be. ### It’s More Expensive Having to build and maintain integrations, especially APIs, can be time-consuming and costly. In addition to the setup work and testing, you need to ensure you’re compliant with the ever-changing platform requirements. Browser-based tracking is more straightforward. ### Risk of Data Gaps For S2S tracking to be effective, it needs to accurately capture and store unique click IDs when users interact with ads. If these IDs aren’t properly collected on the landing page, accurately stored, or passed through the conversion funnel, the conversion can’t be matched back to the original ad click. Even the smallest errors can lead to attribution loss. ## Pixel-Based vs. S2S tracking: Key Differences ### Sending Data Pixel-based tracking relies on the browser to send data to the ad platform. S2S tracking doesn’t need access to a browser; it uses the backend server instead. This makes S2S tracking more reliable and private. ### Vulnerability to Ad Blockers Pixels are vulnerable to ad blockers — which many users employ — and browser restrictions. As such, marketers who use pixels have less control over data collection. By implementing on the server side, S2S trackers can determine precisely what data is shared. ## Implementation of S2S Tracking: How to Set Up Here is a high-level overview of the steps to set up S2S tracking: ### 1. Technical Setup You first need to decide whether to use your own server infrastructure, or pay for a managed service. For example, many advertisers choose platforms such as Google Tag Manager Server-Side or Taboola’s Realize because they take care of the hosting and maintenance. If you prefer having maximum control, you can build your own server endpoint. Either way, you’ll need one to receive data from your website and forward it to the ad platforms. ### 2. Passing Click IDs Once you’ve completed the server setup, you’ll need to decide how to capture and store unique click IDs. This usually requires reading the click ID from the URL when the user lands on the website, and then storing it in a first-party cookie or on your back-end server. ### 3. Handling Postbacks When a conversion happens, your server will send a postback to the ad platform’s server. This postback must include the click ID and any other required parameters. Each ad platform has its own specifications, so make sure you understand what needs to be included. ### 4. Debugging and Testing You must test your setup before going live. Make sure your click IDs are being captured correctly, and simulate conversions to test the postback process. This way, you can confirm that ad platforms are receiving and attributing data correctly. ## Measurement of S2S Tracking ### Log and Audit Server-Side Events Every S2S conversion is triggered from your server, allowing you to log each event for an accurate record. By reviewing these logs, you can confirm whether conversions were set up successfully, check for errors, and match events against platform reporting. ### Run Parallel Tracking During Implementation During the setup phase, many advertisers will run S2S and pixel tracking simultaneously to compare results. This can help you identify discrepancies, make sure events are firing correctly, and validate accuracy before completing the transition to S2S. ### Monitor Attribution Rules and Deduplication When running both pixel and server events, you want to understand how platforms handle attribution windows and deduplication rules. This is because a single user action could trigger two separate conversion events. Ad platforms use attribution windows and deduplication rules to determine how to handle these duplicates. ## Scalability and Long-term Strategy ### Future Capabilities A well-designed S2S setup doesn’t just fix current tracking issues — it also sets the stage for future capabilities. One advantage is the ability to integrate offline data. For example, you can match in-store purchases or call center conversions with online campaigns through the same server-side infrastructure, making for deeper offline-to-online attribution. ### Cost and Maintenance Always consider your costs and maintenance! This includes the initial setup costs, as well as ongoing server or managed service fees. That said, while S2S tracking is more expensive than pixel-based tracking, its superior optimization and value of recovered conversion data usually outweigh the additional costs. ### Website Performance S2S tracking can improve your website’s performance by reducing the number of client-side scripts and pixels on a page. The result is often faster load times and better Core Web Vitals, leading to a better user experience and boosting search engine optimization (SEO). ### Overall Data Strategy Perhaps most important is how S2S tracking fits into a long-term data strategy that relies less and less on third-party cookies. It supports first-party data initiatives and allows marketers to measure performance with maximum accuracy. It also shifts measurement from the browser to the server, providing more control over data. This will be critical, as privacy regulations and platform policies continue to evolve. ## Key Takeaways Server-to-server tracking represents a major shift in how marketers measure ad campaign performance, by shifting tracking from browsers to servers, giving advertisers more tracking accuracy, resilience, and control. This is critical in a world with increased privacy controls and less reliance on third-party cookies. Whether you implement S2S tracking on your own or through a managed service like Realize, now is the time to make the shift. ## Frequently Asked Questions (FAQs) ### How can we use S2S tracking to better understand the customer journey and inform our programmatic creative strategy? S2S tracking provides a more complete picture of user behavior across devices and channels, capturing conversions that pixel-based tracking might miss. This allows marketers to understand the whole customer journey better, refine their audience segments, and optimize their creative strategies. With Realize, for example, advertisers can integrate S2S data directly into Taboola’s platform, allowing for programmatic creative decisions that are based on accurate, cross-channel insights. ### How can we use S2S data to provide our affiliate partners with real-time conversion analytics? Because S2S tracking happens server-to-server, advertisers can send postbacks to multiple endpoints simultaneously. Realize also supports flexible postback configurations, allowing advertisers to share conversion data instantly with affiliates while maintaining full control over data governance. ### How can we use S2S to run more reliable A/B tests on our ad creative and landing pages, without the risk of client-side tracking issues? A/B testing relies on accurate conversion data. Pixel-based tracking can create noise through lost conversions, which makes testing unreliable. With S2S, conversion data is captured even when pixels are blocked or cookies are restricted. Realize centralizes this data, allowing advertisers to run tests with confidence and make creative decisions based on complete information. --- ### Single Touch Attribution: One Click to Rule Them All URL: https://www.taboola.com/marketing-hub/single-touch-attribution/ Last Modified: 2025-10-26 11:24:48 If you’ve ever asked the question, “Which campaign actually drove this conversion?” you’ve thought about the problem of attribution. Attribution models are frameworks for assigning credit to the marketing touchpoints that lead to a conversion, from the very first impression to the final click. Marketers generally start with simpler models and progress to more sophisticated ones as their data and budgets grow. For a fast primer on the major approaches and when to use them, see our broader guide to attribution models. ## What Is Single-Touch Attribution? Single-touch attribution is the practice of giving 100% of the conversion credit to one interaction on the buyer’s path. It’s simple by design and comes in two forms: first-touch and last-touch. In first-touch, the credit goes to the interaction that first brought the user into your orbit (say, a prospect’s initial click from a social ad). In last-touch, all value is assigned to the final measurable interaction before the conversion, for example, a click that comes from branded search or a direct visit to your website. ### How Does First-Touch Attribution Assign Credit for a Conversion? First-touch attribution assumes the most important metric is how a customer discovers your product, through the interaction that initiated the relationship. If someone sees your top-of-funnel ad and later converts after many steps, first-touch attributes the conversion to that initial ad. This model is popular for brand growth and upper-funnel analysis because it surfaces which channels and creatives are best at opening new customer journeys. ### How Does Last-touch Attribution Assign Credit for a Conversion? Last-touch attribution assumes the most important moment is when the customer commits to buy — the interaction that immediately preceded the conversion. If a customer clicks a retargeting ad and makes a purchase, last-touch assigns the sale to that ad, regardless of earlier influences. This model is common for performance reporting because it maps neatly to transactions and is widely supported in advertising and analytics platforms. ## Advantages of Single-Touch Attribution Single-touch is analytically useful for three practical reasons: clarity, speed, and compatibility. ### Why Is Single-Touch Attribution Considered Simple to Understand and Implement? Because these models only have two aspects, they’re simple to implement and track. Many ad tools and analytics platforms offer these models out of the box, so setting tracking up typically means configuring a setting rather than building a complex, multi-step model. Reporting is straightforward, comparisons are fast, and you can tie spending decisions to a small set of familiar metrics without much training. ### In What Limited Scenarios Might a Single-Touch Model Be Sufficient? Single-touch can be “good enough” when journeys are short, there aren’t many channels, and your sales objectives are focused. Here are a few examples: - Free trials where the last click to conversion reliably captures the decisive action. - A brand-awareness sprint where you only care which sources are best at first exposure. - Early-stage startups that are looking to validate channel fit with limited data volume. More broadly, single-touch can provide a clear, unambiguous view of either the beginning or the end of the journey. If you’re optimizing purely for net-new discovery, first-touch highlights the channels that most often start paths. If you’re optimizing for immediate conversions at the bottom of the funnel, last-touch shows which levers most often seal the deal. ## Disadvantages and Limitations of Single-Touch Attribution What you gain in simplicity, however, you lose in granularity. ### Can It Lead to Inaccurate Conclusions About the Effectiveness of Different Channels? Yes. Customer journeys are rarely one-and-done. E-commerce marketers report using roughly six touchpoints on average in their strategies, while in complex B2B motions, a single deal can involve hundreds of impressions and touches across months. In such environments, crediting only the first or the last step systematically overlooks everything in between, including content, community, PR, partner referrals, organic search assists, and retargeting. ### Why Might It Result in Misallocation of Marketing Budgets? Generally, single-touch attribution models bake assumptions into the outcomes. The first-touch model tends to over-reward awareness channels (social, PR, upper-funnel content) and under-reward closing channels (brand search, email, retargeting). Last-touch does the reverse, over-rewarding channels that show up late and under-rewarding the ones that created demand in the first place. If you scale budgets from these distorted readings, you can starve mid-funnel programs that move people forward. You may also inadvertently chase short-term wins at the expense of long-term growth. ## Comparing First-Touch and Last-Touch Attribution ### What Are the Key Differences Between First-Touch and Last-Touch Attribution? First-touch answers, “What sparked the customer journey?” Last-touch answers, “What closed the deal?” The first emphasizes demand creation, and the second emphasizes demand capture. First-touch is directional for brand growth, whereas last-touch is operational for performance reporting. ### When Might You Focus on the First Touchpoint? You want to focus on the first touchpoint when launching a new brand, feature, or market and you need to see which awareness channels actually pull net-new audiences into your funnel. It’s also useful when you’re evaluating top-of-funnel creative and messaging to learn what most often starts qualified journeys, for example, which ad or post tends to drive that crucial first click to a free trial. ### When Might You Focus on the Last Touchpoint? You’re more likely to use the last touchpoint when you’re optimizing conversion paths and closure tactics, such as checkout flows, cart-abandon retargeting, or branded search bidding. It’s equally relevant when managing channel partners whose commercial arrangements hinge on the final referral or click. ### What Are the Inherent Biases in Each of These Models? First-touch systematically favors channels that appear early (and may give too much credit to novelty or reach). Last-touch systematically favors channels that appear late (and may give too much credit to brand familiarity that earlier touches created). Neither is “wrong,” they’re just two simplified ways of looking at how a customer converts. ## Alternatives to Single-Touch Attribution ### What Are Some Common Multi-Touch Attribution Models? - Linear: Distributes equal credit to every recorded touchpoint. Simple, egalitarian, and useful for collaboration incentives across teams. - Time-decay: Assigns more credit to interactions closer in time to conversion, reflecting recency effects without ignoring early influences. - Position-based (U-shaped, bathtub): Heavily weights the first and last interactions while distributing the rest across the middle. Good for sales motions where both discovery and closure matter. - W-shaped and custom rules: Adds major weight to a third milestone like lead creation or opportunity creation in B2B. ### How Do These Models Attempt to Provide a More Holistic View of the Customer Journey? Multi-touch models spread credit for the customer journey across multiple assists, rather than awarding it all to a single click, making it easier to see where supporting interventions made the sale. That helps you protect mid-funnel programs, justify content investments, and understand how channels work together. ### When Is It Generally Recommended to Move Beyond Single-Touch Attribution? Consider moving from single-touch to multi-touch models when: - Sales cycles exceed a few days and involve repeated nurture. - You’re running integrated campaigns across three or more channels. - Budget decisions keep swinging wildly when you switch between first- and last-touch reports. - Stakeholders need to understand influence, not just origination or closure. ### How Can Data-Driven Attribution Offer a More Sophisticated Approach? Data-driven attribution (DDA) uses statistical methods and machine learning to estimate how each touchpoint changes the probability of conversion, distributing credit accordingly. Many analytics stacks now support such models: for example, modern analytics and ad platforms provide DDA or privacy-preserving last-touch models as defaults or options. DDA requires more data and care to interpret, but it better reflects reality in multi-channel journeys and aligns teams around incremental impact rather than raw clicks. ## Key Takeaways Single-touch attribution assigns all credit to one interaction — either the first or the last — and is popular because it’s quick to implement and easy to explain. You can use first-touch when you’re measuring discovery and early-journey effectiveness, and last-touch when you’re optimizing close-rate and immediate conversion tactics. Be aware that first-touch over-credits awareness, and last-touch over-credits closers: Both can misallocate spend if taken as full truth. ## Frequently Asked Questions (FAQs) ### Is single-touch attribution still commonly used? Yes. Despite its limits, single-touch attribution persists because it’s available in nearly every tool, aligns with straightforward reporting needs, and works passably for short journeys. It also maps neatly to platform constraints where last-touch is the primary supported model. ### How can I easily implement single-touch attribution in my analytics platform? In most cases, you can select first-click/first-touch or last-click/last-touch in your analytics or ad platforms’ attribution settings and then use the model comparison views to see how results change. If your stack offers a default data-driven model, you can still switch specific reports to last-click when you need a simple, conservative read on closing channels. ### What are some signs that I should switch to a multi-touch attribution model? Red flags include whiplash between first- and last-touch reports, growing investment in mid-funnel content without clear credit, longer sales cycles with many nurturing steps, and channel teams arguing over ownership of results. If you rely on more than a couple of channels and routinely see multiple assists before a conversion, multi-touch (or data-driven) will likely improve planning and budgeting. --- ### Contextual Targeting: Effective Marketing in a Cookieless World URL: https://www.taboola.com/marketing-hub/contextual-targeting/ Last Modified: 2026-03-08 09:15:04 In recent years, internet privacy laws have made it harder for marketers to harvest user data from websites, which previously allowed them to place relevant ad campaigns based on user demographics. Contextual targeting seeks to overcome this limit by placing ads based on the context of the web page, rather than user data. ## What Is Contextual Targeting? Contextual targeting is the first step in a form of programmatic advertising that places ads on platforms within an appropriate, relevant environment. Contextual advertising refers to the ad content being placed, while contextual targeting refers to the choice of platform or publisher where the ad runs. Contextual targeting can be used in native advertising, display ads, on search engine results pages, in-app advertising, and more, for a true omnichannel approach. Contextual targeting allows brands to place relevant ads on websites and in other places without having specific user demographic data. Instead, ads are placed based on the surrounding context, established by keywords, meta-data, and other indications of the topic. ## How Does Contextual Targeting Work? Contextual targeting relies on the context — that is, the surrounding content — of a web page or other platform to place ads similar to the content. One study found that 69% of consumers are more likely to look at an ad that’s relevant to the content they’re currently viewing, while 44% have tried a new brand after seeing an ad alongside content they were viewing or reading. As an example, if you’re reading an article about camping because you’re planning a trip, you’re more likely to respond to an ad for sleeping bags or a new tent. That’s contextual targeting at its best. Sophisticated algorithms (powered by AI, in some cases) identify the correct context for ad placement without relying on any user data. Contextual targeting doesn’t know, or care, who you are, where you live, or how much money you make; it also doesn’t care what sites you’ve visited previously. You’re here now, and you’re reading about camping. Since you’re reading about it, contextual targeting will assume you’re interested in camping and, as a result, may want to buy camping gear. This can significantly shorten the sales funnel, especially in a world that’s moving rapidly toward one-click or no-click search. ## What Are the Benefits of Contextual Targeting? Contextual targeting has many benefits, especially in an increasingly cookieless internet, where tracking behavior becomes harder without violating user privacy rights. ### Reaches User in the Right Mindset and Ready to Buy Performance marketing, although it’s focused on the decision-stage of the sales funnel, has potential to reach users at every stage of the sales funnel and shorten the cycle from recognition and research to buying. When people are consuming content on a specific topic, perhaps researching their options, they are already approaching “buy mode.” Contextual targeting, then, reaches people when they are most receptive. ### Precision Targeting Trying to reach a niche or even a micro-niche market? There’s content out there on virtually any topic. Contextual targeting helps you find that content and place relevant ads at the right moment. ### Cost-Effective Because you can reach your audience when they’re ready to buy, contextual targeting shortens the sales cycle and improves ROI. ### Easily Scalable With real-time performance data, certain contextual advertising platforms make it easy and affordable to scale. ### Maintains User Privacy Unlike behavioral targeting, which tracks users’ actions across the open web, apps, and other platforms to serve up relevant ads, contextual targeting focuses on the content, not the user. This eliminates the need for cookies or other data collection methods. ## What Are the Limitations of Contextual Targeting? ### Harder to Optimize Contextual targeting relies on one factor alone to target users: the content they are consuming. This makes it harder to optimize, for instance, based on demographics like audience income, geography, or past shopping behavior. All targeting is based on keywords, meta data, and website topics. ### Harder to Measure Performance Based on Audience With a lack of user data, it’s harder to measure why a campaign is successful or who it’s converting. Advertisers only know their ad resonated with a specific reader, but they lack additional data. By combining contextual data with user demographic and user behavior data, though, you can create even more targeted, effective campaigns. ### No Retargeting Because contextual targeting doesn’t “follow” users, advertisers miss the opportunity to share similar content across platforms and build affinity through familiarity. ## What Are the Different Types of Contextual Targeting? To determine proper ad placement, contextual advertising platforms must evaluate the content. There are three primary ways to do so. ### Keyword Gauging topics through keywords is, perhaps, the most straightforward way to identify the right content for contextual ads. It can lead to an accurate match — for instance, an ad for a Nissan Rogue might appear near an article reviewing the best compact SUVs of the year. That said, relying on keywords alone can also cause mistakes if keywords have two meanings, are unclear, or you’re trying to reach a more nuanced market. For instance, B2B advertisers might find themselves reaching consumers searching for similar keywords. Additionally, if a website in the advertising network isn’t using keywords properly or effectively, it could result in less accurate targeting. ### Category For broader reach, advertisers might choose content based on broad categories or topics. For instance, an e-commerce company that sells luggage could place ads on a Prime Day suitcase round-up if they used a keyword-based strategy, but they could reach a larger audience by targeting broad-based articles about travel or planning a trip. ### Semantic Finally, sophisticated advertising platforms use AI-powered analytics to pinpoint ad placements based on multiple factors: keywords, context, topics, and even user sentiment and intent. This is important, since research shows that people are 20% more engaged when viewing content on platforms they view positively, and also spend 15% more time looking at ads on those platforms. By using software that gauges user intent and sentiment while consuming content, advertisers can place ads where they will have the most impact. ## How to Implement Contextual Targeting in Advertising Platforms When you’re ready to get started, the process for implementing contextual targeting varies depending on the platform. In all cases, you want strong creative, and you want to make sure you’re targeting the right audience. ### Google Ads In your Google Ads account, under Content, click Campaigns. Then, select “content” from the dropdown menu. Google uses keywords and topics to choose display ad placements. You can also choose specific websites that allow display ads. ### DV360 DV360, or Google Display & Video 360, is a robust programmatic advertising platform owned by Google. By its nature, it’s more complicated to use than Google Ads, but it allows marketers to manage campaigns across channels, aligning media, data, and creative. To get started, you’ll want to create an insertion order that includes your budget and pacing. Choose whether you want display or video advertising. Then, the key part comes in: Select your targeted keywords, choose a category, and then choose any specific websites, apps, or channels. You can also exclude negative keywords. Finally, DV360 will show your potential reach, and you can adjust your settings to broaden or narrow the reach. You can optimize as you go by monitoring performance and adjusting keywords or categories. ### Meta While Meta advertising does not offer contextual targeting, you can replicate the results by choosing interests, behaviors, and page categories you want. Meta uses a combination of interests, demographics, and behaviors, but it won’t be as exact as pinpointing contextual targeting through a DSP with contextual capabilities. ### TikTok Advertisers can use the TikTok Pulse Suite to place ads next to related user generated content, or next to premium content from top brands like Disney, NBC Sports, or top digital publications like BuzzFeed. Ads can be placed surrounding tentpole events, like the Super Bowl, or adjacent to evergreen content. TikTok offers ad placements to fit different budgets, depending on the Pulse product you choose. ### Realize Realize places your ads on top-performing websites so that your reach expands far beyond the walled gardens of search and social. Realize uses first-party data and AI trained on 17+ years of proprietary data to deliver ads to consumers who are in the consideration and action stages of the sales funnel, for faster conversions. To get started, create an account and you’re on your way. The Campaign Performance Simulator can help you gauge the success of your ads before you spend a dollar, allowing you to estimate potential conversions based on your CPA (cost per action/acquisition) and budget. ## Is Contextual Targeting Privacy-Friendly? Unlike advertising strategies that require collecting user data to create lookalike audiences or using cookies to track past consumer behavior, contextual targeting preserves user privacy. It doesn’t require cookies or any user data: Contextual targeting cares about what the prospects are watching or viewing, not who they are. So, if you’re reading a top 10 list of camping gear, they don’t care if you’re a 25-year-old man or 65-year-old woman. Your actions, not your demographics, indicate your intent and your interest. ## Contextual Advertising Campaign Strategies Marketers can choose their contextual advertising campaigns based on keywords, categories, or semantically (i.e., analyzing the web page as a whole, going beyond keywords to look at user intent, the credibility of the website, and a host of other factors). Let’s consider the pros and cons of each. ### Keyword Strategy Highly effective, but limited, using a keyword strategy for contextual advertising relies on thinking in the same language as your ideal customers. Keywords can open your campaign up to misunderstandings, though: For example, if your ideal reader is searching for the best types of apples vs. Apple technology, they might stumble upon your ad for an iPhone case when they’re looking for the best orchard to go apple picking. That’s an obvious, and broad example, but subtle nuances can make a difference when you’re trying to find your ideal customer via keyword matching. ### Categories Placing ads based on categories, such as “smartphone technology,” may help you find those iPhone users that need a new case. But, you might also find people looking for a new phone or for ways to use their smartphone. ### Semantics Determining ad placements through semantics overcomes the challenges of categorical or keyword campaigns by using AI to determine context based not just on keywords and categories, but user intent, images, and even the reputation of a website overall, and the page’s author. ### Contextual Advertising Strategy Success Tips To create a successful strategy: -  Use a mix of tactics to determine your target audience. -  Clearly define your campaign objectives. - Develop eye-catching, informative creative. - Test your ads and placements. - Track your results. - Refine your campaign. ## Contextual Advertising Performance and Measurements Understanding your goals for a campaign helps you track and refine your creative and your targeting for greater success. Some possible goals for middle- and bottom-of-the-funnel (BOFU) campaigns include: - Signing up for a newsletter. - Providing their email for a content asset (lead generation). - Making a purchase. - Downloading an app. A metric for these would be “cost per action” or “CPA.” ### Understanding and Tracking Key Performance Indicators (KPIs) Metrics can help show the success of a campaign. If your goal is new customer acquisition, e.g., your KPI might be measured as cost per acquisition (CPA). You can also track Return on Ad Spend (ROAS), which indicates how much money you make on each ad you place. It’s calculated as ad spend for that campaign divided by total sales from that campaign. ## Contextual Advertising Optimization Optimizing your contextual advertising campaigns involves tracking results and seeking ways to improve your KPIs. ### A/B Testing A/B testing, where you run multiple campaigns and change one facet of each one to optimize performance, is one of the most common ways to optimize your contextual campaign. You can adjust a variety of factors in your contextual targeting: - Campaign dates and times: If you’re planning a campaign around a specific date, you can use historical data to determine how soon to start pushing out related creative. But, you’ll also want to experiment with specific days of the week and times of the day. - Ad creative: Test multiple versions of your creative, switching up headlines, tags, colors, images, and more. - Channels: Think beyond search and social when you’re testing various channels for your campaign. ### Doubling Down on What Works Once you’ve drilled down to the most successful combination of creative, timing, and placement, increase your ad spend until you’ve exhausted the market and started to see diminishing returns. ## Contextual Targeting vs. Behavioral Targeting Behavioral targeting focuses on what your audience has done in the past to predict future behavior. It often relies on cookies and as such can create privacy concerns. It may also not be reliable, as it may reach consumers after they’ve made their buying decision, e.g., if a user has previously visited several car dealership websites and clicked on ads for new cars in recent weeks, behavioral targeting would continue to show those ads, even after the customer has purchased a car. In contrast to that, with contextual targeting, it’s highly likely a consumer who has already purchased a car would no longer be reading or viewing content on the best cars; they wouldn’t see an ad placed near a list of the top 10 SUVs. Your creative would reach consumers who are still in the consideration phase and potentially ready to buy. ## Contextual Targeting vs. Audience Targeting Audience targeting, often called “lookalike” targeting or predictive targeting, seeks out people similar to your existing customers. This tactic can take advantage of first-party data, such as information from your CRM, to create campaigns without using third-party cookies. It relies on the assumption that your audience demographic has similar traits and interests: Data like age, occupation, income, and geography may be useful in finding your ideal audience. Audience targeting may also pull from data like how a user navigated your website, topics they are currently reading, or keywords they have searched. This can be a solid tactic to expand your reach, especially if your predictive data is accurate and reliable. ## How Does AI and Machine Learning Enhance Contextual Targeting? Artificial intelligence, large language models, and machine learning contribute to more accurate contextual targeting. Rather than looking at keywords alone, Performance AI can analyze user intent, semantic themes, photos, and more. This allows for more meaningful matches, even without cookies. Machine learning also allows you to drill down to more specific market segments in real time, enabling you to scale your campaigns faster and reduce your overall ad spend. ## Key Takeaways Contextual targeting can match your ads to relevant content based on keywords, categories, or semantic context. Using semantic targeting coupled with AI and machine learning can deliver better results in less time, since this method analyzes every aspect of the media to determine the topic down to a granular level, as well as user intent. Contextual targeting also helps maintain user privacy, and addresses the challenges of audience targeting as websites move away from using cookies that track user behavior. ## Frequently Asked Questions (FAQs) ### Does contextual targeting still work in a cookieless world? Contextual targeting is uniquely suited for a cookieless world, since it does not rely on user behavior to match ads with users. ### What is semantic contextual targeting? Semantic contextual targeting evaluates content based not just on keywords or category, but user intent, images, videos, and the overall meaning of a web page, to place relevant ads. ### What are contextual keywords? Contextual keywords help digital advertising platforms implement the proper placement for ads, by determining the goal of a web page based on words used in the content. ### How does contextual targeting work for video advertising? Marketers can use contextual targeting in video advertising by placing ads in content related to their topic. For instance, if someone is watching a video review of the latest Ford truck, Nissan might place an ad for their truck within the video. ### Can contextual targeting be used for content recommendations? Contextual targeting is highly effective for content recommendations, as it can assess what a user is viewing or considering and recommend similar content. --- ### Four Strategies For Online Shopping Store CRO: Realize Expert Recommendations URL: https://www.taboola.com/marketing-hub/enhance-cro-online-fashion-store-with-realize/ Last Modified: 2026-01-20 11:20:59 From the time they wake to the second they shut off their devices and go to bed, consumers are bombarded with marketing messages. The steady stream of ads and influencer promotions makes it tough for products to stand out. For fashion brands, conversion rate optimization (CRO) is key to overcoming those challenges, offering a chance to stand out in an oversaturated market. That’s where Realize comes in. As Taboola’s performance advertising platform, Realize offers the data, testing frameworks, and campaign tools to take you from clicks to loyal customers. But, success with Realize starts with the right strategy, so our expert Patrick Coyle, senior growth advertisers sales manager at Taboola, walks us through how to use these tools to transform traffic into profits. ## 1. Align Tracking to Improve Data Reliability Fashion e-commerce moves at a fast pace: You need to stay on top of trends and be able to pivot with little notice. But, moving at breakneck speed can create accuracy issues, particularly when it comes to tracking your data. How can the data keep up when you’re shifting directions every few weeks? Your dashboard can understandably lag behind. If you use Realize, you may notice a discrepancy between what you see on your e-commerce platform dashboard and what you see in Realize’s reporting, which may stem from the following reasons: - Time Zone Mismatches: The simplest explanation is that the time zone in your store settings doesn’t match your ad platform’s settings. This can lead to day-to-day discrepancies. - Complex Funnels: Do you use upselling or cross-selling features in your funnel? Do you have subscription flows outside the primary purchase activity? If these events aren’t tracked properly, your reported revenue will come in lower than what you’re actually earning. - Multi-Domain Funnels: In some cases, brands have their checkout or upsell flows hosted on separate domains. If that’s the case, you’ll need to ensure cross-domain tracking is in place to capture all activity. - Attribution Model Differences: Your internal analytics tool assigns credit to activity, as does Realize. If you’re unaware of the discrepancies, you’ll constantly second-guess whether your ads are performing. This results in a lack of a single source of truth. Without unifying your data, it can be tough to confidently scale your ad spend. The key is to take control of your analytics. Without a reliable tracking system, every decision you make about budgets, creatives, or scaling is essentially guesswork. For fashion e-commerce brands, accurate data can help you avoid pouring money into a campaign that isn’t working. Locking down tracking can help you gain a transparent, trustworthy view of your campaign performance. This gives you the confidence in your numbers, and that confidence can help you scale your ad budgets based on facts, not assumptions. Even better, you can start layering in micro-events for deeper insight: - Engagement: This includes time spent on page and scroll depth. By looking at how long users stay on a page and how far they scroll, you can quickly spot pages that aren’t resonating with site visitors. - Click-outs from advertorials: Monitoring which users click on an advertorial can help you understand which types of content drive interest. This data can be useful in testing content until you find storytelling angles or product highlights that motivate users to learn more. - Visits to product detail pages: When a user clicks on your product detail page (PDP), you’re one step closer to a sale. When your analytics show users are dropping off before landing on your PDP, you may need to take steps to better optimize your funnel. - Checkout initiations: When a visitor starts the checkout process, you get a clear intent to buy. If a customer exits at this point in the journey, it typically signals an issue with the checkout experience or a mismatch between the expectation and the offer. These additional signals can reveal where the user journey thrives, as well as where it breaks. You can narrow down pain points in your online presence and take measures to correct them, then analyze again. This granular view brings another benefit, too: The information can help build powerful custom audiences, fueling smarter retargeting campaigns while also nurturing customers deeper in their journey. Below is a breakdown of seven steps necessary to effectively align your tracking and improve the reliability of your data. ### Use Realize Tracking Solutions “Realize’s tracking solutions for platforms like Shopify and WooCommerce can help resolve data discrepancies by keeping setup and usage extremely simple, which greatly reduces the risk of errors — even for users with little to no technical background,” says Patrick Coyle. “Alongside the main pixel implementation across all pages, the system automatically tracks a wide range of relevant soft conversions. This makes it much easier to analyze the customer journey, understand where users drop off, and spot potential message mismatches.” “Because the base pixel is consistently deployed, even clickout events from advertorials to product detail pages are properly captured when those advertorials are hosted as subpages within the shop,” Coyle continues. “The same applies to upsell apps — purchases remain recorded without any loss in tracking. Altogether, this creates a much clearer and more reliable data picture, providing the foundation for better optimization of landing pages and product pages.” ### Use Server-to-Server Tracking For more complicated setups, a tool like RedTrack can tackle tracking across servers. RedTrack captures events across domains with far more accuracy than you get from browser-level pixels. ### Sync Your Time Zones If your setup is simpler, start here. Check your time zone on your Shopify account and make sure it matches your Realize settings. This may sound obvious, but this small step can make a big difference in the accuracy of your daily reporting. ### Add Custom Tracking Events For brands that incorporate upsells or unique funnels, your support team can help create custom tracking events that account for every revenue source. This will help capture the post-purchase activities that tracking tools often miss. ### Test Events Directly in Realize With the Realize Ads Manager, you can test whether events are being received correctly. It’s important to do this testing before launching each campaign to avoid hiccups. ### Install Taboola Pixel Helper This Chrome extension shows exactly which events are firing on which page. It’s a troubleshooting tool that can be invaluable for spotting gaps or misfires. ### Align Attribution Models Your account manager should be able to clarify which attribution model is being used in performance evaluations. Make sure this model watches how you measure success internally. Without alignment, you’ll always struggle to get consistent, trustworthy data. ## 2. Test With Campaign Groups Testing is at the heart of CRO, but many fashion marketers get bogged down by how messy and manually intensive testing can be. Typically, testing a new ad creative or landing page means: - Building a separate campaign. - Waiting for data. - Manually shifting budgets to good performers. This approach is not only slow, it also wastes spend on underperformers while you’re waiting for test results to come in. It consumes valuable time, which is a problem when you’re in a fast-moving industry like fashion. Realize simplifies testing by using Campaign Groups. This feature lets you combine multiple campaigns under a single budget, where the Budget Distributor automatically shifts spend toward the best performers. Campaign Groups accelerate your learning cycle. Instead of manually shuffling spend around, you can let Realize do all the heavy lifting, ensuring your budget flows to the activities that are working. By letting Realize handle your budget allocation, you’re free to focus on strategy and creativity rather than babysitting your campaign. Once it’s in place, you’ll enjoy faster insights, smarter spend, and more conversions at a lower cost. Here are some examples of how you can apply this to your fashion e-commerce business: ### Creative Testing Run two campaigns. With the first, focus on lifestyle photography. With the second, feature product close-ups. Place both campaigns in the same Campaign Group. Realize will quickly start shifting spend toward the option that’s getting the best results. ### Landing Page Testing With your landing page, try pitting an advertorial against an approach that sends visitors directly to your product page. Campaign Groups will automatically optimize budget distribution. ### Product Line Testing Say you have a new collection like a swimsuit line. Pit those new, seasonal products against one of your evergreen bestsellers. Put both in the same Campaign Group and see how the new collection performs compared to your tried and true. ## 3. Master A Seamless Journey Experience Imagine a shopper sees your ad promising 70% off. Excited, they click, landing on your PDP. Only, once they arrive, they see no mention of a discount — after searching for a minute or so, they exit and never return. This is what’s known as a message mismatch. Fashion e-commerce is especially vulnerable here because trends and offers move fast. Your ad might highlight a hot discount or trending style, which can scare off impulse-driven shoppers, who have a lower tolerance for friction in the buying journey. You won’t just lose a sale, either — you’ll lose that customer’s trust. This also applies to advertorial funnels. If the ad teases a story, but the landing page jumps abruptly to a product push, users can feel misled, at which point they’ll drop off. Optimizing the on-site experience reduces bounce rates and boosts both your conversion rate and your average order value. By aligning expectations with your offerings, you can transform high-intent visitors into paying customers, increasing your average order value. Over time, this compounding revenue will lead to big gains. Here are some practical steps to ensure the user journey is seamless from first click to checkout: ### Match Ad Promises On-Site If an ad promises 70% off, make sure a banner or headline reinforces it on the landing page. This makes it clear, from the time a customer clicks, how the offer applies. ### Integrate Trust Signals at Checkout Reinforce buyer confidence where it matters most — right before they spend their hard-earned money. Adding elements like customer reviews, secure payment icons, money-back guarantees, and “free returns” messaging can give customers confidence in their decision to buy from you. “To reduce bounce rates during checkout, the on-site experience should closely mirror the ad’s promise,” reiterates Coyle. “Headlines, copy, and visuals need to stay consistent, while price and offers remain clear from entry to purchase. Trust signals like reviews, guarantees, and secure payment badges ease hesitation, and streamlined checkout steps — especially on mobile — keep users moving. When every element reinforces the ad messaging, the journey feels seamless and drop-offs decline.” ### Duplicate Stores for Testing Create a duplicate version of your store and test different product pages and/or checkouts. This can help you determine which version converts best without disrupting your live site. ### Leverage Shopify Plus If your budget allows for it, you can customize your checkout flows. Add upsells, urgency timers, and guarantees to boost conversions. ### Use Storytelling Shift your focus from promoting products to selling through storytelling. You can use advertorials or listicles to introduce your brand in a way that engages customers. This approach can help you build trust, highlight use cases, and warm up users before they see a buy button. ### Test Multiple Narratives Different audiences respond to different messaging. One user might click based on a limited-time discount, while another will take action when a product seems exclusive or boasts sustainability. Realize can help you experiment to find the angle that works best. ## 4. Set Smart Retargeting Strategies Most brands set up basic retargeting to convert previous site visitors. Those brands serve all of those potential customers the same ad that initially brought them to their site. The problem? Creative fatigue. Users who didn’t convert the first time won’t be persuaded by seeing the same message a second, third, or fourth time. In fact, some customers might find the repeat ads annoying. “The clearest indicators of creative fatigue are declining CTRs combined with rising CPCs,” says Coyle. “This usually shows that the audience has seen the ad too often and its impact is wearing off, signaling that it’s time to refresh or replace the creative.” With a smarter retargeting strategy, your ads will feel more personal and relevant. Best of all, customers won’t see the same ad multiple times, reducing annoyance and creative fatigue. This renewed approach to your retargeting efforts will likely bring higher conversion rates, improved return on ad spend, and a better overall user experience. You’ll stop wasting money on ineffective impressions and have a better chance at turning warm leads into buyers. Unlock your full retargeting strategy through a multi-layered approach: ### Tailor creatives to the stage of the user’s journey Cart abandoners are close to buying, so a discount or free shipping can push them over the edge, while checkout drop-offs can be converted through urgent offers like limited stock or countdown timers. For customers who’ve visited one of your PDPs, highlight that product’s benefits or reviews, or provide styling inspiration. ### Use fresh creatives Don’t recycle the same messaging. Instead, create ads that acknowledge previous engagement (“Still thinking about this sweater?”) and show fresh visuals. ### Link to relevant destinations Don’t send customers to your homepage or a generic section of your site. Instead, send them to the exact product or collection they viewed. ### Segment audiences Build distinct audiences for each funnel stage. One segment could hold those who viewed your PDP, for instance, while another includes those who initiated checkout or added an item to a shopping cart. ## Key Takeaways Accurate tracking is non-negotiable: Without clean data, you don’t have the insights necessary to scale with confidence. Campaign Groups make testing easier by utilizing Realize’s Budget Distributor to speed up learning while you focus on creating and implementing winning strategies. Message match is equally critical, ensuring that your advertising remains consistent from start to finish, helping you avoid drop-offs. When it comes to retargeting, it’s essential to steer clear of repeating previous messaging and instead tailor both your creative and your landing pages to each user’s stage of the funnel. ## Frequently Asked Questions (FAQs) ### How can I ensure my conversion data is accurate across both my ad platform and my Shopify store? To make sure your conversion data is accurate, start by comparing Shopify’s sales data with Realize’s Tracking tab. This is a way to quickly identify discrepancies. Once you’ve done that, take a look at the Event Test Tool. This will help you verify that key events like purchases, add-to-cart actions, or page views are firing the way they should. Then, install the Taboola Pixel Helper Chrome extension. This will help you troubleshoot any page-level issues. Each of these steps, when combined, can give you the foundation you need to move forward with confidence in your data. ### What's the best way to test different ad creatives and landing pages to find what works for my fashion brand? The most effective testing approach is to set up different campaigns for each angle you want to test. You can then split them by device (i.e., mobile vs. desktop). By dividing your testing, you can see performance differences clearly and avoid skewed results. You should also assign a separate landing page for each campaign, and those landing pages should align closely with the ad creative. Make sure every element — including your ad copy, imagery, layout, and even the product detail page — follows the same narrative. This will ensure you’re testing one clear angle per campaign, taking it all the way from first click to purchase. You’ll also want to stay organized. Campaign groups and clear naming conventions can help you cluster your test structures. If your data shows significant behavioral differences, consider splitting that campaign further. Keep the number of ads per campaign to a minimum so ads come in cleaner and faster, and only test one major variable at a time. Isolate a headline in one round, then switch to an image with the next. Testing too many variables at once muddies the waters and makes it tough to pinpoint exactly what’s driving a campaign’s performance. ### My ads are getting a lot of clicks, but people aren't buying. How can I fix my website to turn more visitors into customers? Clicks without follow-through typically signal one of two things: #### Message Mismatch Your ad promises one thing, but the landing page doesn’t deliver. For instance, your ad may tell a dramatic story, yet when users click, they’re dropped straight onto a product page. This disconnect causes confusion and kills trust. To fix it, you’ll need to take a look at the ad and make sure the hook flows seamlessly to the landing page. The tone, narrative, and brand promise should remain consistent throughout. With all of your ads, it’s essential that the product detail page feels like the logical next step after a user clicks over. You put hard work into creating clickworthy ads, so it would be a waste for the ad to grab a user, only to have that interest evaporate instantly when the landing page doesn’t live up to expectations. #### Wrong Placement or Publisher Sometimes, the issue isn’t your website — instead, it’s where your ad is running. You might have a high click-through rate, but that doesn’t necessarily equal high intent. Maybe your ad is attracting casual clickers instead of motivated shoppers. Context matters: Native ads require a different storytelling approach than display or email ads. Always stop and ask yourself if the creative aligns with the mindset of the typical user in this particular environment. If not, no matter how much work you put into your website, you won’t see the conversions you expect. In both of the above cases, the key is alignment — between the ad, the platform, and the landing page experience. When all three work together, you’re likely to go beyond clicks to paying customers. --- ### Boost Your Lead Generation with Realize to Overcome Low Contact Form Conversions URL: https://www.taboola.com/marketing-hub/overcome-low-contact-form-conversions-realize/ Last Modified: 2025-10-15 07:48:43 Affiliate agencies and media buyers know that lead generation is both rewarding and a relentless challenge. Even with strong traffic, many teams struggle to achieve profitable, scalable growth over time. This is often due to issues at the same point in the funnel — poorly converting contact forms, missing or incomplete conversion data, or limitations in targeting that result in low-quality leads draining your budget. With one or all of these issues, maintaining performance consistency across campaigns can be even more difficult. Taboola built Realize to tackle these problems head-on, creating a performance marketing platform designed to eliminate these ongoing frustrations. The tool offers seamless contact form integration, intelligent optimization, and the backing of experienced account managers to help you stop wasting money and scale more confidently. Lead generation is more than simply looking at contact form completions. Instead, it’s about a continuous cycle of engagement, qualification, and conversion that reduces wasted ad spend and maximizes revenue. By blending data feedback with machine learning, Realize ensures that every impression works hard and allows affiliates to adapt faster in an ever-competitive landscape. ## Common Pain Points Every Affiliate Experiences with Lead Generation Campaigns ### Sub-Optimal Contact Form Conversions Regardless of how strong your creative or how effective your traffic acquisition strategy is, a poorly put-together contact form will naturally decrease your overall performance. Across industries, the average conversion rate for forms is sitting at 1.7% in 2025. This means that the vast majority of visitors to your site never actually become leads. When only 9% of people who view a contact form actually submit it, even the smallest design flaws or excess number of fields can cost affiliates significant revenue. For media buyers tasked with proving return on investment (ROI), these missed conversion opportunities represent wasted impressions, ad spend, and opportunities for optimization. Consider a campaign that drives 100,000 visitors to a landing page each month. At a 1.7% conversion rate, that would yield 1,700 leads. But, improving conversion by 1% to 2.7% could generate 1,000 more leads without spending additional money on traffic. That type of marginal gain is often the biggest difference between an unprofitable campaign and scalable success. For many advertisers, this type of gain is often impossible when efforts are focused too heavily on top-of-funnel tactics like constant creative refreshes or publisher placement. While these are, of course, important aspects to consider, Realize makes it possible to transform the point where traffic becomes value: your forms. ### Lack of Data Feedback for Optimization Failing to channel a full data picture back into your advertising platform is another serious problem that many affiliates and media buyers come across. Without closing the loop, algorithms can’t learn what a valuable conversion looks like for your business, leaving campaigns stuck at the top-of-funnel stage. From this, budgets continue to fuel clicks and impressions without the intelligence needed to improve overall lead quality. This can be incredibly costly for marketers not using automated lead generation tools. Without integration of data at every stage, potential leads remain untapped. In today’s privacy-first environment, where third-party cookie tracking is becoming more of a challenge, being able to understand exactly what data means and capturing as much as possible is essential. By leveraging first-party conversion data from forms to train algorithms like those used in Realize, you can gather and benefit from a more robust set of data points that actually make a difference in your campaigns. ### Challenges With Lead Quality and Targeting Driving the right amount of traffic for your campaign is only half of the challenge — ensuring the leads that you do receive have genuine interest in what you’re offering is where the greatest profitability lies. If forms fail to capture key demographic data such as age, income, or location, or if your platform can’t act on that data, you risk filling your sales pipeline with low-quality leads. This makes it almost impossible to filter out your unprofitable segments, leading to inflated cost-per-lead (CPL) figures and inconsistent returns across campaigns. Media buyers often find themselves scaling campaigns that look successful on paper, only to discover that the downstream value of the leads generated doesn’t justify the overall campaign spend. Granular targeting has never been more important, particularly in industries like insurance or financial services, where acquiring a lead outside of your core demographic (like age range or income) is an immediate loss for the campaign. ### Integration Complexity and Bandwidth Technical integration is one of the biggest obstacles that marketing teams face. Between different site forms, custom landing pages, and legacy systems already in place, connecting each setup to your advertising platform takes time, resources, and, in some cases, specialized knowledge. Lean affiliate teams rarely have the bandwidth to work on these projects to any great extent and, even when integrations have been completed, they’re often fragile or inconsistent. This ongoing issue creates a drain on campaigns and can result in some of the many issues already discussed — campaign performance is hindered due to unoptimized or inconsistent data. ### Inconsistent Lead Flow Performance Even once everything for a campaign is live, lead flow performance is rarely stable. A form or landing page that performed well during one flight could decline once ad fatigue sets in, or when targeting parameters change. Without a reliable framework for identifying which form variations are true winners, affiliates run the risk of being caught in inconsistency cycles. As a result, budget allocation becomes riskier, scaling is more difficult, and profitability hits a standstill. Reporting also becomes more difficult with these inconsistencies, forcing affiliates to explain week-over-week volatility to clients who are looking for and expecting steady campaign performance. ## End the Frustration with Realize Seamless Integration Taboola’s Realize platform was built to solve these exact problems, designed to generate high-quality leads through re-engaging high-intent audiences. By leveraging display advertising on the open web and optimizing contact form integrations, Realize combines technology, data, and human expertise to create a single solution for more effective lead generation. ### Seamless Integration with Contact Forms With Realize, integration is straightforward. Robust pixel capabilities work across numerous platforms like Shopify, with every form completion automatically tracked and attributed. Instead of having to hunt down lost data, agencies and media buyers can feel confident that your campaigns are backed by accurate, comprehensive data collection and reporting. Realize also reduces implementation timelines, taking what once required several weeks of developer time and achieving the same result within hours. This means that affiliates can launch faster and spend more time focusing on creative strategy and audience testing, giving you a competitive advantage. ### Algorithm-Driven Optimization via Form Data Once conversion data is recorded, Realize can learn from this information to provide greater insights for campaign optimization. The machine learning algorithms built into the tool incorporate both completed forms and micro-conversion data such as partial form fills or click-to-call actions, training on these signals to identify patterns of user behavior or find similar audiences to market to. Over time, these campaigns become more efficient, with lower cost-per-lead (CPL) and scalability that’s more achievable than without these functions. As your campaign evolves, Realize adapts to support. This is particularly helpful if you have seasonal trends that impact user behavior — the algorithm learns this and recalibrates as necessary. In terms of how much data is required to meaningfully train Realize’s optimization algorithms, “Typically, the more the merrier when it comes to signals,” says Jeremy Bade, advertising sales manager, growth, at Taboola. “However, if I can get at least two signals, I can get plenty of leads into a form. If CPA is high and we have only one signal, we don't know where the drop-off is within the form. A third event which is focused on either qualified high intent leads, or leads that converted with a monetary value attached, would be the gold standard — that way, we can optimize towards a more ‘raw’ lead to begin with, then shift the algorithm down the funnel to the most qualified users once we hit 50 qualified leads/high monetary value leads.” ### Precise Targeting Based on Form Insights Another key advantage of Realize is its ability to sharpen audience targeting with form-derived data. If analysis shows certain age brackets or demographics that consistently fail to convert (e.g., over 55s), advertisers can exclude that audience to preserve overall campaign budget. Likewise, high-performing segments of your audience can be prioritized, investing more of your ad spend into the most likely-to-succeed segments. Rather than generic top-of-funnel targeting, affiliates gain a feedback loop that continuously refines audience quality. ### In-House Expertise for Integration and Optimization While the technology behind Realize is incredibly powerful, the real differentiator is Taboola’s in-house expertise. Dedicated account managers support our clients every step of the way, from solving integration challenges to analyzing campaign lead flow performance. This human layer to every campaign ensures transparency across the board. Affiliates never have to wonder why performance metrics are changing, as account managers are on hand to provide context, analysis, and offer actionable recommendations that create confidence and build a long-term trusting partnership. By adjusting targeting and form design, clients can see significant increases in lead form completions and a drop in CPL, thanks to insight from account managers. With expert oversight, you can turn their knowledge into real revenue gains for your business. In terms of common design mistakes made on lead forms that lead to drop-offs, for example, Bade has this advice: “Think about your lead form from the point of view of an end user: If your hook gets them on the page, but your lead form seemingly adds no value to what they were just hooked by, it's likely to fail. Put yourself in the user's shoes and design a contact form that follows that train of thought. If, say, I’m working on auto insurance lead generation, I might write, ‘Drivers may be getting overcharged by these companies, find out if yours is on the list!’ But, if it then just goes straight into the form without asking, ‘What provider do you use?’ then it's likely to turn folks away, as they’ll see it as a cheap tactic to get their information.” “You have to give before you take,” Bade continues. “You shouldn’t ask for things that are too personal on the first click into a form — you need to gain trust by assuring them that you have reputable information for them. Gone are the days of, ‘Enter name, DOB, state, etc.’ right at the beginning of the lead funnel.” ### Flexible Lead Flow Support Every lead generation campaign is different and unique, which is why Realize supports multiple acquisition flows. Whether you’re driving traffic to a traditional lead form, exploring click-to-call options, or even working with a hybrid strategy, Realize can seamlessly adapt to what your business needs. This flexibility means that advertisers can continue to explore new lead generation channels, without sacrificing optimization. ### Cost-Effective Lead Acquisition Through Form Optimization Ultimately, the goal of any affiliate or media buyer is to achieve profitability through low cost-per-lead and maximum return per lead. Realize makes this possible by tying optimization directly to form completions, ensuring that algorithms work towards these intended outcomes. By lowering acquisition costs overall, while maintaining or increasing lead quality, affiliates can feel confident in reinvesting and scaling with data-driven insights that generate the results you’re looking to achieve. ## Key Takeaways Lead generation will always come with its challenges, but you don’t have to keep facing the same issues again and again. Contact forms that barely convert, algorithms trained on poor data, low-quality leads, and unpredictable results can all become pain points for even the most experienced media buyer. With Realize, these obstacles can be turned into opportunities. By ensuring seamless integration with existing tools and harnessing algorithm-driven optimization, Taboola equips advertisers to transform underperforming funnels into scalable growth engines that maximize revenue. The future of lead generation belongs to marketers who can effectively combine technology with actionable insights. Platforms like Realize not only automate the repetitive tasks that drain valuable resources, but also empower marketers to focus on strategy and long-term growth planning. Start your journey with Realize today and discover how Taboola can help you unlock new growth for your business. --- ### The Best A/B Testing Tools to Use With Realize URL: https://www.taboola.com/marketing-hub/best-ab-testing-tools-with-realize/ Last Modified: 2025-10-23 13:00:01 A/B testing is one of the essential tools for understanding your audience and continually getting better at reaching and connecting with them. It continues to be a foundational part of performance marketing, especially in these days of oversaturated digital channels and hard-to-grab user attention. A/B testing practices and platforms keep maturing, so there are many options available for marketing teams depending on your audience, goals, and resources. While Realize includes wide-ranging A/B testing features, there are other platforms on the market that, when used in conjunction with Realize, can help optimize your experience. I’ll do a deep dive here on what you need to know about A/B testing with Realize, including the capabilities Realize brings, the partners it’s integrated with, and the complementary testing tools available in the market. ## Realize’s A/B Testing Capabilities: Quick Overview ### Ad Creatives Realize lets you A/B test ad creatives by uploading combinations of headlines and images, then running multivariate tests. Set campaigns to the "A/B Testing" options to distribute traffic evenly, allocating impressions more evenly across creatives, and allow the algorithm to determine winners after the test concludes. Realize also offers an Optimized mode, which automatically prioritizes top-performing creative combinations. For more dedicated A/B testing, you’ll want to do some up-front work that follows testing best practices: ensure your ads have clear differences, use consistent naming, and test a variety of creative angles. Run the test for enough time to gather sufficient data, then analyze to see which headline and image combination performed best. Once the A/B test ends, the campaign will switch to Optimized delivery. You can scale your best-performing combinations by uploading more of them. ### Landing Page Optimization Support Optimized landing pages are the next step for conversion success after Realize’s performance ads. A/B testing can help by identifying high-performing elements that resonate with your audience and drive conversions. Implementing the winning testing variations can increase the conversion rate of landing pages, and marketers can continue this optimization cycle over time with more A/B testing. ### AI and Predictive Engines Taboola’s predictive AI capabilities can power A/B testing to optimize campaign results in real time. Realize’s generative AI, called Abby, can help advertisers create and manage campaigns, then test the different variations that Abby generates to see which performs best. SmartBid, Taboola's predictive bidding strategy, uses deep-learning AI to adjust bids for each impression. Try different campaign approaches to see how SmartBid can improve your conversion rates. While the Realize platform offers a wide selection of built-in A/B testing features, it doesn’t go as far as deep, on-site landing page experimentation. For highly granular changes, the use of visual editors, complex user flow testing on landing pages, and dedicated third-party A/B testing platforms are often necessary. ## Integrated and Partnered A/B Testing and Optimization Tools for Realize In light of the above, Realize also integrates or closely partners with other popular marketing tools. Depending on the marketing stack you’re already running and your specific goals, one of these options might best fit your needs alongside Realize. These tools complement Realize’s functionalities by enhancing data flow, attribution, automation, and audience management. They support A/B testing efforts either directly or indirectly by providing superior data for analysis, or by offering integrated landing page testing capabilities. Let’s dig in to these integrated and partner options. Here’s a quick overview to start, with more detail below: Tool Name Primary Function With Realize A/B Testing Relevance Key Benefits for Realize Marketers AnyTrack Server-side conversion tracking, data sync (CAPI), attribution. Provides accurate, real-time conversion data essential for A/B test analysis. Its Advance plan offers A/B testing features. Maximizes ROAS with precise attribution, leverages first-party data, automates retargeting. Leadpages Landing page builder. Built-in A/B split testing for landing pages. Optimizes post-click conversion rates, direct integration for digital advertising. TheOptimizer Campaign automation and optimization. Automates budget/bid adjustments based on performance data (from A/B tests or native optimization). Scales winning campaigns, blocks underperforming elements, saves time on manual optimization. RedTrack Performance marketing analytics, attribution, automation. Provides comprehensive, independent data for A/B test analysis; automation rules for optimization. Unifies data across channels, accurate conversion/revenue matching, automates actions based on insights. GrowthLoop Audience segmentation and sync. Enables A/B testing on highly targeted audience segments within Realize. Refines audience targeting, facilitates personalized campaigns, leverages first-party data. ### AnyTrack AnyTrack’s no-code solution makes it easy for content and performance marketers to do server-side tracking, transmitting real-time conversion data directly to Realize with the AnyTrack conversion API (CAPI). This means you can ensure precise conversion attribution and take advantage of first-party, cookieless data so you’re complying with privacy regulations. Plus, AnyTrack unifies conversion data across ad networks, which translates into a fuller picture of performance. You can also use AnyTrack and Realize integration for automated retargeting optimization processes. Why AnyTrack and Realize for A/B testing: Succeeding at A/B testing requires accurate, complete data to gauge what performed better. This is what AnyTrack offers — real-time, unified conversion data alongside cookieless tracking across ad networks, so marketers can trust testing results. If you’re managing complex funnels or campaigns that span a range of ad networks, this might be a good option. ### Leadpages Leadpages is a dedicated landing page builder designed for marketers running A/B testing campaigns. The A/B split testing functionality was designed with landing pages in mind, so it’s easy for marketers to create and test variations against a control page. Plus, it continuously enhances the conversion rate of digital content. While Realize drives targeted traffic and engagement with its content recommendations, Leadpages can then optimize the post-click experience with its high-converting landing pages. The result is maximized overall campaign ROI. WhyLeadPages and Realize for A/B testing: Leadpages directly integrates with Realize for digital advertising purposes, and can optimize the post-click experience for traffic generated by Realize ads. Its A/B testing capabilities for landing pages are a natural fit to accompany Realize’s optimized ads. ### TheOptimizer This automation platform manages and optimizes Realize campaigns, letting users implement advanced rules to adjust budgets automatically for high-converting campaigns, adjust bids, and pause underperforming ads, publishers, and campaigns. Users can upload CSV data files into TheOptimizer or integrate external trackers to create the right automation rules. Why TheOptimizer and Realize for A/B testing: TheOptimizer is a fit with Realize if you’re focused on managing and optimizing complex campaigns and want to add automation. When A/B tests identify the best creative or other campaign elements, TheOptimizer scales the winners and pauses the losers. You can save lots of time and effort and quickly scale the impact of top performers once testing identifies them. ### RedTrack This performance marketing analytics platform can track, measure, and manage campaigns across a variety of channels, offering conversion tracking, revenue attribution, ad spend synchronization, and CAPI integrations with ad networks that include Realize. Users also get reports, dashboards, and can set automation rules for self-optimization. The independent, unified conversion data that RedTrack provides (across 200 integrations) is important to analyze A/B test results as well as automate adjustments based on results. Why RedTrack and Realize for A/B testing: RedTrack’s automation and measurement capabilities help marketers to see attribution across channels and capture the real impact of a Realize ad, even if it’s run across multiple channels. Users get precise ROI calculations and can make solid budget allocation decisions with RedTrack’s data. ### GrowthLoop GrowthLoop’s platform covers CRM and touchpoint intelligence, with a focus on intelligently re-engaging customers and precisely segmenting audiences. Advertisers can create highly targeted campaigns with GrowthLoop to then conduct more sophisticated A/B tests, such as by tailoring offers to specific segments based on value, recent purchases, churn risk, and more. Why GrowthLoop and Realize for A/B testing: GrowthLoop can help marketers target specific, synchronized audience segments within Realize. That targeting leads to a better understanding of what different user groups respond to and thus better strategies to reach them. Users can then apply insights to hyper-personalize campaigns and maximize relevance and conversion rates. ## Leading Third-Party A/B Testing Platforms for Enhanced Realize Optimization Beyond these fully integrated or closely partnered platforms, there are other optimization options to accompany your Realize implementation. These dedicated experimentation platforms are ideal for optimizing landing pages and broader on-site user experience to which Realize ads direct traffic. These tools don’t directly execute ad creative tests within the Realize platform, but they’re critical for marketers looking to maximize the overall campaign ROI by significantly improving the post-click conversion rate. Depending on your marketing goals and resources, one of these might be a good fit for your A/B testing program. Here’s a quick overview of these complementary platforms, with lots more details below on particular features to consider. Tool Name Primary A/B Testing Focus Key Features for Realize Marketers Strengths for Realize Campaigns VWO Web, mobile, server-side experimentation. Comprehensive A/B, MVT, split URL testing; AI-driven ideas; heatmaps, session recordings. Deep optimization of landing pages and user journeys, data-driven hypothesis generation. Optimizely Digital experience and feature experimentation. Robust A/B, MVT; personalization; warehouse-native analytics. Enterprise-grade experimentation, linking tests to true business outcomes, advanced data insights. AB Tasty Experience optimization and personalization. A/B, MVT, split URL; AI-driven personalization; multi-channel testing. Holistic user journey optimization, advanced segmentation, agile experimentation. Unbounce Landing page builder and A/B testing. Integrated A/B testing for landing pages; AI copywriting; smart traffic. Rapid landing page creation and testing, easy for non-developers, focused on conversion rates. Crazy Egg Visual analytics and simple A/B testing. Heatmaps, scrollmaps, session recordings; built-in A/B testing for on-page elements. Qualitative insights to inform tests, easy visual A/B testing for quick wins, understanding user behavior. ClickFunnels Sales funnel builder and A/B testing. A/B testing for multi-step funnels (landing pages, order forms, emails). Optimizes entire conversion funnels, ideal for lead generation and sales processes driven by Realize. ### VWO (Visual Website Optimizer) VWO offers integrated testing tools along with conversion optimization capabilities, as well as qualitative analytics for users to get a better understanding of user behavior. VWO can optimize the landing pages and user journeys that Realize ad traffic is directed to, and performance marketers can use it to test headlines, body copy, calls to action (CTAs), images, forms, and overall page layouts. Why VWO and Realize for A/B testing: Realize users can add VWO to increase landing page conversions and improve the overall user journey, completing the work of a high-performing ad. Successful Realize campaigns can extend even further with holistic optimization beyond the ad platform and ensure that a good ad is only the start of a good user journey. ### Optimizely Strategic A/B testing is easier with Optimizely’s capabilities, which help marketers perform testing with consolidated customer data and then measure the impact of experiments against business goals like revenue and pipeline generation. Realize advertisers running many campaigns can see how they are contributing beyond simply conversion rate optimization (CRO). Why Optimizely and Realize for A/B testing: Traffic that originates from Realize ads can then arrive on Optimizely’s landing pages and optimized on-site experiences for better business impact. Plus, warehouse-native analytics allow marketers to link Realize-driven traffic and A/B test results to understand campaign ROI at a high level. ### AB Tasty AB Tasty offers experimentation and personalization features to help users deliver tailored customer experiences. Its testing types include A/B, split URL, and multivariate, and its incorporated AI features support the creation of more meaningful digital experiences across websites and mobile apps. Marketers using AB Tasty can personalize and optimize experiences from the native ad click to the final conversion. Why AB Tasty and Realize for A/B testing: A/B Tasty supports multi-channel A/B tests without negatively impacting performance, and its tools can optimize the entire journey for a user after they click a Realize ad. That includes broader website experiences, so consistency and personalization show up throughout. And, its multi-channel testing options can help marketers who are running integrated campaigns with Realize as one of the traffic sources. ### Unbounce Unbounce is a landing page builder with integrated A/B testing capabilities, AI-driven copywriting tools, and other features that route landing page visitors to the best-performing variants. Marketers can quickly create and A/B test multiple landing page variations without much developer support. Why Unbounce and Realize for A/B testing: Performance marketers have to iterate quickly on landing pages to align with ad creatives, market conditions, and other factors. To stay ahead, they have to continuously optimize the conversion funnel without technical blockers, and that speed can help beat the competition. Unbounce can help enhance performance from Realize campaign traffic with its quick post-click optimization. ### Crazy Egg Crazy Egg is a visual analytics and user behavior tracking tool that offers data insights as heatmaps, scrollmaps, and session recordings. Its built-in A/B testing tool lets users test changes to ad elements like headlines, images, button placements, and more, then tracks variants and delivers conversion and ROI insights. Its visual analytics show areas that are working (or not) to help marketers develop better A/B test hypotheses. Why Crazy Egg and Realize for A/B testing: Landing pages driven by Realize ads should be engaging and highly optimized; Crazy Egg can offer qualitative insights into how users are interacting with those pages. Beyond straightforward numbers, these insights show why something is happening, so that Realize campaign managers can build better tests and ultimately better user experiences. ### ClickFunnels ClickFunnels users can build sales funnels that integrate A/B testing capabilities directly into various steps to create different versions of landing pages, order forms, and other elements in a multi-step funnel. ClickFunnels provides analytics for each variant to show a complete performance view. Why ClickFunnels and Realize for A/B testing: Performance marketers using Realize may be driving traffic into multi-step sales or lead gen funnels that include opt-in forms, sales pages, upsells, or order forms, and ClickFunnels enables A/B testing at every stage. That in turn prevents conversion leaks and enables optimal conversion rates throughout the process — maximizing the value of every Realize ad click and, eventually, ROI. ## Key Takeaways Realize offers A/B testing capabilities for performance marketers out of the box. Depending on user goals and resources, there are also strong integrators, partners, and other complementary tools that can boost the impact of Realize ad campaigns. These platforms include specialists in landing page optimization, audience segmentation, and many more. ## Frequently Asked Questions (FAQs) ### What are the best practices for A/B testing? A/B testing, whatever the industry, budget, or scope, should start with a clear hypothesis and testing only one variable at a time. Make sure you’re using a big enough sample size, running tests long enough to capture data (generally a few weeks), and focusing on the right elements to test, whether it’s headlines, CTAs, images, and placement of these. Document the process and exactly what you tested to then inform your decisions about what to do differently next time. When something is performing well, or not, figure out why and then double down on that information to make the next iteration better. ### What are key metrics for monitoring A/B test performance? There are some particular marketing metrics that work well for A/B test performance monitoring, such as conversion rate, click-through rate (CTR), bounce rate, retention rate, and more. But, before creating and launching an A/B test, be clear on why you’re running it and what you hope to achieve. If it’s a sales goal like pipeline or lead gen, you might create a different test than if you’re trying to increase user engagement or brand interest. Keep an eye on metrics that track negative impact, such as page load time and other functional numbers. ### How does AI and machine learning impact the future of A/B testing? AI and machine learning (ML) are generally enhancing A/B testing tools and adjacent platforms. These capabilities bring automation, predictive analytics, and real-time optimization, adding speed and accuracy and supporting increased personalization. A/B testing will continue to serve as an essential pillar of data-driven marketing strategies, and smaller or resource-limited teams can take advantage of AI and ML to test more often and faster, and get test hypotheses and other creative ideas. --- ### Uncovering Seasonal Ad Engagement by Vertical: A Deep Dive Into Our Platform's Data URL: https://www.taboola.com/marketing-hub/peak-engagement-periods-by-vertical-realize-data/ Last Modified: 2025-11-13 12:08:42 Understanding when consumers are most engaged is a vital piece of knowledge for advertisers. It enables them to time campaigns with consumer intent, which increases clicks, maximizes ROI, and improves conversions. To help, we collected Realize data from 2022 to the present, to determine the peak engagement periods for several key verticals. The insights in this report reveal when user clicks on vertical-specific ads peak, helping you decide when to prepare, scale, or adjust your campaigns for maximum impact. ## Fitness & Exercise/Healthy Living Peak months: January, December Unsurprisingly, fitness and healthy living campaigns perform their best at the start of the year, when consumers are setting their New Year’s resolutions and trying to adopt a more health-oriented mindset. Ad clicks in this vertical surge as people look to purchase exercise equipment, join gyms and fitness programs, and commit to new eating habits. The spike in clicks doesn’t suddenly appear on January 1st, though: Engagement begins to rise in December, as consumers think ahead and begin to plan for the new year. But, it’s a narrow window: By March, interest in this vertical falls dramatically as people lose their motivation to stick with their resolutions. Our recommendation: Your best bet is to front-load your campaigns in December to capture those early planners. Then, double down in January when consumers are at their peak engagement. In other words, prepare to scale your budget before cutting back in March. ## Television Peak months: February, March, and the fall months Television engagement trends align with seasonal programming and significant cultural events. For example, fall is a key period because it’s when many TV programs premiere their new seasons and streaming platforms offer new releases. That said, some of the biggest audience spikes happen earlier in the year, in February and March, when nominations for the Oscars, Golden Globes, and Emmys drive audiences to search for shows, actors, and other related content. Our recommendation: Align your ad campaigns with these peak seasonal periods. Fall will offer more consistent opportunities, but prepare for concentrated bursts of ad spend early in the year. ## Automotive Peak months: March, April, October, November The automotive vertical sees very consistent surges during spring and fall. In March and April, consumers, armed with cash from their tax refunds, may be feeling confident about making big-ticket purchases. Seasonal promotions during this period don’t hurt, either. Vehicle searches spike again in October and November. This is due to several factors, including new model announcements from manufacturers, pre-winter car preparation for many drivers, and dealership incentives designed to clear inventory. It’s worth noting that ad competition peaks in April and May, which reflects the heightened interest during the spring season. Our recommendation: When creating your annual strategy, plan automotive ad campaigns around the spring and fall peaks. Consider the intent driving consumer demand and optimize your ad creative and budgets accordingly. ## Business Peak months: April, May Business-related ad clicks spike in April and May, which aligns with Q2 business launches and spring marketing budgets, as well as the rollout of business strategies set during Q1. Clicks and ad competition remain high throughout the fall, demonstrating sustained demand for B2B solutions, tools, and services for most of the year. Our recommendation: Advertisers in the business vertical should consider spring as the prime time to launch campaigns and plan to maintain momentum into Q3 and Q4 to capitalize on ongoing demand. ## Shopping/Style & Fashion Peak months: November, December, May, June, September, October Not many verticals are as cyclical as shopping and fashion. The largest spikes occur in November and December, driven by shopping holidays like Black Friday and Cyber Monday, as well as holiday gift-buying, which fuel enormous volumes of clicks. There are smaller peaks in May and June, due to Mother’s Day, Father’s Day, graduations, and the start of summer sales. And, don’t forget September and October, as they are a key period for advertisers looking to ramp up holiday preparations. Our recommendation: Anchor your strategy around the holiday season, making sure that you test your campaigns early and warm up audiences who have started browsing. Don’t ignore smaller yet significant mid-year opportunities. ## Hobbies/Interests Peak months: January, May, June, July January brings a consistent spike in engagement around hobbies and personal interests. This may be due to a mix of seasonal and motivational drivers — for example, people often set resolutions to start new personal projects, like crafts or music. Another spike comes when the warmer weather hits, with increased activity in May, June, and July. During this time, consumers are taking up summer hobbies and outdoor activities. Our recommendation: Take advantage of the January “restart” and the summer surge, and tailor your messaging to resolutions in Q1 and lifestyle pursuits in Q2 and Q3. ## Travel Peak months: March, April, June, July Our data show that travel ad clicks increase steadily from March through June and July. This aligns with summer vacation planning. Similarly, like the trend seen in the automotive industry, many consumers are looking for ways to spend their tax refunds. Our recommendation: Begin ramping up ad spending in early spring to capture consumers' attention during their peak travel planning season. Ensure you sustain your investment through the summer to reach consumers right up to the point of booking. ## Home & Garden Peak months: April, May, June, July Unsurprisingly, the spring and summer seasons are synonymous with home and garden projects. From April to May, consumers spend time on spring cleaning, home improvements, and gardening. Our recommendation: Align your campaigns with these seasonal behaviors. Spring is the kickoff, but summer provides additional opportunities. ## Alcoholic Beverages Peak months: November (extends over the holiday season) Ad clicks for alcoholic beverages spike in November, likely due to holiday drinking during Thanksgiving, Christmas, and New Year’s celebrations. While the shopping window extends into December, you want to get ahead of consumer demand in November to maximize the increase in sales. Our recommendation: Consider November your core advertising season, but make sure campaigns continue into December to maximize the entire holiday season. ## Personal Finance Peak months: January, April The personal finance vertical is predictable, with consumers particularly interested in budgeting, savings, and debt repayment at the start of the year, and tax software and preparation around the April tax-filing deadline. Our recommendation: Focus heavily on these two seasonal peaks. During January, position your messaging around budget and savings tools. In April, lean into tax assistance and compliance products, like tax software. ## Key Takeaways Now that you have insight into peak engagement periods, you can maximize ROI by aligning your campaigns with predictable seasonal peaks across several verticals. For example, fitness and personal finance surge in January, while the spring tax season also fuels finance and business engagement. Shopping, fashion, and alcoholic beverages see their strongest spikes during the holiday season, while consumers are interested in travel, automotive, and home and garden during the spring and summer seasons. By aligning your ad spend and creative strategy with these specific windows, you can capture consumer intent at its highest point. --- ### How to Optimize Cyber Monday Ad Performance With Realize URL: https://www.taboola.com/marketing-hub/optimize-cyber-monday-ads-with-realize/ Last Modified: 2025-10-15 06:39:23 Cyber Monday is one of the biggest shopping days of the year, which makes it particularly competitive for advertisers. In the days leading up to Cyber Monday, shoppers are bombarded by splashy ads offering deep discounts, so you really have to fight to get their attention. Realize can help your campaign succeed with advanced targeting, predictive audiences, and automated bidding strategies. To help you get the most out of Realize this Cyber Monday, I’ve put together some expert insight on how to optimize your campaign using Realize. ## Use Realize to Optimize Your Cyber Monday Campaigns, With These 6 Expert Tips ### Focus on the Right Metrics Cyber Monday campaigns move fast, and with so much data available, it’s easy for advertisers to become overwhelmed. That’s why it’s crucial to focus on the metrics that will give you the best chance of success. According to Katherine Pickles, product marketing director at Taboola, four metrics stand out: click-through rate (CTR), cost per click (CPC), cost per acquisition (CPA), and return on ad spend (ROAS). Your CTR indicates how engaging your ads are. A high CTR is a good signal that your creatives, headlines, and placements are resonating with shoppers. On the other hand, a low CTR is a sign that you may need to make changes to one or more ad elements. CPC tells you how much you’re paying for each click and is a good measure of efficiency. However, on its own, CPC doesn’t tell the whole story. CPA takes things further because it connects cost to outcomes and shows you how much you’re paying for each conversion. Finally, ROAS ties it all back to revenue, which lets you see whether your campaign is actually profitable. ### Leverage Audience Targeting Capabilities One of the biggest mistakes marketers make during the Cyber 5 period is trying to reach as many people as possible. Reach is great, but if you’re not taking advantage of Realize’s audience targeting capabilities, you risk not connecting with the shoppers who are most likely to convert. Pickles advises marketers to “focus on high-intent, 1st-party audiences, mail domain, and search keyword targeting.” And, of course, don’t forget to retarget. Your first-party audiences are people who have already interacted with your brand — they provide a high opportunity for engagement because they’re already familiar with your products. Next, expand beyond first-party through mail domain and search keyword targeting. Both of these allow you to reach consumers who have taken some action that shows purchase intent. ### Run Predictive Audiences to Boost Conversions Realize offers Predictive Audiences, which, per Pickles, “helps advertisers discover untapped, high-converting customers so they can meet their performance marketing goals at scale.” Predictive Audiences uses machine learning to identify users who are likely to take action, at scale. It does this using first-party data, which reduces your dependence on third-party cookies. Predictive Audiences can also be more effective than lookalike models, because it allows you to assess your ad performance and make adjustments in real-time. Pickles believes it’s a more dynamic approach, and a “great tactic to have running on your campaigns to get more conversions for your budget.” ### Harness Performance AI Realize has made enhancements to its Performance AI features, making it easier for marketers to drive measurable results throughout the funnel. For example, Optimize for Engagement enables you to easily define custom engagement metrics, such as time on site or session depth. From there, Realize automatically optimizes to find the right users for the criteria you’ve set. “When using Maximize Conversions, Realize's performance AI ensures that you’re maximizing conversions within your given budget, regardless of the CPC,” says Pickles. “This will ensure depletion during the highly competitive period.” ### Launch Early With a Staggered Budget Strategy Too many marketers wait until Cyber Monday to activate their ad campaigns, but by doing this, you’ve missed any opportunity to build momentum. “Launch your campaign using the Maximize Conversions bidding strategy at least five to seven days before peak days, with appropriate targeting and a daily budget up to 50% less than your target peak budget,” recommends Pickles. By launching early, you can also identify and fix potential issues before Cyber Monday, rather than scrambling to troubleshoot problems during the peak shopping period. ### Set Adequate Budgets A good ad budget requires planning. During Cyber Monday, there is typically a massive increase in consumer demand, which can make it tempting to boost your budget at the last second. This, however, can backfire, and it’s better to take a strategic approach to avoid wasted ad spend. Pickles advises you to hold off on boosting your budget until you’ve reached 50 conversions: “Once 50 conversions are reached, or it’s two to three days before peak days, increase your budget closer to the target amount, but no more than 50% at a time, to protect the CPA.” It’s a gradual approach that gives Realize’s Performance AI time to make adjustments without hurting your performance. This will allow you to scale while protecting your CPA. ## Key Takeaways Realize can help you succeed on Cyber Monday through precise targeting and intelligent automation, but you need to do your part by planning early and using the tools at your disposal. To avoid getting overwhelmed, make sure to focus on the most effective performance metrics and take advantage of Predictive Audiences. Don’t forget to take advantage of features like Maximize Conversions, which simplifies the bidding process. Finally, remember to launch at least five to seven days early, and scale your budget carefully after getting at least 50 conversions, to maximize your campaign’s visibility and profit. ## Frequently Asked Questions (FAQs) ### How can I best utilize Realize's "Predictive Audiences" to identify and target high-converting customer segments for my specific product/service? The best approach is to run Predictive Audiences in conjunction with your standard targeting. Realize’s machine learning can identify new customer segments that you may not have considered. This may result in you gaining additional conversions while you continue to focus on audiences you already know well. ### How can I effectively "pre-qualify" clicks through my headlines and thumbnails on Taboola to ensure users have a clear expectation of the landing page content? Combine headlines that clearly communicate the value of your offer, visuals that align with your product or promotion, and messaging that makes it clear what the reader will see after they click. This should improve your conversion rates and reduce wasted ad spend. ### When should I use "Maximize Conversions" versus "Enhanced CPC" or "Fixed Bid" strategies within Realize for different campaign objectives? The best strategy depends on what you’re trying to achieve. For example, “Maximize Conversions” is often the best choice during Cyber Monday and other peak holiday shopping periods. Want to balance conversions with cost control? If so, “Enhanced CPC” can be very effective. Finally, a “Fixed Bid” strategy is best for A/B testing or situations where you want to maximize control over CPC. That said, it’s not as flexible during peak shopping times like Cyber Monday. --- ### 6 Best Black Friday Promotion Ideas for 2025 URL: https://www.taboola.com/marketing-hub/best-black-friday-promotion-ideas/ Last Modified: 2025-10-15 06:23:27 As an advertising copywriter who's been in the trenches for more than a few holiday seasons, I've learned a thing or two about cutting through the noise (and there’s a lot of it to cut through). It’s not just about slapping a "sale" sticker on everything: Black Friday nowadays is more about crafting a strategy that genuinely resonates with shoppers and drives conversions. Here are some insights on how to make your products shine when the biggest shopping day of the year rolls around — and it starts earlier than you might think. ## Best Ways to Promote Your Products on Black Friday Black Friday isn't just a day anymore, it's an entire phenomenon. This is the unofficial kickoff to the holiday shopping season, a high-traffic period where consumers are actively looking for deals. To truly get the most out of it, you can’t just react: You need to anticipate and strategize before it starts. ### 1. Start Early My first, and maybe most important piece of advice: Don't wait until Thanksgiving week to launch your big push. Consumers are becoming increasingly proactive, with nearly half of holiday shoppers planning to start before November. I’m one of those shoppers, too: Holiday gift-buying is stressful and time-consuming, and I want to check it off my list as early as possible. This means you need to be visible when people start their research and initial purchases. I’d advise you to begin your campaigns early, with a clear core strategy, at least five to seven days before the peak shopping days. This allows your campaigns to learn and optimize and, ultimately, hit your goals. ### 2. Embrace the Power of Mobile (and Video!) Think about how you shop. As in, you, personally. Chances are, a good chunk of it is on your phone. It's no different for your customers: Mobile accounts for 55% of holiday e-commerce, and mobile holiday revenue now outpaces desktop. All this to say, your ads absolutely need to be mobile-first. While you're at it, don't shy away from video, either — holiday-themed creatives, especially video, can boost conversion rates significantly, sometimes with up to 4.4x higher CVR than generic creatives. A quick, looping HD motion creative can grab attention far more effectively than a static image on a crowded feed. ### 3. Leverage AI for Creative Variations and Audience Targeting The days of manually tweaking every ad variant are over, and honestly, they will not be missed. AI-driven creative variations are a game-changer and you should absolutely use them. From my perspective, this isn't just a cool tech fad of the moment — it's essential for optimizing performance at scale. Imagine being able to test dozens, even hundreds, of different headlines and images automatically to see what resonates best with different segments of your audience. That’s what leveraging AI can do. This is also where a platform like Realize comes into play. Realize is built as a performance engine, specifically designed to drive prospecting and conversion outcomes on a cost-per-click (CPC) model. It taps into unique data and AI to help you reach high-value shoppers that go beyond your core audience through Predictive Audiences. This means you're not just guessing, but using high-level methods to find people most likely to convert during that crucial peak season. It’s essential if you’re looking to boost your reach and conversions. ### 4. Offer Compelling Deals — Don't Just Discount Of course, everyone's looking for deals on Black Friday, with about three out of four holiday shoppers on the hunt, especially when it comes to Gen Z. But, simply slashing prices isn't really a strategy. Think about value propositions, bundles, limited-time offers, or even exclusive access. What makes your deal irresistible beyond just the percentage off? Consider how you frame these offers in your copy. Instead of, "25% Off," try something more like, "Unlock Massive Savings," or, "Your Holiday Haul Starts Here." Don’t be afraid to have fun and get clever with it, too — a unique bit of funny or smart copy is a breath of fresh air for customers who’ve been looking at offers all day. ### 5. Monitor Performance and Avoid Disruptions During Peak Once your campaigns are live and performing, aim for stability and continuous optimization, and try to avoid major changes during these peak days. Tinkering with targeting or bidding settings during this critical window can actually re-trigger the learning phase for your campaigns, potentially hurting performance. Instead, focus on monitoring your creative. Keep four to six creatives in rotation, and be ready to pause any underperforming ones quickly. It’s about fine-tuning, not overhauling. ### 6. Don't Forget the Post-Purchase Experience Black Friday is about acquisition, but the holiday season is about retention, so ensure your post-purchase messaging is on point: Thank them, offer complementary products, and pave the way for future engagement. A good Black Friday haul sets the stage for a loyal customer base throughout the year. ## Key Takeaways Begin your Black Friday promotional efforts well before November to capture early shoppers. Optimize all creatives and campaigns for mobile, leverage high-quality video, and utilize AI-driven solutions to optimize creative variations and target high-value audiences effectively. Maintain campaign stability during peak times, focusing on creative performance optimization rather than major structural changes. ## Frequently Asked Questions (FAQs) ### What is Black Friday busiest for? Black Friday is primarily busiest for retail foot traffic, and is still the most popular day for both in-store and online shopping. Specifically, consumer electronics and cell phone stores often see their traffic peak. It’s also emerged as the top digital shopping day overall. ### What is the biggest sale after Black Friday? Cyber Monday is widely considered the biggest sale event immediately following Black Friday. It began as an online counterpart to Black Friday's in-store focus, but has since grown to be equally, if not more, significant in terms of online sales. Other notable post-Black Friday sales events globally include Singles Day (China's largest shopping day), Small Business Saturday, and Green Monday. ### How much money does the average shopper spend on Black Friday? In 2024, the average Black Friday shopper was estimated to spend approximately $674. During the broader Cyber Week period (Thanksgiving through Cyber Monday), global online shoppers spent a collective $314.9 billion. For U.S. shoppers specifically, online spending reached $10.8 billion on Black Friday and $13.3 billion on Cyber Monday in 2024. --- ### From Frustration to Success: Recover Abandoned Carts with Realize Retargeting Solutions URL: https://www.taboola.com/marketing-hub/realize-retargeting-abandoned-carts/ Last Modified: 2026-01-20 12:18:38 For online retailers and e-commerce businesses, few challenges are as persistent and costly as an abandoned cart. Shoppers have made their way to your site, compared products, and even added items to their cart — all before leaving instead of completing their checkout. Across all industries, from fashion and electronics to groceries and subscription services, abandoned carts are one of the most significant leakages in the sales funnel. On average, around 70% of shopping carts are abandoned on sites worldwide, representing billions in lost revenue for businesses every year. Even for brands investing heavily in user experience and checkout optimization, the challenge remains — how do you win back high-intent shoppers before they move on to the competition? Funnel optimization is no longer about simply driving traffic to the top of your sales funnel or building brand awareness: It’s about protecting the revenue you’ve already worked hard to attract and ensuring that potential customers with clear purchasing signals don’t fall through the cracks. Rising acquisition costs across all digital channels means that ignoring abandoned carts is not only inefficient, but unsustainable. Jeremy Bade, Advertising Sales Manager at Taboola, gives us recommendations to deliver retargeting precision and diversified outreach with Realize, that’s specifically designed to recover these lost sales. ## Common E-Commerce Pain Points Around Abandoned Carts ### Sales Funnel Leakage (Abandoned Carts) Cart abandonment is one of the clearest symptoms of sales funnel leakage. Each time a shopper adds an item to their cart, they’re signaling a purchase intent. But, when that intent isn’t converted into revenue, it highlights clear gaps in the sales funnel experience for that user. Some of these gaps are more functionally based: extra costs at checkout, long forms to fill out, or unexpected issues with their payment method. Others are more psychological: hesitation, decision fatigue, distractions, or the instinct to compare prices elsewhere. Whatever the reason may be, abandoned carts represent one of the most critical leakages in digital commerce. Brands may be aggressively spending at the top of the funnel to drive more traffic, but without plugging these bottom-of-the-funnel gaps, valuable sales opportunities are escaping. Protecting your business revenue by proactively addressing abandoned cart issues is essential. In many ways, cart abandonment is a sign that businesses are strong at awareness but weak at conversion. Until the data is analyzed, it’s difficult to know what’s causing this drop-off — it could be an issue between product view and being added to the cart, friction during the checkout process, or hesitation around delivery times or other concerns a user might have. When scaled across thousands of shoppers, these micro moments can lead to a significant leakage issue. This is particularly the case on mobile devices. With more than 50% of online shopping now taking place on mobile devices, distractions are something that all retailers need to factor into their checkout optimization. Push notifications, social media alerts, or other distractions can make shoppers abandon their experience with your brand and forget to return. Proactive re-engagement is essential to bring them back quickly to complete the purchase. ### Stagnant Performance and Rising Costs When abandoned carts aren’t re-engaged, the impact can be felt throughout the whole marketing ecosystem. Brands are left overspending on acquisition while struggling to improve on return on ad spend (ROAS), while conversion rates stagnate and cost per acquisition (CPA) continues to climb. As a result, marketing teams are often forced to justify their budgetary spending without demonstrating any real growth. With the average CPA in e-commerce coming in around $60 in 2025, costs are certainly rising for both search and display advertising. Abandoned carts only add to this overall marketing expenditure as revenue continues to be lost, making it almost impossible to hit sales targets across any digital channel. What makes this problem even more pressing is the gap between acquisition and retention of customers. Businesses paying more to acquire customers while simultaneously losing efficiency at the bottom of the funnel creates dual pressure that erodes profitability, making leadership less likely to increase budgets without proof of revenue growth. Stagnant performance not only wastes marketing budget, it also impacts morale on the team. When strategies show little to no growth, experimentation declines and risk-taking decreases to favor “safe” but not always profitable tactics instead. ### Lack of Diversification in Re-Engagement Many businesses rely on a small set of re-engagement channels, like email campaigns or social retargeting. While these are effective, these channels are also prone to saturation. As a result, audiences tune out, ad fatigue sets in, and incremental conversion decline is likely. Without diversifying re-engagement strategies, businesses are leaving significant digital ground uncovered. Consumers today browse across a wide-scale open web, from new sites to niche publications. Limited retargeting means that many brands are missing out on the environments where users are spending a large amount of time. This over-reliance on certain platforms also makes campaigns vulnerable to changes. Algorithm updates, privacy restrictions, and even temporary downtime all disrupt campaigns focused around a single channel. By spreading retargeting efforts across multiple platforms, brands can build greater resilience against these possible disruptions. Not only that, but diversification is a significant boost for brand awareness in the user experience. When a consumer sees a brand repeatedly mentioned across different platforms, it starts to reinforce recognition without triggering fatigue. If the message feels both fresh and relevant, users are more likely to re-engage. ### Untapped Content Potential for Re-Engagement E-commerce marketing teams often invest in high-quality content like blogs, buying guides, comparison pieces, and editorial storytelling to help build trust and authority. Yet, in many cases, these pieces of content are heavily under-used in direct performance campaigns. By leveraging existing content for re-engagement, brands can address consumer hesitations that may stop someone from completing their purchase. Whether it’s a guide explaining how to find the right size for you or a blog post discussing durability of a product, relevant content can be the perfect bridge between interest and action. Using existing content is also considered a softer path back to checkout than hitting users with a “Buy Now” retargeting ad campaign. Instead, brands can position themselves as helpful advisors with guidance that resonates with consumers, rather than pressuring them into a purchase. This approach also opens the door for personalization. The same shopper who abandoned a cart with a fitness product could be re-engaged with a blog post about training tips. This type of tailored experience deepens user engagement and increases the chance of conversion. ### Competitive Pressure The urgency to retarget your abandoned cart users is amplified when considering the competitive nature of online retail. Every abandoned cart isn’t just a missed opportunity for you, but a potential win for your competitor. In industries with aggressive discounting or highly substitutable products, shoppers can easily abandon a cart on your site and be somewhere else in minutes. For businesses actively looking to grow their market share, recovering abandoned carts is more than a tactical fix — it’s a defensive strategy against your competitors by investing in customer re-acquisition that translates into measurable outcomes and growing revenue. Shoppers these days have more distractions, more options, and more incentives to abandon brand loyalty for better pricing or availability. This means that every moment of hesitation is potentially dangerous for any online retailer. The brands that win aren’t simply those with the biggest budgets: Having smart recovery strategies in place and acting quickly to re-engage cart abandoners reduces leakage and prevents competitors from stealing the sale. ## Six Ways Realize Can Recover Abandoned Carts ### 1. Precision Retargeting for Abandoned Carts Realize begins with precision. By integrating our pixel across your site, Realize identifies users who have demonstrated strong intent, such as adding items to their cart or starting the checkout process. These users form the highest-value retargeting audience, the segment most likely to convert with the right nudge. Instead of spreading ad spend across broad marketing lists, Realize ensures that investment is focused on the audience most likely to deliver immediate returns. It helps you transform abandoned carts into recovered sales and maximize your campaign efficiency across all platforms. It’s this kind of approach that allows Realize users to compete against larger retailers with bigger budgets. “Retargeting on Taboola allows for buffering your CPA, while giving your other campaigns time to learn effectively,” says Jeremy Bade. “Once your main campaigns have properly learned and have allowed for some scale, it's easy to find pockets of success within your own ecosystem, without having to worry about what the larger brands are doing. If you're hitting your goals, you don't have to be the largest, you just get to scale as fast or as slow as you'd like!” ### 2. Diversified Reach for Re-Engagement What sets Realize apart is its scale. With access to more than 9,000 premium publisher sites such as Yahoo, Apple News, and MSN, Realize lets your business reconnect with abandoned cart users across billions of impressions in the spaces where they already consume content. This diversification not only means that your ads appear outside of saturated social feeds — it allows users to see your brand in trusted editorial environments. With this visibility in multiple contexts, Realize combats ad fatigue, extends reach, and increases the likelihood of recapturing high-intent audiences. “Our creative shop team has given guidelines for different creatives types based on targeting parameters, which helps with ad fatigue,” says Bade. “Retargeting is low-hanging fruit, so creating assets specific to retargeting is a great way to call out customers even more clearly.” Some marketers may worry that diversifying their re-engagement channels might overextend their budget or spread their campaigns too thin, which is why Bade recommends creating the audience in Taboola, using the traffic generated there. “The number of site visitors will increase due to the amount of clicks,” he explains. “I always wait about two weeks prior to starting retargeting. Just retargeting every channel with no plan can spread you thin, so you have to look at your engagement metrics — like site visits and add to carts — prior to assessing which channels make the most sense to retarget customers.” ### 3. Content-Driven Re-Engagement One of Taboola’s greatest strengths is its “advertorial” format, which allows your business to re-engage users by promoting valuable content like blog posts or guides. The ability to blend content and performance through Realize means that you can win back users with both direct product advertising, and valuable, informative content simultaneously as an alternative entry point back to your site. This is an effective strategy in high-consideration categories where shoppers want reassurance before completing a purchase. Rather than pushing a direct offer, you can use content to educate, build trust, and reduce friction for shoppers. By delivering this value first, you position your brand as a partner in the decision-making process, increasing the chance of a user returning to their cart and converting. ### 4. Intelligent Algorithm Optimization for Conversions Realize goes beyond static targeting with advanced algorithmic optimization. Campaigns begin by learning from immediate signals received once the pixel is installed, such as “add to cart” or “start checkout,” and gradually evolve to optimizing campaigns for completed purchases. Over time, the Realize algorithm identifies placements, audiences, and formats that will deliver the best outcomes for your campaign. This intelligent learning loop not only drives higher conversions overall, but also improves cost efficiency across digital campaigns. ### 5. Flexible Ad Formats and Geo-Targeting for Retargeting In performance marketing, creative flexibility is key. With Realize, you can deploy ads in a range of formats such as static images, motion ads, or videos to capture user attention. Combined with geo-targeting features that segment users into location-based groups, you can flexibly tailor your campaigns to specific markets, focusing on regions with the highest potential value. ### 6. Dedicated Support and Strategic Partnership Cutting-edge technology is only part of the equation. Realize is backed by Taboola’s team of experienced account managers, who bring hands-on experience to every campaign. From diagnosing performance challenges to refining your creative strategies, your dedicated account manager ensures that campaigns remain aligned with business objectives and goals. For performance marketers, Realize is more than another marketing automation platform: It’s a strategic choice that comes with a dedicated partner who understands the nuances of retargeting and how to maximize the value of every impression. ## Key Takeaways Abandoned carts are one of the largest leakage areas in the e-commerce sales funnel, but Taboola’s Realize tool provides the support you need to bridge the gap between hesitating shoppers and a sale. In today’s competitive market, abandoned cart recovery is not only about improving metrics, but defending your existing market share and protecting profitability. If your team is ready to recover lost revenue, reduce CPA, and turn your intent into action, contact Taboola today to see how Realize can transform your abandoned cart strategy into a driver of profitable growth. --- ### Optimizing the Lead Funnel for Conversion at Scale: Insights from Xevio URL: https://www.taboola.com/marketing-hub/lead-funnel-optimization/ Last Modified: 2025-09-01 09:24:35 In the third and final part of our series on lead generation with Xevio co-founder and CEO, Nadim Kuttab, we’re looking at how to approach end-to-end optimization of your lead gen funnel, from initial ad impression to final conversion. This involves a more holistic view of the customer journey and financial outcomes, rather than isolated campaign metrics like cost-per-lead: Think not just tactical optimization, but strategic business growth. When you talk about building a funnel from scratch for lead generation, what are the foundational steps, and how do you define success at each stage? I’d give the same advice as when we talked about content strategies for high-quality lead generation — having a clickable trigger question followed by a couple more concise clickable things, then asking for personal data. How do you approach A/B testing across the entire lead funnel, from ad creative to landing pages, to continuously improve conversion rates at scale? We use a tool called Voluum — it's a native tracking tool. It can be used for other things, but it's built for native and we're constantly rotating in tests. At any given point in time, we might have between two and five different pieces of content live per campaign and one to three different lead forms per lead gen campaign. We're constantly trying to see what thread of ad, publisher, content piece, and lead form is driving the best ROAS for us and for our clients. So, how do I approach AB testing? We test everything all the time — that's it. If you find something that works, scale it while on a small budget and try to find the next big thing. That's essentially our testing philosophy: Always testing, but not blindly. I see a lot of companies just 50/50 splitting their traffic between different pages at random. Do it systematically: Have 90% or 95% of your traffic go to things that have been proven to work, but always have a little bit of traffic that explores newer things. Doing it at an even mix is a terrible idea unless you're starting from scratch. You have to buy data to see what works. As you know, Realize offers tools like Pacing Health Score and Custom Rules. How can performance advertisers leverage these features to optimize their lead generation campaigns efficiently? Custom rules — we call them parachute rules — are always a good idea. If you’re just one person, or a person who likes to have their days off or their evenings to themselves without their laptop open, set custom rules that essentially catch your campaigns when they fail. Say, if CPA goes above 2x your target and spend is above X a day, it should stop. That limits your losses when tracking breaks, or when there's some sort of tech issue. Yesterday or the day before, there was a huge outage from one of the bigger trackers out there. Three hours of data, wiped out! If you didn't react, those three hours of clicks were sent into the Nether. Realize's Custom Rules to catch the campaign when that happens would have saved you money and time. What are common bottlenecks or drop-off points you've identified in lead generation funnels, and what are your go-to strategies for fixing them? Using a crappy form, or a form that comes from Google or Meta and not adapting it to native or content-based marketing in general, can really halve your efficiency. It's unbelievable how much money can get lost with a bad form. A lot of people use unoptimized forms, or try to build their own rather than using off-the-shelf solutions like HeyFlow that cost $100 or $200 a month, but just work. There's a lot of firepower behind those servers — they’ll last if you scale up, right? Trying to save $5 in hosting costs by building your own form will cost you dearly if things go wrong. For another bottleneck, the first question on your form is a huge one. Don't ask complex first questions, I cannot stress this enough. As you advised when we discussed content strategies for high-quality lead generation, the form should be mostly clickable rather than typable. Exactly. Our go-to strategies are to look at where drop off rates are bad, then split test different things to see if we can improve them. Sometimes it's actually not a bad thing if your drop off rate is higher at the beginning, because then it tends to taper down towards the end and you might have a better conversion rate overall on the form if you filter people a bit more aggressively. But, again, it's so scenario-specific that it really depends on the product, where you're running it, how you're running it, etc. I'd say look at the data, optimize based off of the data, but don't be too obsessed with trying to squeeze 5% more out of the first question because, at the end of the day, it might actually be counterproductive. How do you ensure that scaling lead generation efforts doesn't compromise lead quality or significantly increase CPA? There is always going to be a plateau. At some point, it’s not worth putting a dollar more in because your efficiency decreases. That plateau can be $5,000 a day, it can be $100,000. It’s never $500 a day. If you’re seeing it plateau that low, there is something wrong with your funnel or your media buying strategy. Beyond immediate conversions, how do you factor in metrics like Lifetime Value (LTV) and Return on Ad Spend (ROAS) when optimizing lead generation funnels? As I mentioned when we talked about AI-powered lead generation, any data we can get back from the call centers or anything else that happens post-lead is massively valuable in helping us squeeze out more from the campaigns. The quicker we’re able to get the data into the click level analysis, the better we are at managing traffic that converts. One last thing on that topic: It really depends on the client. Some clients have built these huge models, but they're built on Google, so when you apply them in native, they fail. I've seen this time and time again. The best way to do it is start slow, spend a couple grand, see how it converts. If it shows promise, continue spending more. Don't 100,000x your budget overnight. With lead gen, it's a slow but steady burn: With e-commerce, you can be very aggressive, but with lead gen, you have to be a bit slower because there are so many factors in extracting value out of a lead. The initial metrics might look good, but then the revenue is zero — that’s not good if you're spending $100,000 a day. What role does post-conversion follow-up or nurturing play in maximizing the value of generated leads, and how does that influence your initial funnel design? You lose 20% of a lead’s value if you call them after 12 minutes versus after 30 seconds. You start losing 20% more if you take from 12 minutes to 30 minutes. So, you’ve lost 40% of a lead’s value if you don’t call them within 30 minutes. That’s wild! So, obviously, a quick follow-up (or AI validation, or whatever) is incredibly important for keeping people warm. They’ve just filled out a form, they don't want to wait three days to hear back from you, because the truth is, most likely you won’t be the only form they filled out, and they will not remember who the hell is calling when you call them three days later. For advertisers looking to scale their lead generation on the open web, what's the single most important piece of advice you can offer for funnel optimization? With lead generation, you should be very fast at optimizing. You have a relatively low CPA compared to e-commerce or higher ticket items, so start slow and increase gradually as you get more confident in the data that you're seeing. I've seen a lot of lead gen people come in with very big budgets and then realize, “Oh crap, we're not on Google. We're not on Meta. This is a different channel, we need to approach it differently.” We've been able to compete with Google and beat Meta almost consistently across our portfolio of lead gen projects — Realize offers unmatched quality if done right, but it is a very slow burn. I’ve yet to see a lead gen project that’s ticked all the boxes right off the bat. There's always something that has to be optimized, whether it's lead quality, price, sources, content, whatever. There's something that needs to be fixed along the way. With e-commerce, we've had campaigns that have launched overnight and done well, but that never happens with lead gen, ever. It's always going to take longer because there's so many pieces involved. So, just hold steady! --- ### Discover When Your Users Are Most Likely to Engage With Holiday Ads from Realize Data URL: https://www.taboola.com/marketing-hub/realize-data-backed-holiday-peaks/ Last Modified: 2025-09-16 07:36:54 When it comes to holiday advertising, timing is everything, especially when you’re planning your campaigns and budgets for maximum return. Advertising and marketing teams have to carefully weigh when to start advertising ahead of a specific holiday or set of holidays and gauge which time periods are likely to be most profitable. It’s a balancing act between starting ads too early versus too late, and giving each holiday the right amount of effort. To succeed at holiday advertising, then, analyze your own data, such as looking at owned traffic trends to understand user behavior on your site. You can also use Realize data that examines peaks and declines around the holidays and tie that information back to your own digital timelines — you’ll find the latest data represented in the graphs below, spanning 2023 to 2025 and reflecting the relative traffic peaks per holiday shopping event, as measured by ad clicks. Check out which shopping days capture the most engagement, and when their particular peaks are, so you can plan accordingly. ## How to Plan for Holiday Shopping Engagement Throughout the Year Generally, marketers should prioritize ad spend and campaign development around high-traffic holidays. Digital holiday events include the big ones — Black Friday, Cyber Monday, and Christmas — as well as a range of smaller holidays throughout the year that present opportunities for retail marketers. Make sure to front-load campaigns, starting promotion at least two to four weeks ahead of major holidays, creating urgency and exclusivity around top-performing holidays to capitalize on intent. You can get even more specific on when to start advertising to capture user attention with our trends data. Here’s what Realize showed us for the top holidays and their traffic peaks as compared to Black Friday, the busiest shopping day of the year: NOTE: The following distribution represents the relative click trends observed across the dataset’s diverse ads based on holiday event. With Black Friday as the top shopping day (~4 times more clicks than Cyber Monday), these other holidays show the relative potential peaks for shoppers, and the best times for advertisers to start running campaigns. These peaks vary widely among events, with consumers starting to research and purchase anywhere from three months ahead for Black Friday to the same day for St. Patrick’s Day. Seasonal gifting triggers peaks, too, for certain events: These days, like Valentine’s Day and Mother's Day, can bring decent traffic for advertisers, albeit much lower than the main events. ## Time Your Holiday Shopping Ads: Realize Data-Backed Engagement Peaks So, what’s the ideal time to start advertising for each of these holidays? Let’s get into the data. ### Black Friday Date: Fourth Friday in November. Duration of the peak: Two months. The year’s biggest day for shoppers can and should be the biggest day for retailers, with well-timed ad execution. In both of the past two years, the peak is two months ahead of the event, which means advertising starting in late September is perfectly reasonable to get shoppers’ attention. ### Christmas Day Date: December 25. Duration of the peak: Two to three months. The wishlists may have been checked off and gifts may have been opened, but Christmas Day is still the next busiest shopping day after Black Friday. In 2024, consumers started paying attention three months ahead (compared to two months the previous year), so starting early and offering targeted deals is a great bet. ### Cyber Monday Date: Three days after Black Friday. Duration of the peak: Two to three weeks. Cyber Monday, a somewhat newer event on the holiday shopping calendar, aims to capture shopper attention online after a presumably busy in-person retail weekend. Its engagement peak is quite a bit shorter than Black Friday or Christmas — two weeks last year, and three in 2023. Advertisers can consider running targeted ads and offers, such as focusing on fast shipping and convenience, with less lead time than others in the holiday marketing mix. ### Halloween Date: October 31. Duration of the peak: One to one and a half months. Halloween shoppers are often looking for fun, unique costumes for trick-or-treating and parties, deals on candy, and decorations for their houses and lawns. Last year, consumer interest started a month and a half before the event, so mid-September isn’t too early to target audiences with engaging creative and copy that aligns with the holiday’s themes. ### St. Patrick’s Day Date: March 17. Duration of the peak: Less than one day to 10 days. This holiday has a short lead time compared to others — the longest was last year at 10 days, while before that it was a same-day event. Shoppers might be looking for activities or items for school-age kids, or accessories for parties or bar-hopping. Target your audience carefully with themed creative to see your own sales peak on this holiday. ### Thanksgiving Day Date: Fourth Thursday in November. Duration of the peak: One to one and a half months. This U.S. national holiday sees groups of family and friends gathering, so housewares and food and beverage purchases top the shopping lists. Once the turkey’s been served, consumers quickly turn their attention to the main event: Black Friday and the busy month that follows. Some retailers may start their deals on Thanksgiving Day itself, both online and in-person. Use what you know about your audience to create discounts or offers on this day. ### Valentine’s Day Date: February 14. Duration of the peak: Two weeks to one month. This quick, one-day holiday is a big one for gifts, and in the past two years, consumers have started browsing and shopping a month ahead of time. Flowers, candy, and jewelry still top lists, as well as projects and cards for school-age children. Use the targeted timeframe to try out some eye-catching creative and copy. ### Memorial Day Date: Last Monday in May. Duration of the peak: Two weeks to one and a half months. Often tagged as the unofficial beginning of summer in the U.S., this holiday revolves around outdoor activities like barbecues, with food and beverage purchases made along with outdoor furniture and pools or pool accessories. It’s also a chance to try some red, white, and blue or other patriotic creative in interesting ways. This holiday is a popular one for auto makers and dealerships, too, who can offer special discounts ahead of the summer travel season. ### Mother’s Day Date: Second Sunday in May. Duration of the peak: One to one and a half months. This gift-centric holiday includes products as well as experiences like dining out or spa treatments. Last year, the peak was about a month ahead of the holiday versus a month and a half in the years before, so consumers start thinking about this day fairly early. Get ahead of that peak with special offers and thoughtful copy. ### Boxing Day Date: December 26. Duration of the peak: Three weeks. This U.K., Australia, New Zealand, and Canada holiday the day after Christmas continues the bigger holiday shopping season, with retailers offering post-holiday sales and end-of-year clearance on everything from beauty and clothing to electronics and home goods, along with luxury items. The peak in the past two years has been about three weeks, so consumers may be considering treating themselves post-holiday. Include this in your broader holiday ads plan to capture interest even after Christmas has ended. ### Father’s Day Date: Third Sunday in June. Duration of the peak: Two weeks to one month. Father’s Day shoppers started engaging with ads about a month ahead of the event in the past two years. Similar to Mother’s Day, this one-day event involves gifting and experiences, and advertisers can experiment with fun, thoughtful copy and appropriate offers for the day. ### New Year’s Day Date: January 1. Duration of the peak: One and a half months. The holiday season has ended, and a new year begins, offering advertisers a lot of room to be creative in their ads to reflect the spirit of resolutions and new beginnings. In the past two years, the peak for New Year’s Day consumers was a full month and a half beforehand, so the shift to a new mindset starts emerging in November. These shoppers may still be looking for great post-holiday deals along with wellness, fitness, and other goal-setting-related purchases. ### Back to School Date: September 1. Duration of the peak: Three weeks. Back-to-school season is approximately September 1, though it may be a bit earlier or later, depending on the region. The consumer peak for this event is about three weeks, so early August is the right time for advertisers to start using relevant copy and creative and calling out any special deals or offers for students of all ages. ### Amazon Prime Day Date: Mid-July. Duration of the peak: One week to one month. Amazon’s Prime Day has expanded to a full four days, and offers businesses an opportunity for a sales spike in midsummer. As the event has grown, so has consumer awareness and browsing ahead of the event. In 2023, the peak was 10 days, while in 2025, it was a full month. If you’re taking part in Prime Day, get started early to target your audience and generally make sure your offers are on your users’ radar. ### Independence Day Date: July 4. Duration of the peak: Three weeks. This midsummer U.S. holiday sees the consumer purchasing peak about three weeks ahead of time. Advertisers in food and beverage can take advantage of the holiday with special offers and discounts, and larger purchases like cars, outdoor furniture, or grills are popular. Themes of patriotism and history are relevant on this day, so see how you can reflect that uniquely in your creative. ### Presidents’ Day Date: Third Monday in February. Duration of the peak: Three days to three weeks. This holiday is a smaller event, with, most recently, a three-week peak for consumers. This holiday often sees specials on things like mattresses, appliances, and furniture. Winter clothing and items like ski gear may also have special offers during this time, so consider how you might target this one-day event wisely. ### Juneteenth Date: June 19. Duration of the peak: Four days. Juneteenth is a newer federal holiday in the U.S., offering advertisers another summer event to highlight items like outdoor furniture and decor, and food and beverage. This holiday also offers a chance for books, media, and other items for education on this Black history event. Consider how to balance celebration and reflection in your ads. ### Labor Day Date: First Monday in September. Duration of the peak: One to two weeks. The end of summer holiday also usually overlaps with Back to School, so you might combine the two or highlight one over the other, depending on your products. Labor Day often sees sales on outdoor gear like grills, and big appliances like refrigerators usually have sales since new models come out in September and October. You can try various creative approaches, depending on your products and how you’d like to position this event. ### Martin Luther King Jr. Day Date: Third Monday in January. Duration of the peak: 11 days. This winter holiday is educational and historical in the U.S., with a peak of 11 days before the event last year. In recent years, streaming services have also offered special deals on or around this day, and some retailers may start TV purchase offers now ahead of the Super Bowl in early February. Though a smaller holiday, you might consider a campaign with targeted celebratory and informational creative. ### Veterans Day Date: November 11. Duration of the peak: Five to seven days. Veterans Day in November is ahead of the holiday season, but has developed its own themes around celebrating military service. Consumers may shop for discounted restaurant meals, clothing, and home improvement goods at retailers, with special offers for veterans and active-duty military personnel. There’s also a creative opportunity for patriotic colors and American flags. ## Key Takeaways Mistimed campaigns risk being buried in ad noise. Make sure you don’t start your efforts too late, and remember that every holiday is different. Don’t ignore post-holiday events like Christmas Day or smaller but gift-centric holidays like Valentine’s Day. Digital marketing demands precise timing and data-informed decisions to succeed in crowded marketplaces during shopping events like the holidays. Realize captures specific trends and data points so that advertisers can make the best decisions for their own budgets and campaigns to get the most out of audience attention during these busy times. ## Frequently Asked Questions (FAQs) ### What time of year do people shop online the most? Generally, the time between Black Friday (the fourth Friday in November) and New Year’s Day is the busiest online shopping period. As the Realize data shows, other holidays are still valuable for advertisers to target, especially if they time their offers right. Smaller holiday events like Amazon Prime Day, Mother’s Day, or Back to School can all offer a lot of opportunity when timed and targeted correctly. ### What is the most popular shopping day during the holiday season? Black Friday tops the list in terms of busy shopping days, as it kicks off the holiday shopping season the day after Thanksgiving in the U.S. Started as an in-person event to draw shoppers into stores as early as Thanksgiving evening, it’s grown broadly online, too. Retailers target this day with special deals and offers, targeting users for peak sales. Last year, 81.7 million consumers shopped in-store on Black Friday, with 87.3 million shopping online. ### What are the trends in holiday spending? For the 2025 holiday season, research found that 53% of shoppers plan to spend about the same amount that they did in the 2024 holiday shopping season, which was a record-breaking year. You can use Realize to discover details on peak times for advertisers to capture holiday shopping timing, and our Trends tool can offer guidance on headlines, trending topics, and more. --- ### Best Marketing Campaign Management Software Options URL: https://www.taboola.com/marketing-hub/marketing-campaign-management-tools/ Last Modified: 2026-01-13 10:34:22 Are you still managing your marketing campaigns the old way, with hands-on A/B testing, siloed tools, and campaigns that presume to be full funnel but don’t reach anyone’s heart? Automation tools, artificial intelligence, and full-service campaign management platforms can take over the mundane or repetitive tasks so your team can work on developing the best creative, speak directly to your target audience, and improve your ROI in less time. ## Eight Best Campaign Management Tools Software Best for Campaign channel Features Pricing HubSpot Social media, email, and content management. Email, social media, inbound. Marketing automation, ad management, CRM, email marketing, social media scheduling. Free plan, paid plans start at $9/mo for individuals, $800/mo for teams. Zapier Integration between web apps. Email, inbound. Unlimited app integrations, AI-powered automation. Free plan, paid plans start at $19.99/mo. Monday Project management. Cross-platform through open API. Customer service, campaign management, project management, unlimited APIs, AI-powered tools. Free plan for up to 2 seats, Basic plan for $9/mos., Standard plan $12/mos., Pro plan for $19/mos. per seat, with custom Enterprise plans available. Nutshel Those looking for an easy-to-use CRM. Email, web, SMS, social media, WhatsApp. CRM, real-time reporting. Starts at $19 per user per month, up to $89 per user per month. 14-day free trial. Zoho Integrating all business functions under one service. Email, social media, web, business listings, webinar. CRM reporting, marketing automation, email, password management, etc., with 41+ business operations in all. 30-day free trial, pricing from $37/month per employee for Zoho One. $14/month for Zoho Marketing Automation, Standard plan, with up to 10 users at 1,000 contacts. Salesforce Most familiar CRM. Email, web, e-commerce, social media, SMS. Email integration, CRM, AI, personalization, Slack integration. 30-day free trial, $25/month per user for Starter Suite, $100/month per user for Pro Suite. Adobe Experience Manager Content creation and management. Web, mobile, apps. Easy content creation, form creation. No pricing published. Realize Performance marketing. Omni channel, native advertising. Draws on proprietary, first-party data, delivers highly targeted ads using AI, AdMaker suite makes content creation easy. CPC. ### 1. HubSpot A tried-and-true solution in the CRM and marketing space, HubSpot is best for marketers looking to manage email and social media campaigns. Larger packages for enterprise-level solutions combine multiple tools for marketing, sales, and customer service. HubSpot is best known for its email marketing and social media management tools, but also incorporates meeting scheduling for sales, ad management, and more. Prices range from a freemium version with limited tools for a single user up to custom, enterprise-level packages starting at $3,600 per month. HubSpot users tend to love the advanced capabilities and ease-of-use, but smaller teams often bemoan the lack of features within the lower tiers. ### 2. Zapier Zapier automates repetitive tasks across marketing and project management platforms without the need for complicated coding, and is good for organizations that want to streamline workflows. Zapier is great for internal functionality, but can also operate across channels, including email, social media, web advertising, and more, through integrations. Although it has an easy to use interface, it can get expensive for enterprise organizations. Plus, if you’re looking for real-time monitoring of campaigns, you’ll have to look elsewhere. ### 3. Monday Best known as a project management and workflow tool, Monday delivers a suite of products to manage CRM functions, IT helpdesk and customer service tickets, software development, workflow and project management, and more, all leveraging AI to streamline and automate tasks. Prices range from a free-for-life version with up to two seats, to three options for pay-per-seat plans, as well as enterprise-level custom plans to meet the needs of any size organization. Customized dashboards and real-time campaign tracking require no coding skills. Reviewers like Monday’s user-friendly interface and real-time collaboration capabilities, but Forbes Advisor noted that the free and basic plans are lacking in administrative features. ### 4. Nutshell If you’ve tested several CRMs and found many of them complicated to use, Nutshell might be the best marketing campaign platform for your organization. Affordable and user-friendly, Nutshell Sales delivers a variety of features to track sales and marketing across platforms and campaigns. Nutshell All-in-One combines email marketing, CRM features, invoicing capabilities, SMS, and web chat, and various marketing functions. All Nutshell Sales plans include AI chatbot and SMS tools, as well as landing page and form builders. Users can start with a 14-day free trial and then pick from one of five plans, ranging from $19 per user per month up to $89 per user per month, with savings if you choose an annual plan. Only the most expensive package, Enterprise, offers API support, however, which could be a deal-breaker for some users, but all plans include ChatGPT and Claude App Connections. ### 5. Salesforce One of the original CRMs, more than 150,000 companies, including many Fortune 500 firms, rely on Salesforce for sales and marketing integration. Salesforce offerings fall into three categories: Salesforce Marketing Cloud, Sales Cloud, and Service Cloud. Salesforce is widely recognized as the #1 CRM, but some companies may find it too robust for their needs, so the company offers a variety of products that integrate to meet the needs of any size organization. Salesforce’s Starter Suite is tailored for small businesses, combining sales, marketing, e-commerce, and AI agents to allow sales and marketing teams to work more efficiently. Salesforce integrates with common workplace apps, and small business users will enjoy having sales, service, and marketing on a single, integrated platform, with add-ons available to scale capabilities as your business grows. ### 6. Zoho Like Salesforce, Zoho offers a comprehensive suite of products that combines CRM, email, helpdesk support, project management, an online password manager, and more for businesses of every size. Zoho has more than 130 million users across the globe. Unlike many CRM platforms, Zoho One can also incorporate human resource functions, legal, and bookkeeping/accounting and payroll. Undoubtedly robust, with real-time collaboration features, Zoho One includes 45+ business apps with centralized administrative control and mobile management capabilities. Having so many features can make it too complex, and pricey, for some small businesses, however. You can sign up for a free 30-day trial of Zoho One or any of the individual services, and after that, plans start at $37 per month, per employee, paid annually, for Zoho One. Zoho Marketing Automation starts at $14 per month, billed annually, for up to 10 users and 1,000 contacts with a Standard Plan and goes up to $1,349 per month for up to 25 users and 500,000 contacts. ### 7. Adobe Experience Manager Many people think of Adobe as a publishing software, with Acrobat being the first service that comes to mind. Adobe Experience Manager is a complete, multichannel content management system that works across web, mobile, and apps. With automation and content creation tools, the platform allows you to create, manage, and optimize omnichannel campaigns and create easy-to-use forms for lead generation and digital enrollments. Adobe’s cloud native service offers complete customization to scale with your business or specific campaigns. Adobe doesn’t publish pricing on its website, so it’s best to request a demo if you’re interested in exploring the platform. ### 8. Realize Realize is a unique, AI-powered tool for campaign management within Taboola’s advertising platform. Realize goes far beyond native advertising, allowing users to scale display campaigns and reach bottom-of-the-funnel buyers on the open web. Focusing on the action stages reduces customer acquisition costs (CAC) while boosting the return-on-ad-spend (ROAS). Rather than pricey monthly subscription costs, with Realize, you only pay for results. Best for performance marketing at scale, the platform uses advanced tools like automated A/B testing, predictive AI scoring, and real-time campaign monitoring. Users can track results, dial in, and double down on the most successful content for better outcomes. Realize also utilizes 17 years’ worth of proprietary data to reach the right audiences at the right time, and makes it easy to craft custom creative assets that move people to action, all while offering centralized management for multiple campaigns. ## What Is Campaign Management Software? Campaign management software is a platform that helps manage every aspect of your digital marketing campaigns. The best campaign management software automates and optimizes elements of every campaign, providing real-time analysis so you can adjust your audience, creative, or placements on-the-fly for better results. ## What Features of a Marketing Campaign Software Are Important? Companies may look for different features in marketing campaign software depending on their specific needs and the platforms and audience they want to reach, but the best marketing campaign software should deliver the following features: ### Multiple Options and Formats for Creative Most companies don’t rely on a single touchpoint for a successful marketing campaign. You want to reach audiences through social media, email, text, and on the open web, to name just a few. The best marketing software allows you to create content in the form of native advertising, display ads, vertical ads, video, and more. ### Allows You to Repurpose Assets for Multiple Platforms Marketers know that repurposing content — turning long-form blog posts into social media assets, compiling articles into an e-book, or even creating a video from a social post — saves time and money. Most marketers (94%) repurpose content, and nearly half (46%) of those say they see the best results from repurposing content to different media platforms, rather than creating new content or updating existing content. ### Intuitive to Use Your marketing team should spend its time on strategy and content creation — not learning a new software platform. The best campaign management software should be easy and intuitive to use. It should also provide tailored one-on-one support when you need it to help you maximize every campaign. ### Effective Predictive Audience Targeting Your marketing campaign won’t be effective if it’s not reaching the right people. Look for a platform that uses first-party data, along with advanced AI, to deliver better results. ### Automation and AI Marketing campaign software should make your job easier by automating functions for best results, with AI-powered tools that help you stay ahead of competitors. ### Performance Analytics When you find creative that works, you want to double down to boost your results. At the same time, you need to shift gears if you’re throwing money away with ineffective creative or the wrong media channel. Your marketing campaign software should make it easy for you to gauge results with real-time performance analytics, helping you keep pace with a fast-changing market. ### Collaboration Tools Many hands can make for light work — or a disjointed mess when you’re conceptualizing and developing creative for your campaign. On-platform collaboration tools can streamline content creation, approvals, and buy-in so you can get your campaigns out there faster. ## Campaign Management Software: Pros/Cons Pros Cons Centralized control of campaigns. Not all software offers real-time monitoring. Manage ad spend efficiently. May come with high costs, especially at the enterprise level. Collaborate with teams. Some platforms have complex onboarding requirements. ## Benefits of Using Campaign Management Software ### Automate Workflows for Efficiency Today’s AI-powered CRM software allows teams to seamlessly automate workflows for efficiency. Your creative team doesn’t have to be adept at coding to integrate and connect various platforms, and they can instead focus on content, while the software seamlessly measures results. ### Integrate Functionality Across Apps Today’s best ad campaigns run across multiple channels: social media, email, even SMS. Campaign management software can integrate these tools and apps to maintain consistent messaging and measure results across channels. ### Track Results in Real-Time for Better Campaign Performance With access to AI-powered analytics, marketing campaign software allows you to track results in real-time and adapt on the fly to maximize conversions. ## Considerations With Using Campaign Management Tools ### Know Your Goals Before investing in any software, it’s important to understand your holistic goals. What do you want to achieve? Many marketers today aren’t looking for a full-funnel approach — they want to reach buyers in consideration mode and ready to convert. Your software, and your ad spend, should focus on those results. ### Understand Integrations What software are you already using? If you bring something new into the mix, will it integrate easily? ### Watch for Hidden Costs Major complaints of software like Zapier and HubSpot are that prices rise as you need increased capabilities. On the other hand, smaller teams may find the basic packages don’t deliver the functionality they need. ### Know Your Team’s Comfort Level Some campaign marketing tools require extensive onboarding or ongoing training as features expand. Find out what kind of help is available: Are experts on-hand to assist? ## How to Make the Most of Your Campaign Management Software for Growth (Best Practices) ### Establish Clear Campaign Goals You can’t improve what you don’t measure. Lead gen, engagement, and conversions are all worthy goals, but only one, conversions, puts money directly in your pocket. Be willing to spend money to convert your audience at the consideration stage of the sales funnel. ### Know Your Audience Dialing in on your audience, and segmenting it based on demographics, needs, or where they are in the sales funnel, can help maximize results. Create content that speaks to their pain points and showcases what makes your offerings different, and reach your audience on the platforms where they will see it and engage with it, to maximize your ROI. ### Test and Optimize for Performance Not every campaign performs perfectly right out of the gate. That’s why A/B testing and analytics are so important. ### Turn to the Experts for Help Any effective campaign management software should have a team of experts available to guide you via phone, email, or chat. Lean on them for guidance! ## Key Takeaways Choosing the right campaign management tools for your company depends on your budget, needs, and company goals. Software that identifies and addresses audiences in the later stages of the marketing funnel will drive meaningful outcomes and help maximize your ROI. ## Frequently Asked Questions (FAQs) ### Which software is best for advertising? The best platform for advertising is down to your specific needs, but should ideally provide real-time campaign management, AI-powered capabilities, and centralized management for multiple campaigns. You should be able to launch campaigns quickly and reach customers in the conversion and action stages, so you’re paying for results, not just visibility. ### What is a CRM campaign manager? A CRM campaign manager is someone who uses data and tools within the CRM platform to optimize workflows and maximize results across multiple channels. ### In which tool are campaigns created and managed? Marketers can create and manage ad campaigns in their CRM, or customer relationship management software, or in specialized platforms for performance marketing campaigns. --- ### Data-Backed Consumer Insights for Cyber Week 2025 URL: https://www.taboola.com/marketing-hub/cyber-week-insights/ Last Modified: 2025-10-15 07:14:51 For retailers, each year typically ends with one final blowout, with the period between Thanksgiving and New Year's Day bringing in 19% of retailers’ annual sales, on average. But, that puts significant pressure on business owners to come up with a strategy to maximize their efforts for the holiday shopping season. By looking at recent data alongside emerging trends for the fourth quarter of 2025, you can build a plan that makes the most of the year’s biggest shopping season. ## The Holiday Shopping Timeline When does the holiday shopping season kick off? If you see Black Friday as the starting line, you aren’t alone, but while Thanksgiving weekend remains the top traffic and conversion period, in 2024, major traffic spikes happened as early as Prime Big Day Deals, which was October 8-9. Fall sales also brought another uptick in traffic in mid-October, with big sales from Hoka and Nordstrom. For 2025, expect that pattern to continue — 27% of U.S. adults start shopping Cyber Week deals in October, with another 26% beginning in early November. Another misconception is that Cyber Week brings the majority of holiday sales. In 2024, merchants continued earning through the end of the year, with a sharp dip just before Christmas. Sales surged again briefly in the week after the holiday, demonstrating a missed opportunity for those retailers who cut off their marketing plan too early. Well-timed gift guides can help extend Black Friday into the holiday shopping season. In 2024, gift guides were a top performer, and shoppers leaned heavily into them for ideas across all recipient and interest types. ## Best Time to Publish The holiday season is hectic for most consumers as they add festivities, shopping, decorating, and planning for houseguests to their already busy lives. For that reason, it can be tough to reach customers with your social media posts in the final weeks of the year. In the last quarter of 2023, Wednesday was the top-converting day for editorial content. That differed during Thanksgiving weekend, of course, with Black Friday converting the highest. Marketers should also pay attention to the best time of day to post: Overall, 3 p.m. Eastern Time worked best during the holiday shopping season, but on Black Friday, activity spiked at midnight. By scheduling promotional content for peak activity times, you can maximize your holiday marketing efforts. If you have flash sales or activity guides you’re launching, consider scheduling your posts for midweek and midafternoon. ## Types of Content That Worked Looking at the top 200 articles posted during the 2024 holiday season, four major content types drew the most interest: ### “Best of” Content “Best of” lists continue to attract shoppers looking for curated recommendations. In the final quarter of 2024, top-performing content included listicles highlighting home decor, holiday decorations, and gifting ideas. These roundups work best when they mix trending high-quality and affordable items that help readers make well-informed buying decisions. ### Gift Guides Gift guides continued to show high engagement in 2024 after topping our list in 2023. Successful roundups included those that organized by recipient and interest, while consumers also responded well to gift guides highlighting subscription boxes. The easier you can make things for your followers, the more they’ll learn to lean on you for help with their shopping. Once you’ve launched your gift guide, pay close attention to pricing changes. You can update your guide regularly with the new prices, giving you yet one more chance to share it with your followers. Don’t forget to use affiliate links for an additional holiday season revenue stream. ### Black Friday Deals Black Friday content saw results throughout October and November in 2024. Top performers included Walmart, Dyson, Carhartt, and Nordstrom, all of whom offered early access to deals. Roundups of deals also did especially well, showing that shoppers want to know where to find the biggest savings. For 2025, focusing your Black Friday coverage on early promotions and standout offers is key. ### Sales Roundups General sales content is also popular with consumers looking to save. In fact, Black Friday and general sales articles outperformed Cyber Monday content, showing that interest peaks at the start of Cyber Weekend. For best results, highlight limited-time offers and retailer-specific discounts to drive urgency and encourage conversions. ## Mobile Optimization Mobile shopping increased by 18% during 2023’s Amazon Prime Day, showing the value of prioritizing user experience. As with other times of the year, it’s important to ensure your site is optimized to capture mobile shoppers, whether they’re sitting at home on the sofa or standing in line to make other Black Friday purchases. In addition to mobile friendliness, you should also make sure you’re offering your customers a variety of ways to pay. Try to eliminate any barriers that can keep a visitor from converting. ## Generational Shopping Trends Generation Z and millennials play a pivotal role in holiday sales, but how they shop sets them apart from older demographics. 14% of millennials say they’ll buy a gift directly through an online ad, which is the highest of any generation. Meanwhile, 35% of Gen Z will make their purchases through social media, making platforms like Instagram and TikTok critical touchpoints. Social media doesn’t just drive direct purchases, though: Gen Z shoppers are 125% more likely than any other generation to say that something they saw on social media influenced their holiday buys. These younger generations are spending more, too, with 29% of Gen Z and 25% of millennials planning to spend more this year. Gen X and Baby Boomers, meanwhile, remain cautious. ## Simple, Deep Discounts Once again, deals will drive holiday purchase decisions in 2025. A majority of shoppers, at 55%, say discounts, coupons, and promotions are the best way to win their money. Deals even rank ahead of convenience, at 43%, and product reviews or ratings, at 35%. As you’re planning promotions, focus on keeping them clear and compelling. Straightforward discounts remain the most effective way to stand out during the crowded holiday shopping season. ## Budget-Conscious Shoppers in 2025 Concerned about the economy, some holiday shoppers plan to be more cautious in 2025, particularly older generations, as noted above. That doesn’t mean they’ll stop spending altogether, though: More than two thirds of shoppers say creating a budget is important, and over half plan to stick with the budget they create, but an impressive 90% still claim they’ll spend the same or more than they did last year, with 35% saying they’ll top last year’s spend. For retailers, this means balancing affordability with premium appeal. Budget-friendly roundups can help catch cautious shoppers, while tiered gift guides (“10 gifts under $25,” for instance) make it easier for consumers to find something within their target price range. Include some higher-priced items in your tiered guides to attract both budget-conscious shoppers and those looking for something higher-end. ## The AI Challenge for Holiday Marketers Artificial intelligence (AI) seems to dominate every marketing conversation these days, so it’s no surprise that this holiday season, it will make a big impact. For retailers, this holiday season will be affected by the notable shift in search behavior: Publisher traffic has fallen 27% since early 2024, and Google search results with AI overviews produce a 34.5% lower click-through rate. Shoppers are increasingly getting shopping inspo from platforms like Reddit, TikTok, and YouTube, only increasing the zero-click search issue. To stay visible, marketers need to adapt. Pair search engine optimization with generative engine optimization: Use bulleted takeaways, question-driven headlines, and purpose-based language — for example, write, “best dress for a bachelorette party” rather than “white dresses.” Beyond search, make direct connections with consumers through newsletters, shopper columns written by experts, and an interactive social media presence. ## Monitor Top Categories Do you have trending products in your inventory? It can be tough to predict what will stand out each holiday season, but looking at previous years can help, as can looking at what’s working so far this year. In 2025, the product categories seeing the biggest growth compared to last year are: - Beauty: Up 34%. - Luxury: Up 30%. - Clothing and accessories: Up 18%. - Appliances: Up 16%. Clothing and accessories are performing very well this year overall, with content themes like “travel outfits” and “flattering shorts for men” bringing the best results. Appliance content on topics like “window air conditioners” and “the best Dyson vacuum” also does well. ## Plan Around Key Q4 Events Again, some shoppers are expected to be more budget-conscious in 2025 than in previous years, but that brings an opportunity for marketers to lean into early deals and limited time offers. With a large chunk of the shopping population starting in October and early November, timing is everything, so time your content to coincide with the hottest shopping days to help you maximize your campaigns. As you’re planning your Q4 content calendar, add the following dates so you don’t miss them: - Mid-October: Prime Big Deal Days typically fall around this time of year. This is an ideal time to launch your fall sales. - October 31: Halloween brings another marketing opportunity for retailers. Even if you don’t sell Halloween-themed products, you can engage customers through costume shopping guides and healthy treat tips. - November 11: This is not only Veterans Day, but it’s also Singles’ Day. Make the most of the occasion with “treat yourself”-themed content and special discounts for veterans. - Mid-November: Start teasing your Black Friday/Cyber Week deals earlier in the month and offer special “Early Black Friday” promotions to get sales moving early. - November 28-December 1: The big shopping weekend kicks off at midnight on Black Friday, and it brings your biggest opportunity to reach the many shoppers looking for deals, both online and in their local stores. - November 29: In between Black Friday and Cyber Monday is Small Business Saturday, when customers are encouraged to buy from locally owned merchants. Even if you don’t have a brick-and-mortar, you can drive business to your store for Small Business Saturday. - Early to mid-December: Customers don’t stop shopping once Cyber Monday ends, so many brands keep discounts coming in the weeks to follow. This is also a great time to post gift guides that drive engagement. - Mid- to late December: Last-minute shoppers count, too. Use this time to be specific about items that will arrive in time to be under the tree on Christmas morning. - December 25-January 2: Both Hanukkah and Boxing Day (December 26) fall in this timeframe, giving you another chance to engage customers while also offering special deals to drive sales. - New Year's Eve: New Year's Eve offers the last chance to generate Q4 sales. Celebrate the end of a great year with countdown sales, party supplies, and big year-end blowout promotions. As we head into the last quarter of 2025, marketers who prepare early, stay agile, and pay attention to data will dominate. From optimizing evergreen content for the holidays to capitalizing on real-time trends, marketers have plenty of opportunities to shine. Unfortunately, competition is fierce, so it’s important to have the right tools and strategies to set yourself apart from the many other brands battling for attention. --- ### Marketing Campaigns: Reaching Your Audience Where They Are, Again and Again URL: https://www.taboola.com/marketing-hub/marketing-campaigns/ Last Modified: 2025-09-11 10:05:21 Do you remember a certain TV commercial from your childhood that just stuck in your memory and that still pops in there to this day? Maybe it was an ad for gum, for a ride-on toy, or for some sneakers. Regardless, that one commercial that really sunk in? That wasn’t a marketing campaign, that was just a tactical piece of a larger marketing strategy. The marketing campaign was that gum not only appearing in a TV ad, but also on posters at bus stops and in train stations; it was the same ride-on toy beginning featured in ads as well as being placed in a primetime sitcom. If those sneakers you loved from the commercial were also on display in store windows and in print ads in magazines, then you had yourself yet another example of a marketing campaign. In fact, had it not been for the larger marketing campaign promoting those (hypothetical) products, the ads you just can’t forget might never have made their way into your deeper consciousness at all. ## What Is a Marketing Campaign? At a basic level, a marketing campaign is a related set of marketing and advertising activities usually involving various channels and media, designed to promote a specific product, service, or brand, with the goal of achieving a defined objective. These objectives can include better sales, heightened brand awareness, new leads, and more. In more exciting terms, a marketing campaign can be everything from the sponsorship of extreme athletes sailing a branded boat around the world, to a selection of subtle product placements in an action-adventure movie, or a clever online video series that blends comedy and storytelling with advertising, and so on. Marketing campaigns come in all shapes and sizes and almost no two marketing campaigns look the same. They do often have shared goals, however. Marketing campaigns are created to achieve specific business goals, and they’re designed to reach a specific group of potential customers, leading to conversions among those most likely to feel that a brand, product, or service resonates with them personally. ## What Are the Different Types of Marketing Campaigns? ### New Product Launches Whether it’s a brand new item from a legacy brand or a new product from a cheeky startup, the more people who are made aware of the new product, the better the chance it has of securing a toehold in the marketplace. New product marketing campaigns can be tricky, as advertisers have a lot of informing to do in what’s often a very confined set of parameters — a single five-second online ad or one slot atop a webpage, e.g. — and they can’t count on recognition of an established product. One clever tactic advertisers can use when trying to launch new products is to create a sense of mystery or drum up interest by not sharing much information, but instead using an ad to entice people to visit a site where more information is then shared. ### Brand Awareness Campaign In many ways, the brand awareness campaign is the most common form of advertising. It can take the shape of ads plastered on the sides of race cars or around the walls of stadiums; they may include branded merchandise handed out at music festivals, or “this message was brought to you by so-and-so” clips on radio programs or podcasts. You’re already well aware of brands like Coca-Cola, Nike, Volkswagen, and other brands that engage in this sort of general awareness advertising. What you might not be aware of is how that marketing worked on you the next time you consider which soft drink to sip, which shoe brand to purchase, and which car to test drive. ### Rebranding Campaign Many companies choose to rebrand at some point, which can include a switch to a product line or a total rebranding of the company itself. Companies may pursue this route due to flagging sales, demographic changes, societal pressure, a scandal, and more. Whatever the reason for a rebrand, when accompanied by a successful marketing campaign, rebranding can help a brand evolve, shedding unwanted baggage but maintaining some of its clout. ### Seasonal Advertising Campaign The same marketing approach that worked great for your brand in the summer might make very little sense as winter (and the winter holidays) approach. Creating marketing campaigns that are relevant to the season is a good way to keep your advertising fresh and relatable. ## What Are the Different Channels Used for Marketing Campaigns? For many years, marketers had several major ways to reach their potential audiences: TV, radio, posters and billboards, print ads, direct mail, and word-of-mouth. Granted, one could also argue sponsorships and a few other channels existed, but those were the heavy hitters. In our digital world, things have changed. So, allowing that all of the above marketing campaign channels still exist, here are the best ways for marketers to reach potential customers via the digital landscape. ### Email Marketing This modern-day direct mail outreach involves sending targeted emails to subscribers for promotions, updates, and more. The more email addresses a marketer can gather, the more people they can potentially reach, and the more conversions they can drive. ### Online Videos Online video ads are distinct from TV commercials in many ways. First of all, users have the option to skip them, usually after just a few seconds, so marketers need to make these ads high quality and engaging. Second, these ads can be placed with much more precision than a TV ad, which runs whether or not anyone sees it, and regardless of the interest of its viewer. Third, online video ads can lead to immediate action, thanks to embedded links, whereas TV commercials are only really well-suited to new or enhanced product awareness. ### Social Media/Influencer Marketing For better or for worse, people are highly drawn to and often easily convinced by the people they love to follow online. If you can get established and popular influencers to use your products or services, there’s a very good chance of their halo effect radiating out to your brand and leading to increased sales. ## How to Create an Effective Marketing Campaign ### Know Your Audience The most effective marketers know the demographics of their target audience through and through. This means conducting research, segmenting your audience (dividing your audience into segments based on shared characteristics and tailoring your messaging), creating buyer personas, and employing past data and findings. ### Target Correctly While demographics remain important, the emergence of AI-powered behavioural targeting tools means that audiences can now also be targeted by intent, rather than just identity. Knowing which segment of the population you’re selling to is a great start, but narrowing it down to those most likely to actually complete a purchase is the real key to success. ### Set Your Goals The most successful marketing campaigns are informed by SMART goals, which stands for “Specific,” “Measurable,” “Achievable,” “Relevant,” and “Timebound.” This allows for goals that are realistic, motivating, and focused. ### Lean Into A/B Testing A/B testing is vital to the success of any marketing campaign. By launching different iterations of the same creative, you can get hard data on which version performs better, and continue tweaking it with further testing until you’re sure you have the best possible version. Utilizing AI-powered A/B testing tools will let you optimize in near-real time, keeping your creative constantly refreshed and engaging. ## How to Measure Your Marketing Campaign Efforts ### Return On Investment The most basic — and, arguably, the most important — way to measure the success of a marketing campaign is simply to see how much money you made compared to how much money you spent, also known as return on investment, or ROI. If you spent $5,000 on a new product launch marketing campaign, and saw that product generate $25,000 worth of revenue, that is a successful return on investment. ### Click-through Rates and Traffic The goal of many marketing campaigns is to get more people visiting your client’s websites or apps. If you see noticeably increased clicks and traffic in the wake of or during ongoing marketing campaigns, then you know what you’re doing is working. ### Increased Conversions Making more money is not always the chief indicator of a successful marketing campaign. While return on investment is critical, conversions other than sales can often be very important, such as people signing up for a newsletter, subscribing to a channel, sharing personal information, and so forth. ## How to Optimize and Improve Your Marketing Campaign Results To optimize and improve your marketing campaign results, focus on clear objectives and data-driven analysis and prepare for continuous refinement. This can involve setting specific goals, understanding the target audience, testing different approaches, and consistently monitoring performance and making necessary adjustments. ### Always Use A/B Testing Marketers who put all of their proverbial eggs in a single basket risk losing out. Only by running different iterations of your ad campaigns can you tell what’s working and what isn’t. Remember, there's a chance that neither of your campaigns will be all that successful, requiring total rejiggering. In other cases, both might work relatively well. In this latter case, don't abandon one for the other; realize that you might be drawing in different types of users and keep testing both with further tweaks. ### Data Analysis The longer a marketing campaign goes on, the more data you will have to analyze, and study it you must. With a close enough look at the numbers, you’ll be able to see who is clicking through on what ad, what time of day your ads are the most effective, what types of sites or apps are getting the most traction, and so forth. Dig deep into the data and respond to it quickly. ### Refresh Your Goals The advertising goals that initially informed a marketing campaign might need to change as you receive new data. Rather than always changing your marketing efforts to suit your goals, realize that sometimes you might need to change your goals themselves. Ads meant to drive sales that instead are driving sign-ups might not be a failure — they might simply be an indication that now is the time to grow your audience, and later is the time to grow your revenue, for example. ## Key Takeaways A marketing campaign can take many forms: Campaigns may be across a single “channel,” such as all on one radio station, but they usually involve different platforms and media channels and are designed to promote a product, service, or brand, with the goal of achieving objectives like sales or subscriptions. Advertisers need to develop KPIs and track them closely to see if marketing campaigns are working, then use that data to inform adjustments if campaigns are falling short. ## Frequently Asked Questions (FAQs) ### What is the importance of a target audience in a marketing campaign? Knowing your audience allows you to focus your efforts and resources on the most likely buyers (or other converters). This leads to more effective campaigns and better ROI. By understanding who your ideal audience members are, you can tailor your messaging, products, and overall marketing strategy to resonate with that specific group, increasing engagement and conversions. ### How do you define the goals of a marketing campaign? The goals of marketing campaigns are specific, measurable objectives (remember SMART objectives, as explained above) that advertisers aim to achieve through their marketing efforts, often aligned with larger business goals. Goals provide direction, focus resources, and allow for tracking progress and measuring success. They are defined by how they will lead to success in your KPIs. ### How do you determine the budget for a marketing campaign? Once you know how much you’re hoping to generate in revenue following a successful marketing campaign, you can then usually determine how much to spend to get the ideal ROI. In general, a 5:1, or 500% return on investment is considered good, so if you hope to make $50,000 in revenues, it’s safe to spend up to $10,000 on marketing, provided you are confident in your plans. --- ### 6 High-Impact Cyber Monday Ads to Learn From URL: https://www.taboola.com/marketing-hub/cyber-monday-ads-examples/ Last Modified: 2025-10-15 07:16:47 Cyber Monday is one of the busiest shopping days of the year, which makes it one of the most competitive days in digital advertising. If you want your ads to stand out, you have to find a way to cut through the noise. In the past, advertisers could rely on deep discounts, but they’re no longer enough: You need to blend discounts with other creative elements, including urgency, scarcity, celebrity endorsements, and good storytelling. If you can extend the sales period beyond Cyber Monday by connecting the entire Cyber 5 period, you can drive your revenue even further. In this article, I’ll explore six standout Cyber Monday ads from previous years and explore what makes them so effective. ## 6 Cyber Monday Ads Advertisers Should Know About ### 1. Best Buy Best Buy’s Cyber Monday video ad combines humor and convenience to stand out during the crowded shopping season. In the ad, a woman sits in her kitchen, browsing deals on the Best Buy website. Her partner enters, fully dressed for the cold weather, and asks her if she’s ready to camp out for “doorbusters.” Her response drives home the message: You don’t need to line up outside when the same deals are available online. This ad shines in two ways. First, it extends urgency beyond Cyber Monday by promoting weekly “doorbusters” on Fridays throughout the holiday season. This keeps shoppers engaged while capturing those who may have missed sales earlier during Cyber 5. Second, the ad connects with viewers by using humor. In this case, it pokes fun at outdated shopping rituals, making the ad relatable and customer-friendly. ### 2. Walmart: Deals for Days Unlike many Cyber Monday ads, Walmart’s “Deals for Days” ad doesn’t highlight any specific discounts — instead, it focuses on creating a sense of urgency. It does this by using time-based cues, such as “Starts Sunday Online, 11/28 @ 7 pm EST.” This makes it clear that the clock is ticking, so you need to act fast. The ad also highlights three key product categories, all popular during Cyber 5: electronics, household goods, and toys, by showcasing a laptop, a coffeemaker, and a dollhouse. This also reinforces Walmart’s broad appeal across a wide range of products, driving home the message that you can get everything you need in one place. Finally, by branding the campaign, “Deals for Days,” Walmart is shifting the focus from, “How much will I save?” to, “I can’t miss this window!” — an important distinction that, again, highlights urgency. ### 3. Titan Power+ This Titan Power+ display ad stands out by positioning its offer as a Cyber Week Sale, rather than a one-day Cyber Monday event. The ad creative for this maker of mobile chargers opens with: “Sad you missed our epic Black Friday Sale? Don’t worry…” This is designed to re-engage shoppers who feel like they missed out. By extending deals across the entire week, Titan is attempting to capture late buyers and spread its revenue across the whole Cyber 5 period, rather than relying on a one-day spike. From there, the ad leads with value by highlighting a deep discount (up to 80% off) in multiple locations, including alongside the product image. Cyber Week Ad copy, such as “Biggest discounts, biggest savings” and “Epic Cyber Monday Sale!” drives home the discount-driven messaging. ### 4. Walmart: Celebrity Endorsement In this Walmart Cyber Monday ad, the retailer leverages celebrity power and nostalgia to capture viewers’ attention. It’s a playful approach that draws inspiration from the 2004 film Mean Girls. The elaborate sketch features original cast members, such as Lindsay Lohan, Amanda Seyfried, and Damian Leigh: The ad is set in a high school, where “Santa” arrives to hand out Cyber Monday deals to excited students. Throughout the ad, bold text overlays highlight various product categories that are on sale, which drives home Walmart’s value message amid the chaotic scene. There’s no push for urgency, and no specific discounts are mentioned: Instead, the ad aims to be entertaining and culturally relevant. The ad does, however, close with a simple but effective CTA: “Cyber Monday Deals. Shop Online Now.” ### 5. Macy’s This Macy’s Cyber Monday ad lets shoppers know that the sale starts early by promoting “Sunday and Monday specials.” The ad also emphasizes deep discounts by overlaying bold text alongside product images, with messaging like, “Save 60% on sweaters, cashmere, coats and more for her…” and, “Get sweaters and packable down outerwear for him, now 60% off!” By highlighting products from both fashion and household categories, Macy’s takes a similar approach to Walmart, by showcasing its broad product range. This, along with the messaging targeted for both male and female shoppers, helps it appeal to multiple audiences simultaneously. Convenience is also a big selling point, as Macy’s lets people know that the sales are available both in-store and online, and that they can get free shipping at Macys.com. The combination of wide product selection, deep discounts, and convenience make this a classic, yet highly effective Cyber Monday ad. ### 6. Tailored Athlete At first glance, the Tailored Athlete display ad below appears to be a Black Friday ad, incorporating key elements commonly found in Black Friday ads — namely, urgency and a clearly marked deep discount. However, the ad employs another critical strategy by connecting Black Friday to Cyber Monday at the bottom of the ad creative with the following call to action (CTA): “Sign Up Early for BFCM Access.” By using “BFCM” (Black Friday Cyber Monday), Tailored Athlete extends the promotion access across the entire long weekend. This can increase sales opportunities, keeping shoppers engaged through Cyber Monday. ## Key Takeaways As you can see from the Cyber Monday ads featured above, the most effective ads strike a balance between urgency, value, and creativity to stand out during the busiest shopping period of the year. Some, like the Tailored Athlete and Titan Power+ ads, extend the window for deals beyond Cyber Monday in an attempt to capture more buyers. Large brands with substantial ad budgets, such as Walmart and Best Buy, rely on entertainment, nostalgia, and even celebrity appearances to make their campaigns memorable. All of the ads include strong visuals and clear calls to action. Finally, note that not every ad relies solely on direct discounts — in fact, some don’t even mention a specific deal. Other elements, such as urgency, scarcity, and good storytelling, can be just as effective. ## Frequently Asked Questions (FAQs) ### How can we best convey a sense of urgency and scarcity? To drive home a sense of urgency and scarcity, marketers should make sure their messaging is clear and time-bound. You can do this by using countdown timers, start-end dates (“ends tonight at midnight”), or highlighting limited quantities (“Only 50 available”). You can also use copy like “Last Chance” or “Shop Early.” ### What visual elements will immediately grab attention and convey the Cyber Monday theme? High-contrast text overlays that highlight discounts (“Up to 80% OFF”) continue to be the fastest way to get shoppers to stop scrolling. You can also use bright colors and product images to grab attention. Many retailers, including Walmart, showcase their top-selling product categories in hero images to ensure that shoppers are aware of the products on sale at reduced prices. ### How can we segment our audience to deliver highly personalized Cyber Monday ads? A tried-and-true approach is to start by leveraging behavioral data, including browsing behavior, past purchase history, and abandoned carts. You can use this information to retarget shoppers with more personalized offers. You can also segment by shopper interest and demographics — for example, parents are likely to respond to deals on toys, while post-secondary students will be more interested in tech items. Whether you’re using Realize or another ad network, you can leverage its optimization tools to test creatives and swap visuals and Cyber 5 ad copy in real time. --- ### Content Strategies for High-Quality Lead Generation: Insights from Xevio URL: https://www.taboola.com/marketing-hub/lead-generation-content/ Last Modified: 2026-03-15 13:48:11 Welcome back to our series of lead gen conversations with Xevio co-founder and CEO, Nadim Kuttab. In our last article, we discussed AI-powered lead generation, and how it’s transforming lead gen on the open web. In this installment, we’re looking at the ways in which authentic, valuable content can pre-qualify leads and drive conversions. Since successful open web lead generation is no longer merely about direct calls-to-action, but about building trust and demonstrating value through content, it’s clear advertisers need to invest in more robust content marketing and emphasize the quality of engagement over quantity of clicks. Here’s Kuttab’s advice for doing it right. You've previously emphasized creating content experiences that feel natural on the open web. How does this approach differ from typical ad creative, especially for lead generation? The content experience that really defines native advertising is a piece of content that’s both interesting and educational, and selling your product or what you're doing. It’s essentially a soft pitch to a product with a call to action at the end, and it needs to be built in a way that seems natural in a content-rich environment: A blog with either an opinion or fact-based case that it's making to the reader, and one that’s easy to navigate, especially considering that open web users tend to be a little bit older. Even for the newer types of ad creatives that Realize offers, this funnel and this user experience still performs exceptionally well, so we always start with content pages built around native advertising. They segment the user, they introduce a problem, then they introduce a solution to the problem, then they sell a product that is a solution with some testimonials and some social proof. We've taken that and we've applied it to pretty much everything on Realize, and it works well. What role do long-form advertorials/landing pages play in your lead generation strategy, and how do they help pre-qualify prospects? The landing page is effectively a way to take a cold user and warm them up. With long form advertorials, you want to keep that content as concise as possible, but at the same time, depending on who you're trying to reach, you might need longer content — say, if your product's hard to explain or expensive, then you might want to have the user spend more time in the funnel, warm them up more so they’re more willing to take action. The longer somebody stays on your page, the more likely they are to buy. That's why we have clickouts toward the bottom of the page, rather than at the top: You don’t want somebody to accidentally click on the first hyperlink in the first paragraph before they’ve been warmed up, as that’s the same as just sending them from an ad straight to the product page. It's not an effective use of what Realize can offer. How can advertisers effectively balance engaging storytelling with clear calls-to-action to drive conversions, without resorting to clickbait? It's all a matter of quality. If you deceive the user between the landing page and the offer page, they're going to jump from the offer page — they're not going to buy your product. They're going to be like, “Oh, what the hell? I expected something else.” It's the same as having an ad that says, “Click here and win $100,” and then on the advertorial it's essentially trying to sell them a credit card. People will see through that immediately. You want to have a compelling story that doesn't cause policy issues, because publishers will kick you out if you’re too aggressive with how you sell. The idea, then, is to find a healthy balance between interest and clickiness. There are ethical implications as well, but from a purely financial perspective, it's better to be clear with what you're selling, because then the users that click through will actually want your product or service and be much more likely to buy than if you deceive them into clicking on something that they don't want. Realize offers its Optimize for Engagement tool, with top-of-funnel metrics. How can these engagement signals inform and improve content strategies for furthest bottom-of-funnel lead generation? Let me give you an example. Say you have a target CPA of $100: To get enough data to test every single thing that you're doing, you have four or five landing pages, two or three offer pages, and 20 ads live on three different devices. On top of that, each campaign on each of these devices is going to get a different mix of traffic. That means you're going to be getting dozens, if not hundreds, of different publishers on your campaigns. You will never have enough data — you don't have enough money to get the data you need on a $100 CPA, so, you need to look at something else. The best initial metric to look at — obviously, you want to look at whatever is furthest down the funnel, but the best initial metric to look at when you're starting off — is landing page CTR: How many people are clicking through your content pieces to the actual product or service that you're selling? If you have a publisher or an ad piece that’s driving 20% CTR and one that's doing 5%, you can very quickly kill the 5% content piece and shift your focus to the 20%. You’re then more efficiently using your ad dollars before you've even generated a conversion. Those things are super valuable in helping you shift your budgets more efficiently before you have enough data to definitively say yes. Can you share a common mistake marketers make when creating content for lead generation on native platforms, and how best to avoid it? A lot of marketers will look at what others are doing and copy them. That’s not a particularly effective way of doing it, because the others are already established in the market. If you're trying to compete with somebody 10x or 100x bigger than you in the space, at the start, it’s going to be close to impossible, so I always say, look at what they're doing and do it better. What are their angles? Come up with a better or more interesting angle, try different things. We always produce content from scratch and it's been a strategy that's worked very well for us. So, my recommendation is, look at what others are doing, but don't copy. Trying to beat them at their own game won't work. Beyond the initial content, what are key elements of a high-converting offer page for leads, particularly when dealing with traffic from the open web? You want to have one question at the beginning that's very easy to answer. You don't want to have people put in their zip code or type out their full name or their email — that shouldn't be the first question. The first question is a trigger question to get them into your funnel — the likelihood of them continuing if they've already committed to clicking on one thing is much higher, so the goal of the first question is to pull them in, then have a couple of questions. And try to keep them short, for God's sake! Don't have a million complex things. Making it clickable, not typable, is the best bet until you get to the email, telephone number, and the personal details. When you say clickable rather than typable, you're talking about a multiple choice option? Exactly. Think of an offer page where it asks if you’re a homeowner or a tenant, right? You want to be able to click that and then go to the next question without having to type in anything. We actually use HeyFlow for this. How does the concept of "perceived value" (as you've previously discussed) translate into content strategy for lead generation, especially when the core product isn't changing? First off, perceived value is very subjective, which is why I say perceived value, not value. Let’s say somebody is clicking on an ad for a solar panel: It means that they are interested in solar panels in some way, shape, or form. Does it mean that they will buy one from you? No, but it means that they're interested. They've taken that step, they're reading about it, they've clicked through, they’ve filled out your form. You have their attention — they want to get a quote from you, otherwise they wouldn't have clicked this far. The idea next is to make it easy for them to get to the final answer quickly, adding a FOMO effect of, we might have something others don't: “There are potentially offers in your area that are amazing, click here to find out.” I'm curious! I want to know more — what are these amazing offers? What are these subsidies that I don't know about? There are companies that are cheaper than average because they're buying new solar panels today at a cheaper cost than companies were able to buy them at a year ago, and we can connect you with those companies. Put that information in! That’s one hell of a case, with very clear and concise messaging built on facts that you can show people. You’re educating the customer, essentially. You're educating the customer about the different options that they have. That's why it's perceived value in a way that makes you look good, like, hey, your service is going to help them get an offer. What's your advice for testing and iterating content for lead generation to continuously improve performance and lead quality? You need to be creative. A lot of companies don't send the data of the lead quality back to the platform, and that's a big mistake. If you can get that data into the platform, it helps you really improve lead quality. We were able to improve our lead quality 30% across the board for most of our clients over the last 12 months, just by getting lead quality data back to the click. Then, it’s about creating content that's catchy. Just try things — we’re constantly iterating, that's the truth. It's not rocket science, it’s just math: You have an ad and that costs a click, then you have a landing page, then a percentage of those people go through, then you have an offer page. A percentage of those people go through, then you have a solid lead, and a percentage of those leads will go through and generate revenue. You have different drop off points and there’ll be different metrics and different costs for each one. Look at the data and squeeze or turn the little knobs that you have to improve each individual point. There’s no more to it than that: If your CTR is bad on the landing page, that's what you have to put your effort into. --- ### Healthcare Marketing: Strategies to Reach and Engage Patients URL: https://www.taboola.com/marketing-hub/healthcare-marketing-strategies/ Last Modified: 2025-08-11 16:28:44 When consumers have questions, the internet is usually the first stop, and that especially holds true for healthcare. Whether it’s the strange bug bite on your ankle or the sinus issues that have plagued you for a few weeks, a simple online search can bring answers. For healthcare providers, this demand for information brings a valuable opportunity. Marketers can connect with patients by providing information at each step of their healthcare journey. But, what’s the best way to reach patients, and how do you measure your efforts? This guide is here to help. ## What Is Healthcare Marketing? Healthcare marketing is the practice of strategically promoting services, products, and wellness information to consumers. In addition to attracting new customers, healthcare marketing can nurture existing patients by providing information they can use at each step of their medical journey. Healthcare marketers typically spread their efforts between two major categories, traditional and digital, each with its own benefits and challenges. ## Types of Marketing in Healthcare ### Traditional Marketing The best way to differentiate between traditional and digital marketing is to think back to the days before the internet. While online marketing methods tend to dominate conversations, some traditional methods are still extremely viable. They include: - Print ads: While having marketing materials printed can cut into your budget, they can be well worth it. You can reach nearby consumers through ads in local publications like newspapers, for instance, or have brochures and posters in your waiting rooms. - TV and radio commercials: If you want to reach a large local audience with one campaign, TV and radio spots can still work well. This is especially useful for health awareness campaigns. - Billboards and transit ads: Healthcare marketers can gain high visibility with strategically placed ads on billboards and public transportation. This works best for brand recognition, since the high cost makes it tough to get a return on investment. - Direct mail: This marketing tactic tends to work best for nurturing existing customers. Postcards and reminders can be a great way to nudge patients who are due for a visit. - Community outreach: Appearances at local events like health fairs can be a great way to build community trust. When you can’t appear in person, sponsorships can get your name in front of potential customers. ### Digital Marketing With digital marketing, you’re leveraging the internet to get your brand in front of customers. While competition for attention can be tough, digital marketing provides a lower cost and the ability to gather real-time insights on those who are interacting with your brand. Here are some key types of digital healthcare marketing: - Search engine optimization (SEO): Visibility in searches is vital to getting in front of your target audience. With SEO, you design content to rank higher when patients search for providers, treatments, and other issues related to your specialty. - Paid advertising: With pay-per-click ads, marketers can boost visibility. These paid ads appear above organic search results for related queries. - Content marketing: Healthcare providers often use educational blogs, articles, infographics, videos, and downloadable resources designed to address patient questions. This type of marketing can position you as a local authority on topics within your specialty. - Social media marketing: Many consumers spend hours each day on platforms like Facebook, TikTok, and Instagram. You can reach them through posts with health tips and live Q&As. - Email marketing: If customers opt in, you can send regular newsletters and appointment reminders to keep them engaged. - Video marketing: Short-form videos and explainers can help build trust with current and future patients. You can also use video to provide instructions for upcoming procedures and answer frequently asked questions. - Online reputation management: Healthcare providers often find themselves defending their online reputation. Responding to reviews and maintaining an updated profile on Yelp, Google Business, and Healthgrades can help with that. While digital tools offer affordability and real-time analytics, traditional channels can foster brand familiarity and trust, particularly in smaller communities and with older demographics. Both have their benefits, so it’s important to look at all your marketing options and choose the best to reach your intended audience. ## What Are the Key Components of a Healthcare Marketing Strategy? ### Audience Understanding and Segmentation Each of your patients (and prospective patients) has unique needs, and your marketing efforts should reflect that: A new parent, e.g., will respond to different messaging than an older adult managing a chronic illness. In fact, 77% of healthcare advertisers make segmentation by health characteristics a top priority. With in-depth audience analysis, you can identify the unique characteristics of each customer and target your messaging accordingly. This means going beyond demographics like age or gender to look at behavioral data, search intent, health concerns, geographic location, and preferred communication channels. ### Brand Positioning and Value Proposition Trust and credibility mean everything for healthcare providers. Consumers want to connect with practitioners who will listen to them and resolve their health issues, and these days, they often track down trustworthy providers through online searches. That doesn’t just apply to healthcare. In general, 80% of consumers put experience ahead of everything else when it comes to buying from a business. That includes speed, convenience, knowledgeable help, and friendly service, and you can establish that by clearly communicating what sets your brand apart from others. Whether it’s same-day appointments, bilingual staff, compassionate care, or specialized expertise in treating certain conditions, your marketing should communicate it. ### Channel Mix: Traditional vs. Digital In today’s omnichannel landscape, healthcare marketers can benefit from an approach that combines both traditional and digital tactics. Digital platforms can give you reach, but traditional options excel at reaching local customers. That’s where an omnichannel marketing strategy comes in. With omnichannel, you combine your digital and traditional efforts for a comprehensive marketing plan. You don’t have to separate your efforts: You can move customers seamlessly from traditional to digital and back again. Here are some omnichannel marketing ideas that would work well in healthcare marketing: - Direct mail → online booking: Include a QR code or a short URL in your print mailers to make it easy for patients to book an appointment. - TV → search ads: After seeing a TV ad, 34% of consumers head straight to the company’s website, while 30% consult a search engine for more information. Search ads can capture those second-screen viewers as soon as they type in a query. - Community events → social media: First, promote your upcoming community events on social media, then share photos from the event to boost engagement. You can also include QR codes and print handouts at your event to drive in-person traffic to your online channels. - Print ads → website traffic: Print materials like brochures and flyers can include calls to action that drive consumers to your website. - Billboards → online ads: If you’re putting money into billboards, it’s important to reach consumers who see your ad while they’re unable to get online. Target consumers in the zip codes surrounding those billboards to capitalize on brand awareness. The move from TV ads to digital is one of the most common. When surveyed, most consumers admit they use a mobile device while watching TV. This is especially true of younger viewers, although it applies across all age groups. Source: Diray Media By looking at the variety of ways you can reach consumers, you’ll be able to maximize your marketing spend and reach a larger audience. Since healthcare marketers typically target local consumers, a combination of traditional and digital can be the best way to cover all your bases. ### Content and Creative Strategy Education is the cornerstone of any solid healthcare marketing strategy. Whether you’re creating downloadable guides or answering common patient questions in an explainer video, a well-executed content strategy can establish you as a trusted authority. Not only does this type of marketing establish your specialty, and you as an authority in your local community, but it also helps with search engine optimization. For example, a post titled, “When Should I Take My Child to Urgent Care?” could feature prominently in local search results when nearby patients input a query on that topic. In addition to content on your website, you can also share educational material on social media. More than 60% of consumers use social media for healthcare, with Facebook being the most popular, followed by YouTube, Instagram, Twitter, and LinkedIn. You can use this approach to nurture current customers while also reaching new ones. Source: Market.us Media ### Measurement and Optimization No healthcare marketing strategy is complete without measuring the results. Every campaign activity should be connected to a key performance indicator that you then track using the latest analytics. Today’s tools let you measure results and make adjustments to your campaigns in real time, reducing your risk of wasting time on ineffective activities. Once you have the necessary information, put it to use in optimizing all of your current campaigns. The information will also better inform your future campaigns, creating a feedback loop that ensures you’re making the most of every dollar you spend. ## Current Challenges in Healthcare Marketing Healthcare marketers have plenty of tools available, but they still face some challenges. From navigating strict compliance requirements to rebuilding trust in an era of misinformation, each hurdle requires a thoughtful approach to maintaining patient trust. Here’s a closer look at the challenges facing today’s healthcare marketers. ### Trust and Misinformation Misinformation runs rampant online, especially on the most popular social media platforms. With so many consumers turning to social media for medical advice, how do marketers establish their brand as one that can be trusted? The key is to regularly post reliable educational content that relies on medical science, citing your sources when necessary. Over time, patients will seek you out as a trusted source within your specialty. ### Privacy and Compliance Healthcare deals with stricter regulations than other industries. Marketers have to constantly navigate these regulations to ensure compliance. They include: - Health Insurance Portability and Accountability Act (HIPAA): This law applies to anyone who deals with something called Protected Health Information (PHI) on consumers. You’ll need explicit patient consent to use any PHI in your marketing efforts. - Federal Trade Commission (FTC): The FTC regulates claims made in advertising, including in the healthcare industry. Any claims you make have to be backed up with solid proof. - Federal Drug Administration (FDA): The FDA monitors advertising and marketing for misleading claims, misinformation, and deceptive practices. This includes social media platforms and the handling of product samples. Even a small, unintentional misstep can earn healthcare marketers a violation, which can lead to fines and reputation damage. If you’re using third-party platforms, pay close attention to any tools that gather patient data, to ensure they’re compliant. ### Response Expectations Today’s digital-savvy consumers have grown to expect immediate responses from businesses. That goes for healthcare providers, as well, but medical practitioners can be busy. In fact, 91% of patients surveyed say they expect a response to a portal request within 24 hours. One of the best ways to meet demand is to use the latest tools to automatically respond to chats, emails, and phone calls until you can personally interact with the patient. ### Fragmented Audience Attention You’ll face plenty of competition as you strive to capture customer attention. Standing out from the crowd means targeting prospective patients with relevant, personal messaging. Patients respond to personalization, with 79% of consumers expecting healthcare websites to offer information and services relevant to them. ## How Are AI and Data Analytics Transforming Healthcare Marketing? One of the biggest changes to healthcare marketing in recent years has come from artificial intelligence (AI) and data analytics. The global market for AI in healthcare marketing is projected to grow from $39.25 billion in 2025 to $504.17 billion by 2032, which demonstrates the importance of embracing these technologies. Instead of targeting audiences by demographics like age and gender, these technologies pack in predictive modeling, which lets you target patients by behavior and intent. Data analytics makes it easier for marketers to measure the performance of their campaigns, too. You can track every touchpoint, starting with an ad impression and going all the way through appointment confirmation, for better-informed campaigns moving forward. ## How to Create a Successful Healthcare Marketing Plan A successful healthcare marketing plan centers around your organization’s goals. Whether you’re promoting a specialty clinic, launching a new service, or striving to boost appointment bookings, your plan should be detailed but flexible enough to change over time. Here are some steps to take you through creating a successful healthcare marketing plan. ### Step 1: Define Your Audience Your first step should be to define the audience you hope to reach. You might be hoping to attract first-time patients, for instance, or caregivers. Each audience type calls for a different kind of messaging. ### Step 2: Set SMART Goals When setting your goals, go beyond wishes and hopes by setting SMART goals. SMART goals are: - Specific. - Measurable. - Achievable. - Relevant. - Time-bound. For example, instead of setting a goal to “increase web traffic,” your goal should be, e.g., to increase web traffic by 20% within the next six months. ### Step 3: Know Your Competition Before you start executing a marketing campaign, you should familiarize yourself with your most direct competition. Study those competitors to identify gaps and opportunities to make your brand stand out in the space. Analyze your competitor’s rankings, online reviews, service offerings, and ad messaging to see how your brand stacks up. ### Step 4: Choose the Right Channels No marketing channel works for every healthcare marketer. It’s important to know your audience and distribute your messaging on the right channels to reach them. Young parents might respond better to Instagram Reels and blog posts, while older patients may catch your content through TV ads and direct mail. The key is to employ an omnichannel approach and constantly monitor results, then adjust your campaigns to prioritize the most effective channels. ### Step 5: Develop Valuable Content Healthcare marketers should focus on content that provides educational value, informing, reassuring, and guiding patients through their care decisions. In addition to informative blog posts answering consumer questions, consider infographics, videos, and provider interviews that explain conditions, treatments, or what to expect from a visit or procedure. ### Step 6: Build and Optimize Conversion Pathways Have you followed the customer journey through an interaction with your brand? View your funnel through a typical consumer’s eyes, from seeing an ad or post to making an appointment and joining your patient portal. Make sure your landing pages are mobile-friendly, your forms are simple, and your calls to action are clear and accessible across all channels. ### Step 7: Ensure Legal and Regulatory Compliance Familiarize yourself with all regulations that apply to healthcare marketing, including HIPAA, FTC, and FDA guidelines. Be careful not to make promises that can’t be backed by solid science, and regularly audit your third-party providers to ensure they’re compliant with privacy requirements. ### Step 8: Test and Optimize Regularly A/B testing can help you pinpoint the type of messaging that works with your audience. Test different subject lines, ad creatives, and landing page designs and pivot in response to what you learn from each test. ## Leveraging Social Media Platforms to Reach Patients Social media is a great way to connect with your patients and potentially reach new customers. It offers a direct connection to your audience, who likely spends a great deal of time each day on their favorite social media platform. Whether your target demographic is on Facebook, Instagram, TikTok, LinkedIn, or YouTube, you can use social media to connect with users and establish brand authority. Social media isn’t just for marketing outreach, though — today’s consumers use it for customer service queries. If you can be responsive on your chosen social media channels, you have a great opportunity to give fast, friendly service and cement your reputation as a top provider in your area. ## Key Takeaways Effective healthcare marketing requires a strategic but comprehensive plan that puts patients first. For best results, marketers should implement an omnichannel approach to reaching new customers and connecting with existing patients. From segmenting customers based on health needs to monitoring your results and adjusting future campaigns accordingly, success in healthcare marketing comes down to building trust and authority. Social media, AI tools, and performance-driven creative formats give brands powerful ways to connect with patients and foster long-term loyalty. ## Frequently Asked Questions (FAQs) ### How can healthcare providers build a strong online presence? A professional online presence starts with a well-designed website that clearly communicates your services, hours, specialties, and contact options. You should also make it easy for patients to make an appointment, whether they book it through a scheduler, call your offices using contact information on your site, or complete an email form to request an appointment. Healthcare providers should also maintain an updated Google Business profile and ensure they’re listed with Healthgrades and Yelp. ### What are the best tools for healthcare marketing analytics? You’ll find a variety of helpful tools to support your marketing efforts. Google Analytics 4 can help you monitor web traffic, user behavior, and conversion paths. Many organizations also rely on customer relationship management platforms like Salesforce Health Cloud and HubSpot to manage patient relationships. These tools also come in handy for marketers, who can use them to track engagement across email and SMS campaigns. You can also use A/B testing tools like Optimizely to try out various creatives. Performance advertising platforms like Realize offer multiple ways to utilize data effectively, including AI-powered, real-time testing and optimization. ### How can marketing improve patient retention rates? Marketing isn’t just for attracting new patients, it can also be a great way to nurture existing customer relationships. Timely appointment reminders, check-up notifications, and seasonal help tips can keep patients engaged between visits while also establishing yourself as a trusted resource. Ultimately, consistent, empathetic communication helps strengthen patient relationships, making them more likely to stay loyal. ### Is paid advertising effective for healthcare providers? Paid advertising can be very effective for healthcare providers, particularly well-targeted and regulation-compliant ad messaging. When paired with the right creative, paid ads can help you reach clients actively looking for care. When combined with user-friendly landing pages and strong calls to action, paid ads can also boost appointment bookings and help with new patient acquisition. --- ### AI-Powered Lead Generation on the Open Web: Insights from Xevio URL: https://www.taboola.com/marketing-hub/ai-powered-lead-generation/ Last Modified: 2025-09-01 08:36:11 If you read our series of e-commerce interviews with Xevio co-founder and CEO, Nadim Kuttab, you already know he has a ton of useful advice when it comes to getting your brand out there. In this series, we’re going to be discussing all things lead generation, starting with the ways in which AI can be harnessed to identify high-converting prospects beyond traditional targeting methods. This shift from broad demographic targeting to behavioral-based lead identification is fundamentally changing how lead gen campaigns are being conceived and executed on the open web, so we were keen to get Kuttab’s insights. How has the landscape for lead generation evolved on the open web, particularly with the rise of AI? There's a ton that I think is going to change in the lead generation game in the coming years. You already have call centers becoming more automated, with lead verification or interest verification being handled by AI agents and automated callers. Everybody wants a quality lead, and that becomes easier when you use AI to pre-qualify people. Because that step is essentially being automated, we as advertisers are getting faster signals regarding quality, which allows us to scale and grow our lead gen efforts exponentially faster. The more data we have, the quicker we're able to optimize, and the better it is for lead quality and scale. Lead gen is going to be a big spender in the coming years on anything open web-related. We've seen it time and time again at Xevio: We're able to generate 50,000 to 100,000 leads a month on channels like Realize — it's unbelievably good quality if you do it right. For lead generation on the open web, native advertising and Taboola has always worked well, and it'll just continue getting better now with AI. From your experience, what are the biggest misconceptions marketers have about leveraging AI for lead generation? Everybody always thinks AI means trash, right? Like, they're going to have a very bad automated call center that tries to sell you something. When you use AI properly at any step in the lead generation funnel — whether it's in the ads, the content, the call centers, or the lead qualification — it can be extremely beneficial and prove to be a nice experience for both parties. The leads are warmer, but also the person who has demonstrated interest in the service you offer gets connected to the right people and gets a good quote. We all want to be connected to the right people! AI helps us accelerate that process. Realize's Predictive Audiences tool identifies users likely to convert, based on first-party data. How can advertisers best prepare their data to maximize the effectiveness of such AI tools? The challenge for companies like Taboola has always been, how do I get first-party data? Because people don't have to log into anything. With Meta and Google, when you're using their apps or their platforms, you're constantly logged in and identifiable. You don’t have that luxury on the open web, so the fact that Realize can now segment data based on user behavior, contextual settings, device types, location,etc., is extremely promising, because we know that if you show your ad to somebody who is likely interested in the ad, it's beneficial for everybody. In terms of advertisers using their own data properly, they need to pixel everything and send whatever signals they get back to the platform. It's the most efficient way — the best data is the data that people can collect directly on their platform. We talked about this a little bit last time, the idea that the more you put into it, the more you're going to get out of it. Right! We’re moving into a post-media buying era of marketing where you don’t need to micromanage everything: The AIs and softwares and algorithms will do that for you. But, you need to give them the right signals. If you don't feed Realize the right data, it won’t know what you're going after. It’ll be slower at identifying the right consumer pockets and targeting them with ads, therefore it'll drive up your costs, and others will do it more efficiently than you. Feed data into the system across your funnel to more effectively optimize. The more data, the better. Can you share an example of how a brand successfully used advanced audience prediction to unlock a new, high-converting lead source they hadn't previously considered? Well, we see it as incremental. Predictive audiences are always going to be a lot smaller in scale than when you go broad — if you have a broad campaign doing €30,000 or €40,000 a day on Realize, using the Predictive Audience feature might add a couple of thousand euros at a very good CPA, but it's not going to help us triple our spend. So, we see it as an incremental way to reduce overall cost: If we're halving our CPA with Predictive Audiences, we can afford to buy cold or broad users that are a little more expensive. Basically, it lets us spend more and increase our reach. Our clients love it, because it's effective customer acquisition. What are the critical metrics performance advertisers should focus on when evaluating the success of AI-driven lead generation campaigns? I think the phrase “AI-driven” suggests that you're doing it without AI, and if you're currently advertising without AI in any step of your funnel, you're probably not doing it right. Most lead generation nowadays is in some way, shape, or form AI supported. Either way, the critical performance metrics will be the exact same thing as your normal performance metrics: How good are your campaigns performing? If you're able to insert AI into any step — it can be in the ad creation, in the content piece, in the funnel, in the post-lead process — and you're able to improve the metrics on that step, then great. I don't think there's one thing that you should look at with AI and lead generation — at the end of the day, the question is, for every dollar you're putting in, how much are you getting out? That is the most important metric. Looking ahead, how do you see AI continuing to shape the future of lead generation, especially with evolving privacy standards? There's AI that you see and AI that you don't see. It’s the latter where a lot of the magic is going to happen. It’s like when a user types in their data — there's an AI that's already scoring that lead and passing that data back on to the traffic source for optimization in real time. Those are things that the people who fill out the forms won't even notice, but it's happening. It will be really amazing to see the speed at which we're able to become more efficient thanks to the use of AI. With evolving privacy standards, that's the visible side of AI, right? People won't want to talk to AI agents when it’s done badly, and that's not going to change any time soon. Every time I'm in an automated call and I have to press one or two or tell them what I want, my blood level spikes. It's a terrible user experience. But, if you're able to integrate it seamlessly to make the call faster and more efficient, everybody wins. The effective use of AI will speed up processes, then, make it more seamless for us to get from A to B, whether it's as an advertiser or a user. Obviously we'll have a lot of inefficient or bad use of AI that'll drag down the perception of it, but it’s important to really focus on how to use AI to seamlessly improve processes and improve revenue per dollar, without upsetting people by overusing it and plugging it in everywhere just because you can. As a user, AI shouldn't be something you even think about. It’s AI as a sort of invisible helper. Yes. It should be like predictive texting: You're writing something and then it finishes it, it's not annoying and you can continue typing if you want and ignore the suggestion. That is the future of AI, it will shape a lot. You mentioned that all marketers should be using AI in some form these days, but for advertisers new to AI-powered lead generation, what's one actionable step they can take on a platform like Realize to get started? Use AI to take the ads you're running on platforms like Meta or Google to bring them to Realize — that's the easiest step. One of the challenges that Taboola was facing for years was, how do we bring people from Meta or Google to Taboola? AI makes that step so much easier. Realize also has the Social Importer tool for doing exactly that. Exactly! You guys have a million ways to do this, and they all work — that's the truth. So, that's the Social Importer is the easiest starting point. Will it be successful? Maybe, maybe not. But, it'll give you an idea of what you have to work on next. If you missed the e-commerce series, no problem — you can read them here: - Driving Direct Response Sales on the Open Web - Building High-Converting eCommerce Funnels with Native Content - Beyond Last-Click — Measuring Engagement on the Open Web --- ### 5 Black Friday Marketing Strategies for Performance Ads Success 2025 URL: https://www.taboola.com/marketing-hub/black-friday-marketing-strategies/ Last Modified: 2025-08-26 10:59:20 Black Friday 2025 is on the horizon, and if you’re in the advertising game, you know this is our Super Bowl. It’s not just about having great deals — it’s about getting those deals in front of the right eyes, at the right time, in a way that makes people actually click. As an advertising copywriter who has navigated (too) many of these holiday frenzies, I’ve learned a thing or two about what works, what doesn’t, and what simply gets lost in the digital chaos. This year more than ever, advertisers need to be smarter, more creative, and more strategic. So, let’s talk about how to build a Black Friday marketing strategy that truly performs and hits your goal numbers. ## Tailor Your Black Friday Campaigns with these 5 Strategies ### 1. Start Early With Mobile-First Content and Urgency The notion that Black Friday begins the Friday after Thanksgiving is severely outdated. Nearly half of holiday shoppers plan to shop before November, myself included — this means your campaign needs to be a marathon, not a sprint. My advice: Publish mobile-first holiday content by October. Mobile devices now account for 55% of holiday e-commerce, and mobile holiday revenue now surpasses desktop earnings. Your customers are glued to their phones, so your content and landing pages should be optimized for that experience. Launching holiday-specific, mobile-optimized pages early helps you reach those eager pre-season shoppers. Headlines that encourage it, like "Beat the Holiday Rush" or something similar, create that early sense of urgency and exclusivity. This initial push sets the stage for capturing early intent and building momentum. ### 2. Embrace Creative Trends: AI, Authenticity, and Action Your ad visuals are your first impression, and during the holidays, they can boost your Conversion Rate (CVR) by a significant margin. Holiday-themed creatives lead to a much higher CVR than generic ones, but it’s not just about shoving holiday cheer down people’s throats and calling it a day. There are key creative trends to leverage, such as: #### AI-Driven Creative Variations Artificial intelligence enables faster creative variations, driving further optimizations at scale. This means you can quickly test different visuals and messages to see what resonates best with your audience. #### Unfiltered, UGC-Style Authenticity In a world saturated with polished ads, DIY imagery can build trust and feel more relatable than traditional stock photos. This user-generated content (UGC) style can make your brand feel more genuine. #### Looping, Short-Form HD Motion Seamless, human-driven videos capture attention and bring products to life. Think about showcasing your products in real holiday moments, like decorating or cooking, to create an emotional connection. My personal take is that if it feels too "ad-like," it might not perform as well. Aim for content that feels natural and is integrated into the user's feed for a feel that’s real and organic. ### 3. Personalize Content and Gamify Calls-to-Action (CTAs) One-size-fits-all rarely works, especially when shoppers are looking for very specific gifts. Instead, personalization is key. Craft content that speaks to specific audiences, incorporating interactive elements such as: #### Personalized Quizzes and Gift Guides Quizzes and lists engage shoppers and simplify holiday buying decisions. It’s an easy go-to that’s been shown to work. Consider a "Find Your Gifting Style" quiz that leads to a personalized gift list. #### Gamified CTAs Turn interaction into a game. Instead of just "Shop Now," think about "Play to Shop". A quiz like “What Should You Gift Your Hard-to-Buy-for Family Member?" can engage users before directing them to purchase, while keeping them entertained throughout. You can have some fun writing it, too, and readers will pick up on that. ### 4. Prioritize Discounts and a Fast Checkout Process Deals are the main attraction here, so don’t lose sight of that, or the fact that most shoppers are looking for deals throughout the holiday season — not just last minute. Your ads need to prominently display urgent promotions to catch their attention, so be sure to highlight savings and slashed prices clearly. Equally important is a frictionless checkout. After you bring in customers with compelling offers, the last thing you want is for them to abandon their cart due to a clunky experience. Emphasize and communicate how easy the process is, and then be sure to deliver on those promises. Every step removed in the checkout process can significantly boost conversions, so make it smooth and simple. ### 5. Implement a Robust Campaign Checklist and Strategy Success on Black Friday isn’t an accident or a fluke — it takes careful and meticulous planning, and a clear strategy. Here’s a checklist of critical steps: #### Implement Tracking 4–6 Weeks in Advance Start collecting data on users who interact with your brand way ahead of peak periods. This early data is going to be invaluable for optimizing your campaigns. #### Integrate Quickly Take advantage of seamless pixel setup and automatic conversion tracking with integrations for platforms like Shopify and WordPress. This takes away the headaches that come with manual coding and theme limitations, allowing you to focus on strategy. #### Start Campaigns Early With Maximize Conversions Launch your campaigns at least five to seven days before peak, using the Maximize Conversions bidding strategy. Start with a daily budget up to 50% less than your target peak budget. #### Set Adequate Budget and Plan for Increases Once you reach 50 conversions, or two to three days before peak, gradually increase that budget. Generally, it’s recommended to increase by no more than 50% at a time to protect your CPA. Also ensure your budget is adequate for the entire campaign duration, especially during peak days. Be prepared for natural CPA and CPC fluctuations due to the competitive holiday landscape. #### Avoid Major Changes During Peak Time Keep targeting, bidding, and other settings stable during peak days. This is crucial to avoid re-triggering the learning phase of your campaigns. #### Monitor Creative Performance Always maintain four to six creatives in rotation and pause any underperforming ones. Continuous monitoring and optimization are key to maximizing your return on ad spend. ## How Realize Makes Black Friday Marketing Easier Implementing these strategies can feel like a lot, especially during the insanity of Black Friday. That’s where a platform like Realize comes into play: Built as a performance engine specifically for driving conversions, it offers creative formats and placements beyond traditional native ads, providing diverse ways to captivate audiences. It can help you publish mobile-first content, allowing you to make your holiday creatives festive, authentic, and kinetic. Realize also supports the integration of personalized quizzes and gift guides, helping you to tailor content and gamify CTAs effectively. “Realize’s GenAI AdMaker enables advertisers to write up holiday-specific headline copy, rework images, and create motion ads out of them, as well as generating new visuals that are holiday themed,” says Maayan Leshem, Taboola’s director of creative shop and AI strategies. Realize also offers: - Codeless Conversions (Beta): Easily create conversion events like “Add to Cart” or “Start Checkout” in Realize — no developers or GTM required. - Tracking Test Tool: Validate your tracking implementation in real time, and ensure your Taboola data matches external reports — a crucial step for reliable performance insights during peak season. - Retarget Engaged Audiences Using Taboola Pixel: Run engagement campaigns before conversion campaigns to capture interest and build a warm audience. Use a simple, compelling hook to attract users early, then guide them down the funnel to convert more efficiently. - Reach New High-Value Shoppers: Use tools like Predictive Audiences to extend beyond your core audience and target high-intent users likely to convert during peak season. ## Key Takeaways Starting early is crucial: Begin your Black Friday campaigns by October with mobile-first content to capture early shoppers. Use holiday-themed, authentic, and dynamic visuals, leveraging AI for optimization where you need it, and implement quizzes and tailored content to create interactive experiences. Highlight discounts and ensure a fast, frictionless checkout process. And don’t stop with Black Friday — target audiences post-holiday for continued engagement. ## Frequently Asked Questions (FAQs) ### How do you attract customers on Black Friday? Customers are diligently looking for deals during this period, so focus on compelling offers, create a sense of urgency and scarcity, use visually appealing and holiday-themed creatives, optimize for mobile shopping, and start promoting your offers early. Personalization through gift guides or quizzes can also significantly boost engagement by helping customers find exactly what they need. ### Which marketing strategy is most commonly used by fashion brands during Black Friday? Fashion brands often lean heavily into visual-first marketing strategies during Black Friday. This includes showcasing their products through high-quality, authentic imagery and short-form video content that resonates with the holiday vibe. They also frequently use personalized recommendations, quizzes to help customers “find their style,” and exclusive early access deals or bundles for mobile shoppers to drive conversions. ### What are the top three most sold categories on Black Friday? While specific data varies year to year, traditionally, the top-selling categories on Black Friday include electronics, apparel/fashion, and home goods. These categories often see some of the steepest discounts, driving high consumer demand. --- ### The 2026 E-Commerce Holiday Calendar: Your Guide to Seasonal Marketing URL: https://www.taboola.com/marketing-hub/ecomm-holiday-calendar/ Last Modified: 2026-03-31 11:50:13 Each year, e-commerce businesses have an opportunity to boost sales during key shopping dates, but in 2026, timing and execution matter more than ever. From New Year’s resolutions to Boxing Week deals, consumers are shopping earlier, comparing more options, and responding to highly targeted campaigns. Whether it’s fitness equipment in January, outdoor products in spring, or major discounts during Black Friday, marketers can reach specific audiences with customized strategies throughout the year. The calendar below highlights the most important e-commerce holidays, along with opportunities to plan, launch, and optimize campaigns when demand is at its highest. What’s changed in our 2026 update: - All dates and holidays updated with 2026 information. - New advice added specifically for changes to holiday commerce in 2026. ## What’s Changed for E-Commerce Holiday Marketing in 2026? While the key shopping dates remain consistent year to year, the ways in which consumers shop and how marketers reach them have evolved significantly. For example, many brands are now launching campaigns earlier, using the weeks leading up to major holidays to test messaging, creative formats, and audience segments. This allows them to identify what resonates before competition peaks. Meanwhile, AI tools have made creative production faster and easier to repeat. Marketers can test multiple ad variations and optimize performance in real time. As a result, they’re able to scale top-performing campaigns more quickly than in previous years. Competition has also increased across traditional channels, as advertisers expand into additional platforms to maintain visibility. The result is a more dynamic, data-driven approach to holiday marketing, one that rewards preparation and adaptability. ## E-Commerce Holidays and Dates in Q1 The first quarter of the year (January through March) is about fresh starts, and shopping trends reflect this, as consumers focus on products that can boost their productivity and self-improvement. Shoppers are also looking for big deals on leftover seasonal products. ### New Year’s Day – Thursday, January 1 Who it's for: Health and fitness enthusiasts, self-improvement seekers, and regular consumers. For many, New Year’s Day represents a fresh start. As people embrace new fitness and nutrition habits, fitness equipment, nutritional supplements, and other wellness products tend to be in high demand. Productivity tools and subscription services also do well, as people look to become more efficient. ### Martin Luther King Jr. Day – Monday, January 19 Who it's for: Shoppers looking for deals, regular consumers. Martin Luther King Jr. Day was created to honor Dr. King’s legacy, so marketers must be sensitive when executing their marketing tactics for the holiday. Brands can offer sales on a wide range of product categories, including clothing, electronics, and home decor. Additionally, because the holiday falls shortly after the holiday season, it presents an opportunity for brands to clear out excess seasonal inventory by offering strategic discounts. ### Valentine’s Day – Saturday, February 14 Who it's for: Couples, gift-givers, jewelry shoppers. Valentine’s Day is a celebration of not only romantic love, but also of friendship and affection. While everyone knows that flowers, chocolates, and jewelry sell well, there is increased interest in experiential gifts, such as spa and travel packages, and self-care items like self-help books and skin care items. ### President’s Day – Monday, February 16 Who it's for: Budget-conscious shoppers. President’s Day is one of several three-day holiday sales weekends during the year. It’s well-known for massive sales, particularly with big-ticket items like mattresses, home appliances, and furniture. As such, marketers can promote significant discounts with messaging like “this weekend only” or “limited-time.” Like Martin Luther King Jr. Day, it’s also an opportunity to clear out any leftover stock from the holiday season. ### St. Patrick’s Day – Tuesday, March 17 Who it's for: Party planners, green apparel aficionados. St. Patrick’s Day is a fun holiday, so festive attire, accessories, and party supplies always do well. E-commerce businesses can promote green-themed clothing, Irish-themed decorations and home decor, and various novelty items. When it comes to marketing campaigns, e-commerce brands should aim to make their marketing campaigns fun in order to draw in shoppers who are looking for unique products. ## E-Commerce Holidays and Dates in Q2 In the second quarter of the year (April through June), consumers are beginning to think about the summer months ahead. As a result, shopping trends tend to favor outdoor products and activities. There are also some key family events, like Mother’s and Father’s Day. ### Mother’s Day – Sunday, May 10 Who it's for: Children, spouses, gift-givers. Mother’s Day is a big event for retail, including e-commerce businesses. Consumers are looking for products that show appreciation to the mothers, grandmothers, and other maternal figures in their lives. In addition to traditional Mother's Day gifts like flowers, jewelry, and beauty products, more consumers are looking for personalized, thoughtful gifts. E-commerce brands that specialize in engraved jewelry and customized photo gifts can do very well. Experience-based gifts, like spa packages and weekend getaways, are also very popular. ### Memorial Day – Monday, May 25 Who it's for: Outdoor enthusiasts, military families. With Memorial Day falling so close to the start of summer, there is increased interest in outdoor items, such as patio furniture, outdoor lighting, barbecues, and coolers. Consumers are also looking for tools for around the home, like lawnmowers and gardening supplies. They’re also looking for deals on travel-related purchases, like camping gear and luggage. As such, many marketers tie their promotions to two themes — the holiday weekend and the start of summer. ### Juneteenth – Friday, June 19 Who it's for: Shoppers supporting social justice and cultural products. Juneteenth, which commemorates the end of slavery in the U.S., has long been celebrated, but wasn’t made a Federal holiday until 2021. It presents an opportunity for marketers who promote products that support Black-owned businesses. Popular items include books, art, fashion, and home decor with cultural significance. The holiday is also an opportunity to offer sales on products that support social justice causes, especially where a portion of proceeds goes toward charitable organizations. That said, marketers should exercise sensitivity when approaching Juneteenth-based campaigns to avoid appearing superficial or inauthentic, which could damage their reputation. ### Father’s Day – Sunday, June 21 Who it's for: Children, spouses, tech and hobby enthusiasts. Father’s Day is similar to Mother’s Day in its focus on gift-giving, but the types of products vary. Tech gadgets, tools, and outdoor gear are always popular with Dads, as are products related to hobbies like sports, fishing, and grilling. Marketers can capitalize on this demand by planning their Father’s Day campaigns around big-ticket items. ## E-Commerce Holidays and Dates in Q3 In the third quarter (July through September), shoppers are dealing with the summer heat and the back-to-school rush. For e-commerce businesses, this means a continued focus on outdoor products as well as a shift toward back-to-school supplies and other educational items. ### Independence Day – Saturday, July 4 Who it's for: Patriotic shoppers, outdoor enthusiasts. Independence Day offers a massive opportunity for businesses selling outdoor-related products. Grills, camping gear, coolers, and backyard pools are all in high demand as families celebrate Independence Day with outdoor activities. Patriotic-themed clothing, flags, and accessories are also popular. Many retailers will promote these products with significant discounts. ### Amazon Prime Day – Mid-July Who it's for: Loyal Amazon shoppers, budget-conscious consumers. Amazon Prime Day has become one of the largest e-commerce events of the year, as Amazon Prime members look for huge discounts on a wide range of products. All sorts of electronics, home gadgets, and beauty and fashion items rank among the most popular categories. Marketers should be prepared to offer substantial discounts in order to compete with all the deals that will be available. ### Labor Day – Monday, September 7 Who it's for: Shoppers looking for end-of-summer deals. The Labor Day weekend comes with big sales in a wide range of product categories. Shoppers can score end-of-season discounts on summer clothing and patio furniture, and sales on travel gear, mattresses, and electronics. Many consumers are also looking to purchase school supplies or prepare for autumn. For marketers, there’s ample opportunity to offer promotions across all of these categories. ### Back-to-School – Late August-Early September Who it's for: Students, parents, teachers. Back-to-school shopping is one of the most predictable retail events of the year for marketers. There is always a strong focus on school supplies, electronics, and clothing. Parents often need to replace backpacks, notebooks, laptops, and other items for their kids. Students also love receiving new tech gadgets, such as tablets and smartphones. If you’re a marketer in the education or technology niche, you can leverage this demand by offering promotions for students and parents alike. ## E-Commerce Holidays and Dates in Q4 The fourth and final quarter of the year marks the approach of the holiday season, which is packed with key shopping dates and major sales events. ### Halloween – Saturday, October 31 Who it's for: Families, party planners, costume buyers. Halloween has become a major event on the calendar, and as such, it drives sales in several product categories. It’s also worth noting that Halloween is no longer just a celebration for kids — it’s equally popular among adults. Of course, costumes are top sellers, but Halloween fans are also looking for accessories, themed party supplies, and decorations. Overall, Halloween is a great opportunity for e-commerce businesses to promote limited-time offers and create themed marketing campaigns. ### Veteran’s Day – Wednesday, November 11 Who it's for: Military families, bargain hunters. Veteran’s Day is primarily a day to remember and reflect. However, retailers have an opportunity to honor veterans and their families with special discounts. For example, many businesses offer exclusive deals to veterans and active-duty personnel on a wide range of products, including big-ticket items such as electronics and home goods. Marketers can tie these offers to themes of patriotism and national pride. ### Thanksgiving Day – Thursday, November 26 Who it's for: General shoppers, early shoppers. Thanksgiving Day may be a holiday centered on family, but it marks the beginning of the shopping rush leading up to Black Friday. Shoppers are gearing up for major Black Friday deals, and marketers can capitalize on this by offering pre-Black Friday discounts or early-bird promotions. Many brands promote products they know will be in high demand during Black Friday and Cyber Monday, such as electronics, clothing, and toys. ### Black Friday – Friday, November 27 Who it's for: Shoppers in search of deep discounts. Black Friday continues to be one of the biggest shopping days of the year for e-commerce businesses, driving significant spikes in traffic and conversions. However, in recent years, it has expanded beyond a single day, and many brands are launching promotions well in advance. Deep discounts on high-ticket items like electronics, appliances, and gaming consoles remain a major draw, along with deals on home goods, toys, and clothing. Marketers use “door-crasher” deals, exclusive promotions, and flash sales to create urgency and drive conversions. Given the level of competition, marketers need to plan ahead, test their campaigns early, and aim for high visibility across multiple channels in the days and weeks leading up to peak shopping periods. ### Cyber Monday – Monday, November 30 Who it's for: Online shoppers, deal hunters. Cyber Monday is geared specifically for online shopping and features deep discounts on tech products, fashion, and home goods. It’s an excellent opportunity for brands to offer online-only promotions, with fast shipping options. They may also want to push excess stock left over from Black Friday. ### Christmas Day – Friday, December 25 Who it's for: Gift-givers. Christmas marks the culmination of the holiday shopping season. While Christmas Day itself may not see much shopping activity, consumers continue to purchase gifts in the weeks leading up to the holiday. Popular gift categories include electronics, toys, clothing, home goods, and personalized gifts. Marketers can use promotions and fast-shipping options to help drive a surge in last-minute sales. ### Boxing Day – Saturday, December 26 Who it's for: Bargain hunters, post-Christmas shoppers. Boxing Day is a major shopping holiday in several countries, including the U.K., Canada, Australia, and New Zealand. It’s generally a time for clearance sales, with many retailers offering discounts on items that didn’t sell before Christmas. Customers can find deals on holiday decor, clothing, electronics, and home goods. Due to the deep discounts offered, many people will use Boxing Day as an opportunity to treat themselves after Christmas, as well. Boxing Day sales often extend for the entire week between Boxing Day and New Year’s Day, with many brands promoting Boxing Week sales. ## Key Takeaways By aligning marketing campaigns with consumers’ evolving needs and behaviors, e-commerce businesses can create more effective and timely promotions. While some product categories, like electronics, clothing, and home goods, are consistently popular, each holiday presents a unique opportunity for brands to offer tailored promotions and foster stronger connections with their customers. Brands that plan ahead, test their campaigns early, and adapt their strategies in real time are better positioned to stand out during high-competition periods. By using this calendar as a guide, you can build campaigns that not only align with key dates, but also perform more effectively in an increasingly competitive landscape. ## Frequently Asked Questions (FAQs) ### What is the best promotional strategy for the Christmas season? During the Christmas season, brands should focus on offering early promotions, limited-time discounts, gift guides, and personalized offers. Shoppers are often searching for discounts in the weeks leading up to Christmas. This gives brands an opportunity to create urgency with their messaging. For example, you could use “while supplies last” to encourage customers to buy sooner rather than later. You can also offer free shipping, bundled discounts, and loyalty rewards to boost your conversions. ### What are the top five items consumers are expected to buy in celebration of the Christmas holiday? During the Christmas holidays, top items include gift cards, electronics, clothing, toys, and home decor. In fact, many of these items sell well during most major holidays. ### What do most people consider the official start of the holiday shopping season? Black Friday is considered by many to be the official start of the Christmas holiday shopping season. Increasingly, however, brands are creating sales events leading up to Black Friday. There is also an increasing focus on shopping days at different times of the year, such as Amazon Prime Day, which occurs in mid-July. ### What are the top three retail companies during the holiday season? Unsurprisingly, retail giants Amazon, Walmart, and Target are usually the top performers during the holiday season. All three companies offer a wide range of products and deep discounts. --- ### Beyond the Banner: 7 Creative Ad Strategies for Q4 Success URL: https://www.taboola.com/marketing-hub/holiday-season-creative-strategies/ Last Modified: 2025-07-29 07:46:08 Q4 can easily become a frenzy for marketers and advertisers trying to tap into holiday sales without going overboard on campaign spending. Shoppers are browsing much earlier than December for holiday purchases, and nonlinear buying journeys add uncertainty and additional challenge. Plus, the sheer volume can seem overwhelming: Holiday spending in the U.S. in 2024 was projected to reach $1.59 trillion, and 53% of shoppers in 2025 plan to spend about the same as last year. I’ve gathered some top creative strategies to help you cut through the noise of Q4, with the help of Taboola’s Creative Team, who constantly tracks and analyzes data on the latest trends for marketers and advertisers to inform this year’s holiday playbook. ## Holiday Shopping Creative Strategies for Q4 Success ### 1. Embrace Interactive Formats Marketers can try a range of interactive, engaging formats to reach new prospects among the crowd. Polls, quizzes, and playable ads can all stand out online, and they have the benefit of combining increased engagement with data collection. Consider how you can personalize interactive content and gamify calls to action (CTAs). This ad, for example, calls back to 1980s video games with its design, and offers a fun and easy way to connect with the brand. ### 2. Time It Right The holiday season now starts long before December, so marketers should start planning campaigns well ahead of Q4. National Retail Federation data finds that nearly half of holiday shoppers — 45% — plan to shop before November. Spending will vary widely by industry and business size, of course. “Our clients in education and B2B often prioritize spending ahead of the holidays, heavily in October and early November, then scale back for December,” says LaVana. It’s important to pay attention to browsing and shopping patterns, too, in order to time your campaigns right and attract users whether they’re in research or active buying mode. “We’ll often see a spike in research for education over the holidays due to more downtime, so we do maintain visibility,” adds LaVana. ### 3. Leverage User-Generated Content (UGC) User-generated content remains a great way to attract attention for your brand without having to invest a lot. It can also bring authenticity and social proof to a holiday campaign. You might, for example, use DIY candid imagery to build trust, rather than using samey-looking stock photos. Consider ways to incorporate less filtered and produced content in general, so readers can relate more to your brand and products. And, if you’re working with influencers or affiliates, line up their participation early for your holiday strategies. ### 4. Use Retargeting Wisely As with any other time of year, be focused on your ideal customer profile (ICP) during the holidays, advises LaVana. “Focus on retargeting efforts to previous site visitors or target email or company lists that you know are the right fit for your product or service,” she says, adding that it’s also essential to look for high-intent users who are actively searching for your product, and to prioritize those efforts versus demographic targeting on oversaturated social channels. ### 5. Tell Stories Through Short-Form Video There’s plenty of opportunity to create short-form videos for Q4 shoppers. Mobile devices make up 55% of holiday e-commerce, outpacing desktop revenue, and that number is only expected to grow. Social media platforms like TikTok and Instagram make it easy to post multiple short videos, engaging with influencers and others where applicable. https://www.youtube.com/watch?v=Rx0om7rWhl0 For slightly longer, memorable holiday video content, consider how you can infuse emotion and connection: Think Apple’s 2021 Christmas commercial, or Etsy’s holiday campaigns that focus on authentic gift-giving. https://www.youtube.com/watch?v=h4TpS0MDHfA&t=1s ### 6. Personalize at Scale Q4 is the time to put all that hard-earned data to work in reaching your leads, customers, and high-intent users. Use the audience segments you have, or tweak existing ones for holiday purposes, and see how you can get creative. For example, use a gamified ad to quiz users on their holiday style, then provide personalized lookbooks accordingly, such as one for a party or event shopper. ### 7. Embrace the Power of Scarcity and Urgency Used well, scarcity and urgency can help move a prospect toward a purchase in Q4. It’s important to communicate limited-time offers or low-stock notices without being too aggressive and turning off buyers. Consider how both copy and imagery can help convey a sense of urgency, as well as get holiday shopping on users’ radars early to make sure they have a wide selection. Simple images early in Q4, such as this Christmas-adjacent one of socks, can spark the shopping urge. ## Key Takeaways Marketers looking for creative ad strategies in Q4 should set the course early, since holiday shopping starts much earlier than December. Creative formats and placements that go beyond the banner can help brands succeed: Consider short videos, gamified ads and CTAs, and personalization, and ensure that you start holiday marketing well ahead of time for best results. In 2025, we’re well beyond static banners as the best way to engage new and returning holiday shoppers. While marketers have a lot of options to succeed in a holiday market, it’s not always easy to determine what will deliver the best performance. One rule holds true, though: As in any other time of year, it’s essential to put your customer first. “Know your audience! Prioritize who they are and where they are to break through the clutter,” says Jackie LaVana, founder of 126 North Marketing. “Make sure your imagery and messaging are clear and stand out.” --- ### How to Prep for Cyber 5: 6 Strategies to Win the Season URL: https://www.taboola.com/marketing-hub/how-to-prep-for-cyber-5/ Last Modified: 2025-08-26 11:16:10 The Cyber 5 period, which includes the five days from Thanksgiving Day through Cyber Monday, is one of the highest-stakes shopping periods of the year. With consumers highly motivated, competition among retailers intensifies, and digital ad space becomes a battlefield where marketers fight for attention. But, while Black Friday and Cyber Monday have become synonymous with deals, marketers need to do more than offer deep discounts if they want to win Cyber 5. To help you gain an edge this year, I’ve consulted with marketing experts and drawn on Taboola’s own Realize headline performance data to identify six crucial ad strategies. ## 6 Strategies to Prepare for Cyber 5 ### 1. Create High-Impact Headlines Marketers can no longer rely on generic headlines or discounted pricing when building Cyber 5 ad creatives. Taboola recently analyzed the performance of previous Black Friday and Cyber Monday headlines from Realize and found that the highest-performing headlines connected with consumers on a deeper level. While Direct Discount headlines still resonate, marketers must include a mix of curiosity-driven, problem-solving, and urgency-based messaging in headlines to stand out. These “hybrid” headlines work well because they connect with shoppers on a human level by solving a real problem or sparking their curiosity. For example, here’s how you could transform a generic headline (using Dyson as the brand) into a high-impact headline that incorporates curiosity-driven, product-solution, and urgency elements: - Generic: “Dyson Black Friday Deals — Shop Now!” - High-Impact: “Black Friday Sale Ends Tonight: Save 40% on Dyson — Breathe Cleaner Air Before Winter Hits” As Esther Altomare, head of industry enterprise advertising at Taboola, puts it, “The key is balancing intrigue with clarity, making the benefit obvious while evoking just enough curiosity to prompt a click.” ### 2. Plan and Stress Test Creatives Early If your ad campaigns haven’t been tested in advance, small cracks can become major problems. As such, by the time Cyber 5 arrives, your ad creative should already be a proven winner. According to Aaron Whittaker, VP of demand generation at Thrive Internet Marketing Agency, “The brands that win aren't the ones optimizing in November, but the ones that have started tracking lead indicators by late summer. I have personally witnessed campaigns where a mere 0.5% lift in click-through rates during September will lead to a double-digit revenue increase over Cyber 5.” The key takeaway here is that you should test early and address any issues before November. That means preparing for Cyber 5 months in advance, rather than performing a last-minute sprint. As David Quintero, CEO of NewswireJet puts it, “By the time Black Friday hits, testing season is already over. Enter peak season with high-performing assets dialed in, not unproven ideas.” ### 3. Leverage Historical Data to Forecast Demand and Allocate Budgets You can find valuable clues about where to focus your resources from previous years’ Cyber 5 sales events. While planning your creatives and allocating your budgets, take a close look at last year’s key performance indicators (KPIs), including traffic patterns, sales results, and channel performance. This should give you some ability to forecast which categories are likely to outperform last year. Paul DeMott, chief technology officer at Helium SEO, follows this approach when planning for Cyber 5: “We start by examining past data from similar events to get an idea of the traffic and the key products that will sell. We use this knowledge to develop very specific campaigns that will touch the right users at the right time.” For DeMott, certain KPIs provide additional insight. For example, he looks closely at conversion rates to measure the effectiveness of specific landing pages and offers. He also tracks order values to gauge the effectiveness of efforts to encourage customers to make larger purchases. But, he says, the most important one is return on ad spend (ROAS). According to DeMott, “It comes in handy when tracking the direct relationship between the amount of money used in the ad and the revenue.” Of course, when using past years’ performance data to plan this year’s Cyber 5 campaigns, remember that you will still need to make real-time adjustments as you go. ### 4. Be Ready to Make Real-Time Adjustments Smart marketers plan their Cyber 5 campaigns well in advance, by analyzing last year’s results and stress-testing their ads in the months leading up to Black Friday and Cyber Monday. That said, Cyber 5 is far too dynamic to rely on a set-it-and-forget-it approach. As you’ve likely experienced, the landscape — including consumer behavior, competitive pricing, and even supply chains — can change within hours. A campaign that looks bulletproof on Friday morning could be underperforming by Saturday if your competitor offers a deeper discount or a popular item sells out. As a marketer, you have to be prepared to make adjustments on the fly. That might include swapping out underperforming ad creative, shifting your budget to a higher-performing channel, or changing your messaging to keep up with shifts in consumer sentiment. Himanshu Agarwal, CEO of recruiting firm Zenius, recommends that businesses develop a real-time agility plan: “Monitoring daily performance has helped me switch out underperforming creatives early. I replaced these with backup offers and/or creative assets prepared in advance.” Fortunately, there are several AI-powered marketing tools at your disposal that can make real-time adjustments much easier. Platforms like Realize can monitor multiple KPIs, such as click-through rates (CTRs) and ROAS in real time, and quickly alert marketers to underperformance. They can then recommend next steps, like adjusting audience targeting or increasing bids on top-performing keywords. ### 5. Create Dedicated Cyber 5 Landing Pages and Exclusive Offers By creating a dedicated Cyber 5 landing page, you can consolidate all of your offers. This not only makes it easier for shoppers to find your promotions and deals, it also makes for more accurate performance tracking. Lou Haverty, owner of e-commerce brand Tank Retailer, takes this approach every year: “We negotiate exclusive manufacturer discounts one to two months in advance and direct all traffic to a Cyber Monday landing page, supported by targeted Google display ads and email campaigns.” Ultimately, you can use a well-designed landing page, along with exclusive offers, as the centerpiece of your Cyber 5 campaign. ### 6. Use the BFCM Continuum to Maintain Momentum Across All Five Days Many shoppers view Cyber 5 as one extended shopping event, rather than individual days. Marketers should do the same by building campaigns that cover the full period from Thanksgiving through Cyber Monday. Along the way, messaging can shift to match buyer sentiment at every stage. For example, in the days leading up to Black Friday, you can leverage “early access” messaging to capture shoppers who like to plan ahead. Over the weekend, you can promote urgency and exclusive offers to drive sales. Finally, on Cyber Monday, you can frame your offers as a “second chance” for deals shoppers may have missed earlier. ## Key Takeaways While marketers must be prepared to make last-minute adjustments during Cyber 5, successful campaigns are often the result of advanced planning, including careful analysis of previous campaigns and rigorous testing of creatives in the months leading up to Cyber 5. Creative assets need to be validated before November, with ad variations and backups ready to go in case consumer demand shifts, supply chains break down, or competitors change their pricing strategies at the last minute. Pay close attention to your headline strategy, too: Don’t settle for generic headlines or those that focus solely on price. Direct discount headlines are acceptable, but incorporating elements of curiosity, urgency, and problem-solving is critical if you want to persuade high-intent consumers. Finally, remember to treat Cyber 5 as one continuous shopping event, rather than a series of disconnected sales days. You can choose a single narrative and allow it to evolve through the different stages of the period. ## Frequently Asked Questions (FAQs) ### What are the top trends in 2025 Cyber 5 advertising? For 2025, the biggest trends in Cyber 5 advertising include using hybrid headline strategies that combine persuasive elements, the ability to adjust ad creatives in real time, and producing and testing creatives early, long before November. ### How can I craft abandoned cart email sequences for Cyber 5? You can use a three-step sequence. Start with an immediate reminder letting the shopper know that they’ve left the item in the cart. Then, follow up within 24 hours, using urgency messaging: “Low stock” or “sale ending soon” keywords are good examples. Finally, make one last push before the sale ends. You may even want to include an extra incentive, such as free shipping. ### What are the key KPIs to track during Cyber 5 to measure the success of my advertising campaigns? Conversion rate, cart abandonment rate, average order value (AOV), return on ad spend (ROAS), and incremental revenue per visitor are all important KPIs to track during Cyber 5. Make sure you’re monitoring them in real time so that you can make adjustments as needed. This can include optimizing your creative or reallocating your budget to the highest-performing areas. --- ### One Through 10: What Your Google Quality Score Says About Your Advertising URL: https://www.taboola.com/marketing-hub/quality-score/ Last Modified: 2025-08-21 11:02:21 So your ads are hitting a Quality Score of 10? Great! That means better placement in search engine results pages (SERPs), lower costs paid per click of your online ads, and, almost assuredly, a better conversion rate with your customers. Batting a bit closer to three or two or even… one? A low Google Ads Quality Score can pretty much guarantee your ads are hardly being served to a relevant audience at all — and when they are, you’re probably paying more than you want for clicks that may not even be driving potential customers further down the funnel. Your Google Ads Quality Score is a measure of the quality of your ads, to be sure, but it’s also a self-fulfilling prophecy: With ads that rank well in this metric usually comes more success; with ads that rank poorly come negative outcomes. ## Understanding Quality Score in Paid Search In Google Ads, and similar platforms like Microsoft Advertising, a Quality Score is a metric that reflects the overall quality of a marketer’s ads and the landing pages to which those ads point. The Quality Score is used to determine ad ranking and cost-per-click (CPC). This indicator is a diagnostic tool, typically on a scale of one to 10, that indicates how relevant and useful your ads are to users searching for your keywords. It also tells the search engine how well to rank you and how often to serve your ads. Quality Score is important for paid search advertisers because it directly impacts ad costs and visibility on search engine results pages. ### Where Can Advertisers See Their Quality Scores? Marketers can see their Quality Scores within their Google Ads and Search Ads 360 accounts by customizing the keyword columns. By adding a "Quality Score" column, marketers can view the overall score and its components: Expected CTR (click-through rate), Landing Page Experience, and Ad Relevance. ### Is Quality Score a Real-Time Metric? Quality Score is generally considered a real-time metric, as it’s updated each time your ad is triggered in an auction. That means your score reflects the current relevance of your ad to a user's search query. ## Factors Influencing Quality Score Quality Score is not a one-size-fits-all proposition; there is a lot that goes into the “numbers” the algorithms crunch, and historical performance is a significant factor contributing to your Quality Score. Specifically, the individual keyword and a matched ad, as well as the overall historical CTR of the entire Google Ads account, are factored in when calculating the expected CTR, a key determinant of Quality Score. ### What Is the Expected Click-Through Rate and How Does It Affect Quality Score? In Google Ads, the expected CTR is a prediction of how likely a user is to click on an ad when it appears for a specific keyword. It's a crucial component of Quality Score, which significantly impacts ad rank and CPC. A higher expected CTR generally means a higher Quality Score, which can lead to lower costs and better ad positioning. ### How Does Ad Relevance to Keywords Impact Quality Score? Ad relevance is a key factor in determining an advertisement’s Quality Score. A higher ad relevance score, which indicates how closely the ad copy aligns with the searcher's intent, leads to a higher Quality Score. This, in turn, can result in lower cost-per-click and better ad positioning. ### How Does Landing Page Experience Influence Quality Score? Landing page experience is a critical factor heavily influencing your Google Ads Quality Score. A good landing page experience — characterized by relevance, usefulness, and user-friendliness — can lead to a lower CPC and improved ad rank. Conversely, a poor landing page experience can negatively impact your Quality Score, resulting in higher CPCs and lower ad positions. ## The Benefits of a High Quality Score The better the score, the better the SERPs placement, the lower the costs, and the more the conversions. What’s not to like? ### Does a High Quality Score Impact Eligibility for Ad Extensions and Other Features? A higher Google Ads Quality Score can dramatically impact the eligibility of your ads for display with ad extensions and other ad formats. Ad extensions, like sitelinks or callouts, can in turn enhance the visibility and effectiveness of your ads by providing additional information and functionality. Google's algorithm determines whether or not these extensions are shown based on several factors, including ad placement, other ads on the page, past extension performance, and your ad's rank. ### How Does It Contribute to Better Overall Ad Performance and ROI? A high Quality Score in Google Ads leads to better ad performance and a higher return on investment (ROI) through lower costs, improved ad placement, and increased visibility. A higher Quality Score indicates that your ads, keywords, and landing pages are well-aligned and relevant and useful to users, which Google rewards with lower CPC and better ad positions. This leads to more clicks, conversions, and ultimately, a better ROI for your ad spend. ## Diagnosing and Improving Quality Score A low Quality Score will negatively affect how your ads perform, but it’s not a life sentence. In most cases, with effort, you can improve your Quality Score within a matter of weeks. ### How Can You Identify Which Keywords Have Low Quality Scores? To identify keywords with low Quality Scores in Google Ads, you can utilize the Quality Score filter in your account and examine the individual keyword Quality Scores and their components. A score of seven or above is considered good, while a lower score, typically below five, indicates lower relevance and potential for improvement. ### What Steps Can You Take to Improve Your Expected CTR? To improve your expected CTR, focus on making your ads more relevant and appealing to your target audience. This often involves optimizing ad copy, using more relevant keywords, refining your targeting, and leveraging ad extensions. Additionally, consider A/B testing different ad variations and landing page variations and monitoring your campaign performance to make necessary adjustments. ### What Are Some Strategies for Improving Landing Page Experience? To enhance a user’s experience on your landing pages, focus on clarity of design and copy, page load speeds, and engagement opportunities. Simplify a page’s design to ensure faster loading times, and make the call to action (CTA) clear and prominent. Align the landing page content with the ad that led the visitor there, and optimize it for viewing and interaction on mobile devices. Also, consider using social proof like testimonials and incorporating engaging elements like videos or interactive features. ## Common Misconceptions About Quality Score You need to know what’s accurate and what’s not when it comes to Google Ads Quality Scores so you can approach them properly. ### Is Quality Score a Direct Ranking Factor? Quality Score in Google Ads is a ranking factor for paid search results, but it’s not a direct ranking factor for organic search results. Quality Score is a metric used to assess the quality of your ads and landing pages in the context of paid advertising, specifically in Google Ads. While it influences ad rank and cost per click in paid search, the score does not directly impact how your website ranks in organic search results. ### Can You Directly Buy a Higher Quality Score? You can’t directly buy a higher Quality Score with Google Ads. While your bids and budget can impact the visibility of your ads (which might indirectly influence the data used to calculate your Quality Score), they are not direct factors in determining the score itself. Think of your Quality Score as a reflection of how good and relevant Google deems your ad experience to be for users, rather than a price tag. ### Does Pausing and Restarting Campaigns Affect Quality Score? Pausing and restarting a Google Ads campaign does not directly impact Quality Score, but it can indirectly affect performance. While your Quality Score and historical data are retained, performance may be delayed when a campaign is resumed due to a learning period required for the algorithms to adjust. ### Is Quality Score the Same Across All Match Types for a Keyword? Quality Score is not different for a keyword based on its match type. A keyword's Quality Score is the same regardless of whether it's used in a broad phrase, or exact match. It's calculated based on the performance of search queries that are an exact match to that keyword. ## Key Takeaways A Google Ads Quality Score is a metric that estimates the relevance and quality of your ads, keywords, and landing pages. A higher Quality Score generally leads to lower costs (in the form of lower CPC) and better ad positioning in search engine results, making it crucial for campaign success. While a good Quality Score is beneficial, it's not a key performance indicator (KPI) to optimize directly; instead, focus on improving the underlying factors that influence it. These include expected CTR, ad relevance (how closely your ad copy matches the user's search query, for example) and user landing page experience. ## Frequently Asked Questions (FAQs) #### What is a "good" Quality Score? A good quality score is generally considered to be seven or above, with a 10 being the best possible ranking, while a poor score is seen as below five. With improved ad relevance and landing pages, you can usually start to raise Quality Scores within a month or so. #### How often is Quality Score updated? An ad’s Quality Score is updated by Google Ads every time an ad enters an auction, which can be multiple times a day for high-volume keywords. While the underlying score is constantly recalculated, the displayed quality score in the Google Ads interface might not update immediately and can lag behind. #### What tools can help me analyze and improve my Quality Score? To analyze and improve your Google Ads Quality Score, you can always use tools within Google Ads itself, such as the Keyword Planner and the Search Terms report, but also consider leveraging third-party platforms like Semrush, WordStream, and Ahrefs. #### How important is mobile landing page experience for Quality Score? A good mobile landing page experience directly impacts the Quality Score, which in turn affects ad ranking and cost-per-click. Google assesses the relevance, transparency, and usability of your landing page, and a mobile-friendly, fast-loading, and easy-to-navigate page contributes to a higher Quality Score. --- ### What Is a Session in Digital Marketing? URL: https://www.taboola.com/marketing-hub/session/ Last Modified: 2026-06-22 08:37:24 You might not think of your time browsing a website or scrolling a social media platform as a “session,” but many advertisers certainly do. In the world of online marketing, monitoring user session behavior is vital to understanding how a campaign is performing. ## Defining and Understanding Sessions The moment a user opens a site or app — whether by clicking or tapping a link or by accessing the platform directly — the session has begun. ### How Do Different Analytics Platforms Like Google Analytics Define a Session? In Google Analytics, a session is defined as a group of user interactions with a website that occur within a specific and continuous given time frame. The default time frame, or session timeout, is 30 minutes of inactivity. If a user is inactive for more than 30 minutes, any subsequent activity is considered a new session. Sessions also end by a user simply navigating away from a site or out of an app, or by the closing of a browser or powering down of a device. ### What Are the Default Session Timeout Settings, and Can They Be Customized? As mentioned, 30 minutes is the default time after which a session is considered over on most platforms, but this timing can be customized in most cases. If the timing is not set, after 30 minutes of inactivity, no more web analytics — such as user activity including clicks and taps, videos watched, likes and comments, page views, etc. — will be collected in that session. ## Key Metrics Related to Sessions ### What Is the "Number of Sessions" Metric and What Does It Tell You? In Google Analytics, the "Number of Sessions" metric represents the total number of individual user interactions with a website or app within a specific time frame. It indicates how frequently users engage with the content, providing insights into user engagement and the effectiveness of the site in retaining visitors. ### What Is "Users" Versus "Sessions" and Why Is This Distinction Important? Users represent unique individuals who have visited a website or app, while sessions represent the number of visits, or interactions, a user makes on a website. A user can have multiple sessions, especially if they return to the site repeatedly or remain active for extended periods. The distinction is important because it provides different perspectives on user behavior. Analyzing both metrics helps in understanding user engagement and website effectiveness. ### What Is "Sessions Per User" and What Does It Indicate About User Engagement? "Sessions per user" is an important metric measuring the average number of sessions (sometimes also called “visits”) a user initiates on a site or in an app within a given timeframe. It's a great indicator of user engagement that shows how frequently a user returns and interacts with the platform. A higher sessions per user ratio generally suggests greater engagement with lots of return visits, while a lower ratio may indicate a higher bounce rate and a need for improvement in user experience (UX) or content relevance. ### How Does "Session Duration" Provide Insights Into User Behavior? Session duration information provides insights into user engagement and content efficacy by showing how long users interact with an app or website during a single visit. Longer session durations generally suggest higher user engagement, while shorter sessions can indicate a lack of interest, difficulty in finding information, or poor site function — though a short session can also indicate an efficient user experience that was actually quite a good one! That’s why analyzing session duration alongside other metrics like pageviews and user behavior patterns offers a more comprehensive understanding of user engagement. ### What Is "Pages Per Session" — or "Session Depth" — and What Does It Measure? "Pages per session," also known as "session depth," measures the average number of pages a user views during a single visit to a website or application. It's a key engagement metric, indicating how deeply users explore a site or app during a session. ## The Importance of Sessions in Digital Advertising If someone comes back to your restaurant time and time again and they have a leisurely two-hour dinner every time, they’re probably enjoying the atmosphere and the menu, right? If they come in once, look around the dining room, and bolt, that was not a good dining experience. With online marketing, studying sessions provides the same sort of customer data. ### How Can You Analyze User Behavior Within a Session After They Click on an Ad? Understanding what users do after clicking on your ads is crucial for optimizing your marketing efforts and improving their overall experience. You need to see how long they spent on the page to which they were directed, what actions they took, and if a conversion took place. The more steps the person took, the better the ad worked. ### How Can You Segment Sessions Based on Traffic Sources Like Paid Search or Social Media? Segmenting your website sessions by traffic source (like paid search, social media, and so on) in analytics tools provides valuable insights into user behavior and campaign performance. You can utilize “Urchin Tracking Modules” (UTMs), which are small code snippets added to the end of a URL, to track the performance of online marketing campaigns. In Google Analytics, sessions play a crucial role in understanding and attributing conversions by providing context to user behavior on your website or app. ## Analyzing Session Data for Optimization ### How Can You Use Session Duration and Pages Per Session to Assess Engagement? Session duration and pages per session are important metrics for assessing user engagement on a website. Longer session durations and higher pages per session indicate that users are finding the content valuable and are actively exploring the site. Conversely, short session durations and low pages per session may suggest a lack of engagement, difficulty in finding information, or poor content quality — or again, it may indicate a very well-designed site that allows for immediate action. That said, very high bounce rates — when a user enters and then exits a page without taking any meaningful action — definitely indicate a content and/or ad relevance problem. ### How Can You Compare Session Metrics Across Different Traffic Sources or Campaigns? To compare session metrics across different traffic sources or different campaigns, you can use tools like Google Analytics to analyze metrics such as bounce rate, average session duration, conversion rate, and pages per session. By examining these metrics for each traffic source (organic search, social media, paid ads, and so on), you can identify which channels are most effective at driving engagement and conversions. ### What Are Some Advanced Techniques for Analyzing Session Data? Session data is often a veritable treasure trove of information about user behavior, and it can be analyzed using a variety of advanced techniques to extract deeper insights and drive better decisions. Beyond basic metrics like page views and bounce rates, these advanced approaches — such as session replay, advanced experimentation, and behavioral segmentation — allow for a more nuanced understanding of user journeys and interactions. ## Limitations of Session-Based Analytics ### What Are Some of the Limitations of Relying Solely on Session Data? While session data offers valuable insights into immediate user behavior, relying solely on it presents numerous limitations for marketers seeking a comprehensive understanding of their audience and how their campaigns are working. The main issue is that session data provides a “snapshot” of user interactions within a specific timeframe, but struggles to connect multiple sessions to the same user, making it difficult to understand the full customer journey or identify returning customers. Also, ad blockers and browser restrictions can interfere with session data collection, preventing the loading of tracking scripts and cookies. ### How Can Cross-Device Tracking Impact the Accuracy of Session Metrics? Cross-device tracking can significantly improve the accuracy of session metrics by providing a more holistic and less fragmented view of the user journey. Without it, traditional analytics approaches (that often rely on cookies) struggle to connect user activity across different devices, risking skewed and inaccurate session metrics. ### How Do Single-Page Applications (SPAs) Affect Session Tracking? SPAs can revolutionize user experience with dynamic content updates and by eliminating the need for full-page reloads. While this provides a smoother and more responsive feel, like a native application, it does also significantly impact how session tracking is managed and usually requires a different approach compared to traditional multi-page applications (MPAs). ### Are There Alternative or Complementary Approaches to Analyzing User Behavior? There are several alternative and complementary approaches to analyzing user behavior in online marketing, beyond traditional metrics like page views, session analysis, and the like: - Quantitative research methods use numerical data to identify patterns and trends in customer behavior. - Surveys and questionnaires gather data from a large number of users to understand their preferences, purchasing habits, and satisfaction levels. - Behavioral data analysis synthesizes data from user interactions with websites and apps, such as website visit patterns, purchase histories, and usage data. - Heatmaps visualize user interactions on a website or app by displaying areas where users click, scroll, or move their mouse: This helps identify popular elements or areas for improvement. ## Key Takeaways In digital marketing, a “session” refers to all of a user’s interactions with a website (or app) within a specific timeframe. The metric aims to capture a user's visit, encompassing all the actions they take, such as page views, likes, comments, or transactions, from the moment they enter until they leave or become inactive for a defined period. Understanding sessions is crucial for digital marketers as it provides insights into user behavior, website engagement, and the effectiveness of marketing campaigns. ## Frequently Asked Questions (FAQs) ### What happens when a user is inactive on a website? It varies from site to site, of course, but when a user is inactive on a website, several things can happen, depending on the website's policies and how it's designed. Generally, the website might log the user out automatically. In marketing terms, after 30 minutes, a session is considered over if there is no action. ### Do multiple tabs opened by the same user count as one session? Multiple tabs opened by the same user within the same browser usually count as a single session. When a user opens a website in multiple tabs, they are typically using the same browser instance and the same session data (like cookies) is shared across all those tabs, per multiple online resources, such as Adobe Analytics. This means that any actions performed in one tab will be reflected in other tabs, and logging out of one tab will log the user out of all tabs. ### How are sessions tracked across different visits from the same user? Sessions are tracked across different visits from the same user primarily through the use of cookies and user IDs. When a user visits a website, their browser typically receives a unique cookie, which acts as a persistent identifier. If the user returns, the website can recognize the cookie and associate the new visit with the existing session. User IDs, when implemented, provide a more robust way to track users across devices and browsers by associating a persistent identifier with the user's account. ### Can a single user have multiple sessions on the same day? Yes, a single user can have multiple sessions on the same day. The ability to have multiple sessions simultaneously depends on the specific application or system being used: For example, some web applications allow users to log in from multiple browsers or devices simultaneously, creating separate sessions. Other systems, like Remote Desktop Services, may have settings that restrict users to a single session at a time, but these settings can be configured to allow multiple sessions. ### How can I troubleshoot discrepancies in session data? To troubleshoot discrepancies in session data, start by identifying the source of the issue, comparing data across different sources, and checking for errors. Investigate potential causes like different tracking methods, session timeout settings, or data filtering, and utilize tools for data validation and cleaning to ensure data accuracy and consistency. --- ### Stop the Scroll: How to Design High-Performing Holiday Ads URL: https://www.taboola.com/marketing-hub/design-high-performing-holiday-ads/ Last Modified: 2025-07-31 08:46:06 Holiday ads compete for attention in a busy market during a relatively short period of time. Crowded user feeds mean that marketing and advertising teams have to design ads that stand out and entice readers to buy quickly. That said, the holiday season also brings specific opportunities for ad design, as teams look to creative work to increase conversions and purchases. A successful December can end the sales year on a high note, meeting goals and setting the business up for success in the new year. Check out the tips below on how to design high-performing holiday ads, and the Taboola creative team’s playbook for success. ## Enhance Your Holiday Ads Performance With These 7 Design Tips ### 1. Create Visually Striking Imagery and Video Good holiday design starts with high-quality visuals, festive themes, and a clear product focus. Consider how your design can stand out across images, photos, and videos, paying attention to every detail. “Be unexpected: Remember that your content will be surrounded by a plethora of similar holiday themes,” advises designer Liz Parmalee. “Instead of using the common hunter green and cherry red combination, consider alternatives like lime green and magenta, or orange and pink. Maybe the traditional ‘Silent Night’ navy turns into a deep purple.” Consider every detail as you’re designing ads to grab attention during the holiday seasons. “Let your typography stand out — while others may be using script fonts, you could choose something with more personality,” suggests Parmalee. “Rethinking scale and type hierarchy can also help capture the user's attention without screaming at them with colors or graphics.” ### 2. Tell an Authentic Story Try including authenticity and evoking emotion, depending on your product, market, and advertising channels. For the holiday season, you might use ads to tell stories about family, friends, and connection. Taboola’s Holiday Creative Playbook recommends authentic, kinetic creative, like this: Also consider subtle festive touches or clever, simple wordplay in your ads. You can break out the overall holiday season into sub-seasons, too, like Black Friday, Cyber Monday, Christmas, and New Year’s. Match creative to each of these blocks of time as you also maintain brand consistency and guidelines for good copy, visuals, and calls to action (CTAs). ### 3. Write Compelling Headlines, Copy, and CTAs You’ve only got a second to catch a shopper’s attention, whether it’s in an email, on social media, or in a paid ad. Write benefit-driven headlines and CTAs, and include emotional triggers and a clear value proposition. CTAs should be prominent and action-oriented, and you might test different phrasing ahead of time to see what works best. “Keep it short and punchy,” recommends digital marketing specialist Honey Chhatani. “Make the value proposition immediately obvious — what the offer is, why it matters, and how to act. This is critical when attention spans are short.” In addition, incorporate seasonal elements like urgency cues, gift-giving language, countdown timers, or last shipping date reminders to drive action, says Chhatani. “Add urgency with phrases like ‘Only a few left’ or ‘Offer ends tonight.’ Real or perceived scarcity can dramatically boost CTR and sales.” ### 4. Get the Details Right in Advance The holiday season starts earlier and earlier, with research and initial browsing often taking place in October or even before. Line up your creative with the tactics you plan to use for a robust strategy. “Sync paid and organic creative,” suggests Chhatani. “Make sure your social, email, landing pages, and ads are aligned in theme, tone, and visuals. A consistent story across touchpoints increases brand recall and trust.” While your message should stay consistent, you can still tailor visuals and tone to fit the platform, so Instagram vs. email vs. search may have subtle variations. In addition, start teasing campaigns before they go live, Chhatani advises. “Just like Amazon promotes Prime Day in advance, you can create excitement and build purchase intent with early announcements, preview emails, or waitlists.” Use predictive analytics and real-time personalization to serve the most relevant ad variation and placement to each user segment. The right keywords for SEO are also essential. “Filter out irrelevant or underperforming search terms to prevent wasted spend in paid search campaigns,” says Chhatani. “Focus on keywords that are tightly aligned with your offer and customer intent. Think relevance, not just volume.” ### 5. Emphasize Brand Identity The holiday season isn’t the time to rebrand or otherwise dramatically change your company’s brand identity. Instead, it’s an opportunity to reinforce your existing brand recognition and promise. For smaller companies, their holiday creative can do this alongside driving sales. “Focus on creating something simple, distinct, and scalable,” says Parmalee. “The goal is for people to see your logo or brand color across an event hall or in their web browsers and immediately recognize you.” Smaller businesses or startups in particular can elevate what’s already working without disrupting brand recognition, adds Parmalee: “Try a slow, deliberate enhancement of the elements of your brand that are working well and are already recognizable.” Consistency is also important for timing and messaging during the holiday season. “Ensure your ad launches, emails, and social posts align for product drops or big announcements,” says Chhatani. ### 6. Optimize for Mobile First Holiday shoppers are browsing, researching, and buying on their phones like never before: Mobile devices are expected to make up about 60-63% of online traffic during the 2025 holiday season, with desktop computers making up the rest. “Most holiday traffic is mobile,” agrees Chhatani. “Ensure creatives are designed for small screens — short copy, large and clear CTAs, fast-loading visuals, and vertical video formats where relevant.” Concise messaging and thumb-stopping visuals can also catch the attention of users quickly. The entire mobile shopping experience should be very smooth and easy: “Prioritize fast load times and seamless checkout UX,” advises Chhatani. Check out the holiday creative playbook for more on mobile optimization. ### 7. Test Different Ad Formats and Types There’s a whole world of ad formats to explore in modern digital marketing, such as carousel ads, motion ads, video ads, and traditional static images. They all have their place in a holiday marketing mix and each can contribute to different goals. Draw on past performance data when you’re building a holiday campaign. “Look at what performed well in past campaigns — what messages, visuals, and formats led to higher engagement or conversions?” asks Chhatani. “Document your findings and build on them.” There are lots of ways to experiment with engaging users, too. “Motion is a helpful tool for drawing the eye to your ad,” says Parmalee. “It doesn't need to be an elaborate animation — having an element swish into view or pulse can be enough to catch the user's eye.” Performing testing on any new formats is important, especially to understand the line between engagement and overload. “While larger and more complex animations can certainly attract attention, they may also end up being distracting or annoying,” warns Parmalee. “Start with something simple, and conduct an A/B test comparing your animated ad with a static version to measure any differences in engagement.” ## Key Takeaways Creative ad best practices during the holiday generally rely on the same principles as good ad design during the rest of the year. “Marketing design serves a specific purpose: to support the carefully crafted initiatives of the marketing team,” says designer Liz Parmalee. “Consider the story you want to tell, the emotions you want to evoke, and the actions you want the user to take. Everything else is icing on top.” The holiday season can outperform the entire rest of the year when marketers and advertisers create and promote high-quality, user-focused ads. Designing visually compelling work with short, snappy copy and clear CTAs is essential for holiday success. Teams should also pay attention to consistent brand identity, timing, and ad formats for the best possible holiday ad performance. --- ### Insights Backed by Realize Proprietary Data for Your 2025 Cyber 5 Ad Copy URL: https://www.taboola.com/marketing-hub/realize-data-cyber-5-ad-copy/ Last Modified: 2025-09-11 09:55:28 For years, Black Friday and Cyber Monday have dominated the holiday shopping season. It’s critical for digital advertisers to be strategic when creating ad copy and headlines to capture attention and drive engagement during this time, so to help you prepare in 2025, we’ve analyzed previous Black Friday and Cyber Monday ad headline performance. The findings from our proprietary Realize data sheds light on various headline performance trends, both top performers and underachievers. Here, I’ll provide you with insight-based tips you can implement as you build your 2025 Cyber 5 ad copy. ## Qualitative Headline Categorization We carefully reviewed all of the headlines in our analysis and assigned them to specific categories, based on their central persuasive technique or messaging strategy. Classifying all of the headlines into specific groups allowed us to understand which types of headlines resonated with consumers, and why certain types performed better than others. Here’s a closer look at the different headline categories used: ### Direct Discount/Price-Focused (DD) Direct discount or price-focused headlines highlight financial savings and discounts, and are one of the most widely used strategies during Black Friday and Cyber Monday. Examples, such as “Up to 70% off!” or “Save $320,” focus on the price advantage of a product and are designed to appeal to value-conscious consumers. There is power in a clear, measurable discount, especially during major shopping events like Black Friday, when people are faced with a bewildering array of offers. ### Urgency/Scarcity (US) Urgency and scarcity-driven headlines (think, “Selling out fast!”) trigger an immediate emotional response from consumers by homing in on limited-time offers or product availability. They can tap into customers’ fear of missing out, or FOMO, which can cause them to act quickly, even impulsively, to secure an excellent deal. This approach is consistently effective across Black Friday and Cyber Monday campaigns. ### Product/Brand-Specific (PB) Product or brand-specific headlines target a more intent-driven audience by directly naming a particular product. Examples could include “Dyson Black Friday Deals” or “Save on the iPhone 16.” Shoppers who respond to these types of headlines know precisely what they’re looking for and are searching for the best deal on those particular items. ### Benefit-Oriented (BO) Unlike direct-discount headlines, benefit-oriented headlines shift the focus away from just price and toward the positive impact the product can have on the consumer. Headline examples might include “Save on Energy Costs This Winter,” or “Get Salon-Quality Hair with the Dyson Airwrap.” This approach extends beyond price and appeals to consumers’ quality of life. ### Problem/Solution (PS) Problem/solution headlines are highly effective as they identify a customer pain point before presenting the product as the solution. One example is, "Hate waiting in long lines? Skip the chaos with our exclusive Black Friday online deals!” This highlights the problem of long lineups in shopping malls during Cyber 5 and presents the solution of securing great deals online from the comfort of your home. This type of headline often performs well as it appeals to customers’ immediate needs. ### Curiosity/Intrigue (CI) The goal of curiosity-driven headlines is to create surprise or pique a customer’s interest, making them want to click on the ad to learn more. Examples like “Born Before 19XX?” or “Cyber Monday is almost here — but this exclusive offer might disappear first!” create a sense of mystery and intrigue. The goal with these types of headlines is to prompt customers to click on the ad to discover the rest of the story. According to Esther Altomare, head of industry enterprise advertising at Taboola, Curiosity/Intrigue and Problem/Solution headline styles both “drive engagement by tapping into emotional triggers and offering quick, relatable fixes. The key is balancing intrigue with clarity, making the benefit obvious while evoking just enough curiosity to prompt a click.” ### Generic Sale/Event Announcement (GS) Some headlines are more generic and simply announce a Black Friday or Cyber Monday sale without providing further details. While the purpose of these headlines is to reach a broader audience, they can be ineffective in generating clicks. ### Informational/Guidance/News (IN) Informational headlines tend to be less promotional and mainly focus on providing valuable information to consumers. Examples include “Tips for Avoiding Black Friday Scams,” or “How to Get the Best Cyber Monday Deals.” While these ads are not directly tied to a sales transaction, they still play a crucial role in influencing customer decisions during the Cyber 5 period. ### Call to Action (CTA) Headlines that feature direct calls to action are designed to get an immediate response from consumers. Phrases like “Click Here!” or “Shop Now!” are intended to drive customers to take action. CTA headlines can be effective, but they often lack a strong value proposition, which is crucial for ensuring customer follow-through. ### Social Proof/Expert/Celebrity Endorsement (SP) Using celebrity or expert endorsements in headlines helps brands establish trust and adds instant credibility to a promotion. Headlines like “Oprah’s Favorite Things on Sale,” or “Over 50 Rangefinders Tested and Approved by Experts,” are good examples of this headline style. ## Which Headlines Are Cyber 5 Shoppers Most Engaged With? We analyzed headline performance from 2024 to find out which of the headline categories listed above consistently captured customer attention during the critical Cyber 5 shopping period (the five days from Thanksgiving Day through to Cyber Monday). The data below offers a clear picture of what resonates with shoppers during these high-volume shopping events. Note: The CTR percentages in this table are illustrative, reflecting the relative performance trends observed across the dataset's diverse headlines and their categorized examples. Average Click-Through Rate (CTR) by Headline Category (2024-2025 Illustrative Data) for Black Friday Headline Category Average CTR (%) Curiosity/Intrigue (CI) 10.00% Problem/Solution (PS) 9.00% Call to Action (CTA) 4.00% Benefit-Oriented (BO) 3.00% Direct Discount (DD) 2.00% Urgency/Scarcity (US) 1.80% Product/Brand-Specific (PB) 1.50% Informational/Guidance/News (IN) 1.50% Generic Sale/Event Announcement (GS) 0.25% Average Click-Through Rate (CTR) by Headline Category (2024-2025 Illustrative Data) for Cyber Monday Headline Category Average CTR (%) Problem/Solution/Benefit-Oriented 1.85% Explicit Discount/Savings 1.72% Urgency/Scarcity 1.68% Product/Category Specificity 1.55% Social Proof/Expert/Celebrity Endorsement 1.48% ## Top Performers and Implications The strength of “Curiosity/Intrigue (CI)” and “Problem/Solution” headlines over the Cyber 5 period demonstrates the importance of connecting a product to a direct customer need or desire. These headlines address more than just product features — they highlight the value proposition in a way consumers can relate to. “Urgency/Scarcity (US)” headlines also performed well over the Cyber 5 period, but their widespread use in Black Friday advertising may lead to diminishing returns. Try headlines that provoke a strong emotional response and reinforce the primary customer intent: Find a great deal and act before it’s gone! “Direct Discount (DD)” headlines are seen to successfully capture a Cyber Monday shopping crowd who are approaching the day with clear shopping intent, and specific product/service purchases in mind. With this in mind, marketers in 2025 should perform an in-depth analysis of their products, identifying high-demand products and creating specific, keyword-rich headlines for them. The inclusion of Social Proof/Expert/Celebrity Endorsement in the top Cyber Monday performers, meanwhile, highlights the ongoing power of external validation in influencing consumer decisions, leveraging trust and aspirational marketing to differentiate ad copy. ## Under-Performers and Implications “Generic Sale/Event Announcement (GS)” headlines were significant underperformers, with a CTR of only 0.25% in our analysis. While these broad headlines, such as “Early Black Friday Deals,” serve a purpose, they don’t generate clicks as effectively as more targeted ones. They may get high numbers of impressions, but they aren’t translating to high CTRs. If you’re looking for maximum exposure, you may achieve the reach you’re looking for with generic headlines, but without a compelling hook, your audience will remain unengaged. This only underscores the need for more targeted and persuasive headline messaging. ## Consistencies Certain headline categories remain consistent performers (or underperformers) throughout the Cyber 5 sales events. “Curiosity/Intrigue (CI)” and “Problem/Solution (PS)” headlines drive the highest CTRs, while generic headlines could have more impressions, but less clicks (therefore, a lower CTR). “Direct Discount (DD)” and “Urgency/Scarcity (US)” headlines consistently perform well on both Black Friday and Cyber Monday, proving that urgency-based language is crucial for driving customers to take immediate action. Because of this, consider using time-sensitive and scarcity-driven language, including phrases such as “last chance,” “final hours,” “won’t last long,” and “expiring soon.” These keywords create FOMO, which causes consumers to convert quickly, provided you don’t overdo it. ## 6 Cyber 5 Ad Copy Trends from Realize Data ### 1. Emergence of Narrative Since 2023, Curiosity/Intrigue (CI) and Problem/Solution (PS) headlines have become more prominent and effective, suggesting a change in consumer behavior. Shoppers want more than just rock-bottom prices: More than ever, they are being drawn to headlines that tell compelling stories and directly address their needs. This is critical for marketers, as the Cyber 5 market is becoming more competitive every year. Yes, direct discount headlines are still effective, but they’re so common that they don’t stand out like they used to. “Discount headlines still work, but brands need to layer in storytelling, assurance, and personalization to win attention,” Altomare reiterates. This focus on narrative-driven or curiosity-led headlines appeals to consumers’ deeper psychological motivators, moving beyond price as the sole selling feature. ### 2. Celebrity/Influencer Integration More advertisers are using celebrity names (or celebrity claims) in their headlines. For example, you’ll often see references to stars like Oprah, Taylor Swift, or Martha Stewart. “Influencer headlines work best in fashion, wellness, tech, and lifestyle, especially when they highlight relatability or social proof,” says Altomare. “Effective headlines name-drop strategically and tie the endorsement to clear benefits or popularity metrics.” They can also differentiate your ad in a crowded market. ### 3. Problem-Solution Framing Headlines that clearly define a problem and identify the product as a solution are generating higher CTRs. For example, a headline like “This tiny dongle fixes the biggest problem with Kindles,” or “This video doorbell cured my insomnia,” builds empathy into the ad and addresses shoppers’ pain points, making it more likely to resonate with them. ### 4. Niche Audience Targeting With increasingly sophisticated AI tools at their disposal, along with real-time data, brands can create headlines for highly specific groups — for example, “Cyber Monday Steals for Seniors.” It’s a hyper-personalized approach that increases ad relevance because it speaks directly to particular customer groups. ### 5. Product Category Evolution While electronic items such as laptops, smartphones, big screen TVs, and gaming consoles remain a Cyber 5 pillar, deals are expanding into other product categories as well. This includes lifestyle products like fashion, beauty, and home decor, and home improvement items, such as robot vacuums, smart lighting, and curtains. The deep discounts also apply to services like flights, streaming, mortgages, and office software. It’s a strong indication that in 2025, Cyber 5 will appeal to a much broader range of shoppers. ### 6. Cross-Promotional with Black Friday (BFCM Continuum) Many Cyber 5 headlines now directly link Cyber Monday to Black Friday. For example, phrases like “Black Friday Cyber Monday Sale,” or “Still on sale after Cyber Monday,” frame Cyber Monday as a “second chance” to take advantage of deals that may have been missed on Black Friday. This helps brands maintain momentum throughout the Cyber 5 period. It’s also important to note that Black Friday is also stretching in the other direction, with so many “early Black Friday sales” happening that the actual Friday itself is now just a part of the wider event. ## Strategic Implications ### Blending of Persuasive Headline Elements Our analysis indicates that brands can no longer rely solely on a single approach to Black Friday and Cyber Monday headlines. In fact, the best-performing headlines incorporate multiple persuasive elements. To achieve this, consider framing Product-Brand Specific (PB) and Benefit-Oriented (BO) headlines with Curiosity-Intrigue (CI) and Problem-Solution (PS) messaging. Remember, too, that while Direct Discount (DD) messaging is still effective, it works best when you combine it with these other elements. ### Evolving Consumer Behavior Customers have become more sophisticated and selective in how they respond to Cyber 5 advertising. This means that they are less likely to engage with generic headlines, as evidenced by the category’s low CTRs. Consumers want you to tell a story, spark their curiosity, or address a clear pain point. These types of headlines are outperforming those that solely focus on price. ### "Early Black Friday" As is the case with other shopping seasons, early promotions have extended the sales window leading up to Black Friday. Still, it’s not as simple as just slapping “Early” on the beginning of your headline: You need to create a clear value proposition explaining why consumers should buy in advance, rather than waiting until the actual Cyber 5 period. ## Actionable Recommendations ### Adopt Hybrid Headline Architectures Focus on creating hybrid headlines that combine multiple elements into a single message. This could include Product-Specific aspects with a strong Benefit-Oriented message, and then layered with some Urgency/Scarcity or Curiosity/Intrigue. For example, you could transform a generic “Dyson Black Friday Sale” headline into “Dyson Airwrap: Achieve Salon-Quality Hair at Home — Limited Black Friday Stock!” As you can see, the amended headline is product-specific, yet it also effectively highlights the product’s benefits and creates a sense of urgency. ### Embrace Narrative and Problem/Solution Framing Incorporating narrative and problem-solving framing should be a priority, especially when promoting high-value products or unique offers. Headlines that ask a question, reveal an unexpected feature, or address a pain point tend to create curiosity and build a stronger connection with shoppers. For example, a security company could replace a straightforward “Black Friday Security Camera Deals” headline with something like, “Worried About Home Security? This Black Friday Deal Makes Advanced Protection Affordable!” The amended headline shifts the focus away from the product by addressing a customer concern with an affordable solution. ### Prioritize Explicit Value Proposition For years, clear and explicit value propositions have been key to Cyber 5 success, and that hasn’t changed. Headlines that highlight a specific discount, whether it’s a percentage off, a dollar amount, or a “lowest price ever” claim, consistently perform well with budget-conscious consumers. That said, this strategy is most effective when direct discount headlines are paired with product-specific elements for high-intent audiences. One good example: “Save $320 on Microsoft Surface Pro 11 — Early Black Friday Price!” ### Maintain and Refine Urgency Messaging As I’ve highlighted in previous headline examples, you can use specific keywords to create a sense of urgency with shoppers. For example, terms like “last chance,” “final hours,” and “limited stock” all tap into customers’ sense of FOMO, compelling them to make a purchase quickly. You can also create urgency throughout in Cyber 5, with “early access” language and, at the end, with “last chance” messaging. However, don’t overdo the urgency narrative. As Altomare cautions, “Treat urgency like spice: It’s effective in moderation, but too much can cause users to tune out or distrust the message.” ### Integrate Consultative and Value-Added Content One emerging trend is a shift toward providing value before price. You can accomplish this by offering advice or highlighting complementary products in your content. For example, “Buying a monitor on Cyber Monday? You’ll need these 10 accessories, too,” positions you as a helpful authority. It may also increase your average order value (AOV) if people purchase additional products based on your recommendations. ### Harness Social Proof and Aspirational Marketing Social proof remains an effective headline strategy, especially in niches such as lifestyle, beauty, and some tech categories. If you can include celebrity endorsements or expert recommendations in your headlines, you can build instant credibility and drive engagement. ### Strategic Positioning within the BFCM Continuum When positioning headlines, be mindful of the entire Cyber 5 period. For example, frame Cyber Monday as the last chance to secure deals that were missed on Black Friday. This can help you capture last-minute shoppers and continue momentum throughout Cyber 5. ## Key Takeaways This analysis of previous Black Friday and Cyber Monday headlines reveals that to be successful this Cyber 5, marketers must employ a strategic mix of headline categories to ensure their ad creative connects with consumers on a deeper level. Gone are the days when you could rely on generic headlines or solely focus on discounted pricing. According to Taboola data, curiosity-driven, problem-solving, and even urgency-based headlines are consistently outperforming other categories. Also, as Cyber 5 continues to evolve, you can adapt by following emerging headline trends, such as the integration of storytelling and a greater focus on value. Finally, look for opportunities to leverage social proof and timely messaging to drive clicks and conversions during one of the busiest shopping seasons of the year. --- ### Unlock Holiday Sales: Data-Driven Creative Tips You Can Use Today URL: https://www.taboola.com/marketing-hub/data-driven-creative-holiday-sales/ Last Modified: 2025-09-09 17:03:06 During the busy holiday season, every day and every detail counts. Marketers have to craft the right creative ads to reach audiences, then measure and adjust on the fly to make sure the strategy is working. Every decision should be informed by data to engage users and provide a smoother buyer’s journey. With a data-driven framework and clear KPIs, marketers can go above and beyond on their holiday sales goals. To that end, I’ve gathered tips on how to make your own holiday marketing campaigns shine with data-informed creative. You can also check out Taboola’s holiday creative playbook for more data and market trends to inform your data-driven creative strategies. ## Seven Ways to Inform Your Data-Driven Creative Strategies ### 1. Understand Your Audience (and Their Holiday Mindset) With millions of shoppers converging on the internet toward the end of the year with the intent to purchase, knowing your audience is crucial. Without that depth of knowledge, your ads might be too broad, too narrow, or simply not speaking to the right people. “The better you understand your audience, the easier it is to create content that feels relevant and personal,” points out digital marketing specialist Honey Chhatani. “Start by understanding who you're speaking to. Tailor your messaging to resonate with their interests, motivations, and pain points.” With what you know about your audience, build visual and messaging themes accordingly, telling the stories that will resonate with them. For holiday shoppers, you might tap into emotions or otherwise convey a relevant narrative. ### 2. A/B Test Your Creative Elements A/B testing is an essential part of marketing best practices all year round. Items to test include headlines, visuals, calls to action (CTAs), and ad formats. It’s important to strike a balance with A/B testing, though. “Test thoughtfully, but don’t over-test,” suggests Chhatani. “Start with simple A/B tests, like comparing Headline A vs. Headline B, or video vs. static image. Stick to testing one element at a time. Overloading with too many variations can confuse the results and make optimization harder.” Chhatani recommends using the seasonal holiday spike in impressions to test variations quickly, rotating ads with different headlines, imagery, or CTA placement. In the meantime, monitor performance and reallocate budget accordingly. If you’re only able to measure a few metrics, you can still capture useful insights. “Start small and stay focused,” advises Chetna Rohilla, marketing assistant at NeuroSync and VP of Podcasts at the AMA Boston Chapter. She continues that, during her pro bono work with an orchestra, “One of my first wins came from simply A/B testing subject lines during a concert push. Clear, time-sensitive subject lines consistently outperformed clever copy, and that one insight shaped all email sends that followed.” “I used to think data was just numbers on a screen, until I saw how it could fuel creativity instead of killing it,” Rohilla adds. “Data’s not there to replace creativity — it’s there to sharpen it, so track what you can, and adjust as you go. Even a single metric, if tied to a clear goal, can sharpen your creativity fast.” ### 3. Analyze Performance Metrics for Creative Optimization Many marketers are able to track a wide range of metrics, but focusing on the right data is the best way to inform your decisions. Some of the key metrics to capture for holiday ads are click-through rate (CTR), conversion rates, scroll depth, and engagement. “Use these insights to tweak creative elements, improve landing pages, or even refine your audience targeting,” says Chhatani. “It’s easy to get distracted by vanity metrics such as likes or views, but the most important thing is whether your campaign is generating results. Every dollar should work for you.” The creative playbook for the 2025 holiday season recommends maintaining four to six creatives in rotation, monitoring their performance, and pausing underperforming ones. ### 4. Leverage Past Holiday Campaign Data The data from past holiday campaigns can be a goldmine for marketers. Explore what worked and what didn’t, and see which evergreen creative themes can be repurposed or reused. To use that data to inform your next holiday campaign, “Identify high-performing ad formats, messaging styles, CTAs, and visual elements,” says Chhatani. “Let data, not assumptions, guide your creative direction.” Chhatani also recommends post-holiday campaign follow-through, such as retargeting holiday buyers. “Don’t let a new customer be a one-time buyer!” she urges. “Use loyalty offers, post-purchase emails, or future promotions to nurture repeat sales.” You can also use the insights from holiday campaign performance to guide your next push, whether another holiday period or a product launch. ### 5. Understand the Role of AI in Data-Driven Creative AI tools can be useful to marketers, assisting in creative generation or analysis. It can act as a starting point for holiday campaigns and tactics, giving marketers a springboard for headlines, copy, targeting, and more. For teams looking to freshen up campaigns and rotate visuals, AI can also offer ideas for new perspectives or new language to combat user fatigue. “For time-sensitive or high-volume campaigns, using AI can give you an edge,” agrees Chhatani, “whether it’s generating personalized video ads, writing different versions of ad copy, or analyzing performance trends.” ### 6. Focus on the Right Data When you’re running a multi-channel campaign during the busiest season of the year, it’s easy to get bogged down in numbers. It can become overwhelming and hard to know which numbers really matter. “Don’t get buried in metrics,” advises Rohilla. “, I’ve focused too much on vanity numbers — likes, impressions — without connecting them to business goals. It’s easy to over-track and under-apply.” During peak summer season, for example, Rohilla found behavioral and timing data to be the most actionable for her target audience. She tested which types of phrases worked to convey urgency, and what people actually clicked on, not just what they liked, to understand better where to put marketing efforts. “In a crowded season, knowing when and why someone engages can be more valuable than broader demographic insights,” she says. Uncover which data points might be most valuable and actionable for your business ahead of the holiday rush. The tweaks you make might be small, but can have a huge impact. For example, simplifying a mobile layout or segmenting emails by prior engagement led to an increase in open rates and sales, says Rohilla: “These weren’t major overhauls, they were small data-driven decisions that compounded into real revenue results.” ### 7. Pay Attention to Timing In advertising, timing is everything, and that’s particularly true during the holiday season. Our holiday playbook recommends implementing tracking four to six weeks in advance of peak periods to gather data on users who interact with your brand. Tracking the timing of your audience is also crucial. See when your audience is most active or likely to convert. “Posting at peak engagement times can improve your reach without increasing spend,” says Chhatani. “Tools like email open rate tracking or social media analytics can help identify these patterns.” Marketers should also track KPIs in real time, or as close to it as possible. Dashboards and automated alerts can be very useful to either pivot or double down on successful tactics, Chhatani points out. ## Key Takeaways Marketers and advertisers have a captive audience during the holiday season, with tons of opportunity for sales. Bring together data insights and creative practices to make sure you’re reaching every prospect. --- ### Back-to-School Ad Headlines: Realize Data-Driven Insights for 2025 URL: https://www.taboola.com/marketing-hub/back-to-school-ad-headlines/ Last Modified: 2025-08-18 07:49:45 Heading into the 2025 back-to-school season, advertisers need to know what’s driving shopper engagement if they want to build effective ad strategies. To help, we’ve conducted an in-depth analysis of our 2024 Realize data to assess last year’s back-to-school ad headline performance. From product-focused headlines to celebrity endorsements, our insights can provide you with a roadmap to maximize your impact. Let’s start by breaking down the most effective headline categories and what they tell us about back-to-school shoppers. ## Headline Categorization: Understanding What Drives Engagement To better understand the types of content that appeal to back-to-school shoppers, we carefully categorized headlines based on their thematic content. This allowed us to compare different advertising strategies and assess their performance. Here’s a closer look at the headline categories we used: ### Product/Promotional Product or promotional headlines are straightforward, with a clear focus on advertising products, sales, discounts, or specific brands. They often highlight the discount or the product’s unique selling points. The intent behind these headlines is clear: to drive sales by clearly showing consumers what’s being offered and how much they can save. ### Solution/Benefit-Oriented Solution or benefit-oriented headlines address the common challenges that parents, students, and teachers face during the back-to-school season, and offer solutions that make their lives easier. Focus on the product’s benefits and value, such as saving time, reducing stress, or offering convenience. Examples include "Send your kids back to school with the right protection!" and "Here’s a practical guide to not breaking the bank." ### Lifestyle/Fashion/Outfits The lifestyle/fashion/outfits category features headlines that focus on style, appearance, and lifestyle. They are designed to appeal to shoppers who are concerned with the social aspects of the back-to-school season. They often highlight clothing, accessories, or personal items that can make a student feel confident or trendy. ### Informational/News/Event Informational headlines place a focus on sharing valuable news or information related to back to school. They will often report on events, human-interest stories, or safety concerns regarding the return to school. These headlines aren’t always aimed at selling a product, but rather at providing advice or updates. Examples might include “How to Protect Your Child From Back-to-School Germs,” or “2025 Back-to-School Trends You Need to Know About.” ### Emotional/Relational Emotional or relational headlines evoke feelings associated with the nostalgia and excitement of returning to school. They are designed to resonate with parents who want their children to have the best possible experience, as well as with students who are looking forward to a successful and enjoyable year ahead. Headline examples include “Make Your Child’s First Day Special With These Back-to-School Gifts,” or “Support Your Teen’s Success This School Year With These Essential Tools.” ### Generic/Call to Action Generic headlines are often straightforward and use a simple call to action. The goal is to trigger an immediate action, like a click or purchase, from the consumer. Instead of aiming to educate or highlight a product’s benefits, the marketer is trying to create a sense of urgency. Examples include “Back-to-School Essentials!” and “Don’t Miss Out on These Back-to-School Savings!” ## What Kind of Message Are Back-to-School Shoppers Engaging With in 2024-2025? The table below breaks down the clicks distribution based on the headline categories we analyzed during the 2024-2025 back-to-school season. As the data shows, shoppers were highly engaged with certain types of headlines, and less so with others. Note: The following percentages in this table are illustrative, reflecting the relative click trends observed across the dataset’s diverse headlines and their categorized examples. Headline Category Clicks Distribution (2024-2025 Back to School) Product/Promotional 46.51% Solution/Benefit-Oriented 24.12% Lifestyle/Fashion/Outfits 12.06% Emotional/Relational 8.18% Informational/News/Event 6.46% Generic/Call to Action 2.67% As you can see, shoppers have a clear preference for headlines that address product value or offer practical solutions. This perhaps isn’t surprising, as back-to-school shoppers are looking for competitive prices and ways to make their lives easier during what can be a stressful time. ## Patterns in Headline Effectiveness ### Value and Clear Propositions Drive Engagement Headlines that mention a product, savings, or make a clear value proposition perform consistently well. This indicates that shoppers are drawn to offers that promise significant value or substantial savings. ### Solution-Based Headlines Resonate As evidenced by a 24.12% click ratio, shoppers also respond to solution-based back-to-school headlines. According to Esther Altomare, head of industry enterprise advertising at Taboola, “During high-stakes, time-sensitive seasons like back-to-school, shoppers are under pressure to make decisions quickly. Headlines that are clear, urgent, and relevant instantly communicate to busy shoppers: ‘This solves your problem now.’” Take the following headline, for example: “Here’s a practical guide to not breaking the bank this back-to-school season.” This appeals to parents who want to do more than just spend money and shop — they’re looking for ways to prepare for and manage the school year. ### Interactive Content Performs Well Interactive or engagement-driven headlines, such as quizzes and shopping guides, generated high clicks in 2024. Clearly, content that invites participation taps into consumers’ curiosity and their desire for more personalized experiences. ### Celebrity Endorsements Grab Consumer Attention Advertisements that feature a celebrity endorsement saw a significant increase in clicks. When a celebrity or well-known public figure endorses a product or service, it creates immediate credibility that is difficult to achieve through other means. Celebrities also have large, loyal followings, so they bring their own built-in audience. ### Generic and Informational Headlines Don’t Perform as Well In our analysis, broad headlines that focused on general news, entertainment, or educational topics underperformed, with a click ratio below 10%. It’s clear that shoppers are more interested in finding solutions to their immediate needs through things like product discounts and lifestyle improvements. ## Back-to-School Keyword Trends ### Brand and Celebrity Associations When headlines include keywords of specific brand names and associated celebrities, they generate incredibly high engagement. Established brands working with public figures are able to move from generic product curiosity to specific, high-intent engagement. ### Financial Keywords Our analysis revealed that keywords directly tied to financial benefits, like “Savings,” “Offers,” “Deals,” or “Loans,” consistently drive high engagement. This shows that consumers are highly focused on value during the back-to-school season. This was also true when we compared headlines that emphasized “Savings” or “Loans” and those with more generic terms, such as “Essentials” or “Gear.” In other words, advertisers who directly address consumers’ financial concerns are well-positioned to capture their attention. ### Practical Solutions In 2024, keywords that offered practical solutions were highly effective. This can include terms like “protection,” “healthy,” and “simplified,” and it shows that shoppers want real-world solutions that can make navigating going back to school easier. As Altomare explains, it’s all about cognitive ease: “Simple, benefit-focused headlines reduce mental effort. Busy parents/students gravitate to clear solutions.” For 2025, highly-engaging keywords like “lunch box ideas,” and “sweet and savory lunchbox snacks,” are driving traffic. This reinforces our 2024 findings that consumers respond to content that offers clear, actionable solutions. ## Strategic Advertising Recommendations Based on our findings on headline effectiveness and keyword trends, here are some ways you can optimize your back-to-school advertising campaigns: ### Prioritize Value-Driven and Solution-Oriented Headlines Make sure that your back-to-school headlines focus on financial benefits by promoting discounts, savings, and actionable tips that address the challenges of the back-to-school season. This can include topics such as health, organization, meal prep, and anxiety management. ### Leverage Brand and Celebrity Associations If possible, leverage your brand through partnerships with relevant celebrities and influencers. Effective endorsements can help you cut through the noise and quickly grab consumers’ attention. ### Embrace Interactive and Utility-Driven Content Formats Shoppers love content that engages them beyond a basic product listing. Experiment by incorporating quizzes, surveys, or actionable tips, which makes your advertisement more of a helpful resource than a sales pitch. This approach can build trust and boost client engagement. ### Invest in Multilingual and Localized Campaigns As you build your back-to-school ad campaign, consider using non-English headlines for specific linguistic and cultural markets. Our data shows that Spanish, German, Thai, and Greek headlines resonate well with their respective audiences. Ignoring non-English headlines could result in you missing out on significant market potential. ## Key Takeaways This analysis of 2024 back-to-school ad headlines reveals some key insights. For starters, shoppers are much more engaged with headlines that highlight value and practical solutions. They’re also interested in products that offer financial incentives, such as discounts and special offers. Altomare sums it up nicely, stating that, “solution-oriented creative goes a long way, so focus on lists, urgency, and solutions.” --- ### Back-to-School Retargeting Strategies to Reach Students Across the Web URL: https://www.taboola.com/marketing-hub/retargeting-strategies-back-to-school/ Last Modified: 2025-10-26 17:35:27 Students are a valuable demographic for today’s marketers. This is especially true at the college level, where students have a combined spending power of nearly $600 billion. You’re also reaching customers who are still open to forming new brand loyalties, which means one conversion can bring years of purchases. Reaching those students online brings some challenges, though. Today’s college campuses are filled with distractions, and on top of that, Generation Z has developed a mistrust of traditional advertising. Authenticity wins, but how do you get the word out about your products without overtly promoting them? The good news is, the right tools can help you get around these roadblocks. Here are some strategies to help you as you’re planning your retargeting campaigns. ## How to Retarget Ads for Students: 7 Key Strategies and Best Practices With Realize Let’s say a student visits your website or clicks on an ad, but never makes a purchase. The next step is typically to retarget that customer with an offer designed to convert, but with students, traditional retargeting strategies don’t always work, and that’s where the following techniques can make a difference. ### 1. Layer Targeting Signals One-dimensional targeting isn’t enough for today’s savvy students. The best results come when you layer data signals to reach the right individuals in the right mindset at the best possible time. “The winning combination is contextual plus behavioral, and demographic plus location,” says Tomer Tunitsky, product manager at Taboola. “Target users reading relevant content, like student finance or specific majors. Layer in age data (18-24) and inferred student-like behaviors. Then add geographic targeting like zip codes near college campuses.” The key components of this strategy are: - Contextual targeting: Show ads alongside relevant content (e.g., student budgeting tips). - Behavioral and demographic filters: Combine declared age ranges with Realize’s inferred student-like behaviors. - Location targeting: Narrow audiences by U.S. zip code to home in on campus communities. ### 2. Capture Students Mid-journey on the Open Web Unlike search or social, the open web gives you a chance to catch users while they’re actively researching related subjects. They might be reading an article, exploring career advice, or looking into loan options when your ad appears alongside the content. If you’re marketing an educational or service- or subscription-based business, you can especially benefit from this approach. “Our main edge is combining all these signals on the open web,” says Tunitsky. “We capture users in their moment of discovery, when they are actively researching topics.” Use this approach to: - Deliver native ads while students explore relevant content. - Complement social and search strategies with discovery-driven touchpoints. - Position your message as a useful nudge while they’re researching related topics. ### 3. Combine Interest and Topic Targeting for Flexibility While student populations have many things in common, each person is unique. That means each student comes with a separate set of interests. Realize builds in two tools to help with that: Interest Targeting and Topic Targeting. With Interest Targeting, you get hundreds of prebuilt categories, including: - Education. - Teaching. - Test prep. “Start broad with Interest categories (like ‘education’),” says Tunitsky. “Then, use performance data to identify specific high-performing themes and create more granular segments with the Topic targeting tool to double down on what works.” Another option is Topic Targeting. This extremely flexible tool will let you build custom audiences using free-text searches, like “scholarships for engineering students” and “how to pay for college without loans.” “Topic targeting should be run as a separate campaign,” advises Ed Lovelock, digital and product marketing leader at Taboola. For best results, start with broad, interest-based segments, then monitor performance by theme. Once you have enough information, launch topic-specific campaigns around high-performing interests. This type of layered targeting lets you scale reach, then fine-tune based on what resonates. ### 4. Optimize With Real-Time Audience Reporting Real-time reporting can help you move beyond clicks and impressions. You can analyze performance, adjust budgets, and tailor creatives based on what you learn. “Identify audience segments that are working/not working and adjust bids accordingly,” Lovelock says. “Demographic breakdowns help, too — it may be students or parents buying, and creatives can be tailored accordingly. Tunitsky agrees. “Our reports show you which audiences drive the best results. Use these insights to shift budget to what’s working and refine your messaging.” Using audience reporting, you can: - Evaluate creative performance by age, gender, or location. - Tailor copy to students or parents, based on how they’re responding. - Discover high-value audiences using the Audience Explorer tool. ### 5. Extend Reach Through Multiplatform Tools Students rarely convert the first time they see an ad. To break through the constant stream of information coming at them, you need to follow up with messaging that incites action. Realize can help with that. The built-in AI-powered tools can drive mid-funnel traffic, which can then be retargeted through Meta, Google, or even Realize itself. Each of these sites requires separate pixels, but Realize can handle that for you, integrating with both to easily boost your audience-building efforts. “Realize can deliver traffic (Intent Prospecting) to advertiser websites,” Lovelock says. “If they don’t convert there and then, advertisers can retarget on Meta, Google, or Realize because those pixels will have captured the visit.” Tunitsky adds that pixels placed through Realize can help: - Segment users based on on-site actions. - Exclude users who have already converted. - Build powerful Predictive Audiences using successful converters. This helps you avoid wasting spend on users who’ve already taken action, while also letting you focus your budget on those who are still in the decision-making phase. It also helps you guide each student through every step of the funnel without needing to start over every time they switch platforms. ### 6. Scale Using Predictive Audiences As your campaign begins capturing data, Realize’s Predictive AI kicks in. With each new data point, it learns more about your audience. Soon, it gains the ability to identify common traits of converters and apply that information to its future targeting efforts. Here’s how Predictive Audiences works: - Your Meta and/or Google Pixel conversion data is fed to Realize. - Realize models audiences with similar traits. - You can then target lookalike users on the open web to scale your results. These steps increase your chances of engaging students at scale. It can also be combined with other strategies, such as showing testimonials after an initial engagement to build trust. ### 7. Use Location Targeting to Zero in on Campuses Targeting by physical location remains one of the most efficient ways to advertise to student populations. Homing in on college campuses will ensure you’re working within large concentrations of students. “We can target by zip code in the U.S., which is perfect for focusing on specific university campuses and student areas,” Tunitsky says. The best use cases for location targeting are: - Targeting specific universities or college towns. - Localizing creatives with city or campus names. - Promoting in-person activations, pop-up events, or seasonal student offers. Realize’s granular location capabilities can be combined with interest and behavioral targeting to create more precise student campaigns. With location targeting, you can ensure your message arrives in front of students where they live, study, and spend time. Interest and behavioral targeting delivers your content specifically to the students who are most likely to engage and convert. ## Key Takeaways Successfully targeting today’s students is all about layering. Combining contextual, behavioral, demographic, and location-based data allows you to reach students when and where they’re likely to be most receptive. Realize’s Interest Targeting lets you start broad, then refine based on performance using Topic Targeting. The audience reporting tool helps you easily identify what’s working and further refine your efforts. Once you’ve gathered enough conversion data, you can use Predictive Audiences to find new audiences who behave like your best leads, then layer in zip code-based targeting to localize your messaging and maximize relevance for student communities. ## Frequently Asked Questions (FAQs) ### Who is the target audience for retargeting ads? Retargeting is for people who have previously interacted with your brand. Someone might have clicked on an ad or social media post, visited your website, or otherwise shown an interest in your offerings, then moved on. With retargeting, you can deliver additional content to those users with the goal of driving conversions. ### Where should you advertise to students? As with any demographic, you should advertise to students where they typically hang out. That includes social media platforms like TikTok and YouTube, but you can also gain traction by targeting their favorite place for research — the open web. Ads can be delivered to students while they’re reading about college life, preparing for tests, and researching upcoming events. ### How much does it cost to retarget ads? Retargeting costs between $0.25 to $0.60 per click on average, which is a fraction of what you’d spend on a search ad geared toward prospecting. Best of all, since retargeting focuses on users who’ve already interacted with your brand, they’re more likely to get results, boosting your return on ad spend. ### What is a good CTR for retargeting ads? The average click-through rate (CTR) for retargeting ads is 0.7%, compared to the average CTR for display ads, which is 0.07%. If you’re targeting college student audiences, keep in mind that CTRs are typically higher, particularly when it comes to email campaigns. If you’re seeing disappointing CTRs, make sure your creatives align with their specific phase of the decision-making journey. --- ### Sports Marketing Trends 2026: How Teams Are Winning Fans in the Digital Era URL: https://www.taboola.com/marketing-hub/sports-marketing-trends/ Last Modified: 2026-04-30 19:08:51 Today’s sports fans expect more than just highlights and final scores. They want access, personalization, and a greater connection to the teams and athletes they follow. And, they want all of this on their terms, across multiple platforms, and in real time. As a result, sports leagues and organizations are no longer just competing on the field, they’re competing for attention in a crowded digital landscape, where content, community, and experience might matter as much as performance. In 2026, the most successful teams and brands are thinking like media companies, using AI to scale content and personalize it, and building direct relationships with fans rather than relying solely on third-party platforms. The following trends highlight how sports marketing is evolving, and where the biggest opportunities lie for digital advertisers. What’s changed in our 2026 update: - Five of the seven listed trends are new for this update; new trends added are: - The 2026 FIFA World Cup Will Redefine Global Sports Advertising - Teams Are Becoming Media Companies - AI Is Powering Real-Time Personalization at Scale - The Rise of Creator Ecosystems and Athlete-Led Media - The Surge of Women’s Sports Fandom - All entries include updated and current information, advice, and stats. ## Trend 1: The 2026 FIFA World Cup Will Redefine Global Sports Advertising The 2026 FIFA World Cup isn’t just another major sporting event — it’s set to become one of the largest global media moments in history. With matches located across the U.S., Canada, and Mexico, and an expanded 48-team format, the tournament is expected to reach a record-breaking global audience. What will really set 2026 apart isn’t just the size of the audience, though, but the way it will engage. According to a WARC (World Advertising Research Center) study, FIFA has partnered with TikTok for the event, signing a “Preferred Platform” agreement that will grant creators “behind-the-scenes access and archive rights.” There have already been 1.4M TikTok posts using the #FIFAWorldCup hashtag, months before the event is due to begin. Additionally, per the study, 85% of fans use TikTok as a second screen during live events. Even Netflix is getting in on the action, agreeing to stream Goalhanger’s popular “The Rest Is Football” podcast daily during the World Cup. This expansion of content beyond the actual TV match coverage to the surrounding conversation on various media platforms will be unprecedented. Dan Holt, strategy partner, Havas Media Network, told the WARC that, “The World Cup match will be the base layer. Meaning will be assembled elsewhere, across social feeds, watch-alongs, creator commentary, group chats, memes, and remixes.” For digital advertisers, this shift creates an opportunity to extend beyond traditional media buys and show up where the conversation is actually happening. Brands that can align their messaging with real-time moments, creator content, and second-screen behavior will be better positioned to capture attention throughout the entire fan experience, not just during the match itself. ## Trend 2: Teams are Becoming Media Companies In 2026, sports teams and leagues are no longer just the subjects of content; they’re full-fledged content creators. From behind-the-scenes footage and locker room access to player-led storytelling and documentary-style series, teams are using platforms like TikTok, YouTube, and Instagram to produce always-on content designed to engage fans beyond game day. For example, FC Barcelona (@fcbarcelona) is using its TikTok account, which has over 65 million followers, to share exclusive footage of team practices, locker-room celebrations, and warm-up routines that fans don’t see during regular television broadcasts. Likewise, Formula 1’s TikTok account offers behind-the-scenes footage, including team radio clips, pit crew challenges, and casual driver interactions. At the same time, community-building has become a core part of the playbook. Fans are gathering in Discord servers, Reddit forums, and other private groups to connect, debate, and rally around their favorite teams. These communities provide digital advertisers with a goldmine of insights, and they’re using them to their advantage by offering more interactive content and gamification, such as fantasy sports leagues, prediction games, and live trivia, where fans are rewarded for participating. User-generated content is also playing a bigger role. The most effective advertising campaigns are encouraging fans to participate by sharing reactions, traditions, and moments. This turns passive viewers into active contributors. In 2026, the goal isn’t just engagement, it’s ownership of attention and community. ## Trend 3: AI Is Powering Real-Time Personalization at Scale Personalization in sports marketing is no longer just about segmentation, it’s also about real-time decision-making that’s powered by AI. Sports teams and advertisers now have access to massive amounts of first-party data, from app usage and ticket purchases to viewing habits and in-game interactions. But, the real shift is in how that data is being used. Consider what’s possible using AI-driven systems: - Dynamic ticket pricing based on demand signals. - Real-time content recommendations during live events. - Personalized merchandise offers triggered by key moments. - Automated cross-channel campaign optimization. Instead of manually building campaigns, marketers can define the inputs and let their AI systems optimize the outputs, testing creative, adjusting messaging, and allocating ad spend in real time. Predictive analytics is also becoming more actionable. Sports teams can now identify which fans are likely to attend games, upgrade their seats, or become season ticket holders. They can then tailor experiences accordingly. As personalization becomes more sophisticated, though, so do expectations around privacy. Maintaining trust through transparent data practices remains critical. ## Trend 4: The Rise of Creator Ecosystems and Athlete-Led Media Source: YouGov Influencer marketing has evolved into something much bigger in 2026. It’s now a creator-driven ecosystem where athletes, teams, and popular sports influencers all play a role. Major influencers can deliver authentic, trustworthy, and relatable content to highly engaged audiences across platforms such as TikTok, Instagram, and YouTube, boosting a sports brand’s reach and image. Even smaller influencers are in demand, with sports brands actively sourcing talent in communities like Reddit. These aren’t one-off campaigns, either — they’re long-term collaborations that prioritize authenticity and consistency. This shift gives brands an opportunity to tap into existing fan trust, reach niche audiences through creator communities, and produce content that feels native, rather than promotional. It also reflects a broader shift in how sports marketing campaigns are planned and executed. As Alex Henderson, sports sponsorship lead at M+C Saatchi Fluency, told Zappi, “Plan from cultural moments, not partnership menus. Don’t default to signage or social posts. Do the research to map the moments your audiences care about. Then design brand-led experiences that will deliver meaning in those moments.” In this model, creators play a key role, helping brands show up in those moments in ways that feel timely, relevant, and authentic to fans. To measure the impact of influencer marketing, digital advertisers can track engagement metrics like views and conversion rates. They can also measure engagement on emerging formats, such as livestream takeovers and exclusive fan Q&As on platforms like TikTok or Twitch. Name, Image, and Likeness (NIL) rules are also changing the game in sports marketing. In 2021, the NCAA began allowing student-athletes to earn money from their personal brands. According to ESPN, top athletes have secured deals with brands such as Nike, Adidas, and Under Armour. Even non-sports brands, like T-Mobile and Amazon, have made deals with students. - According to a 2025 Sprout Social survey, 86% of marketers from companies with over 100 employees expected to use influencer marketing during the year. - According to data from FanZGo, athletes have an engagement rate of 5.6% on social platforms, compared with 2.4% for non-athlete influencers. - 63% of fans between 18-34 trust athlete endorsements. ## Trend 5: The Importance of Video Content and Livestreaming in Sports Marketing Unsurprisingly, video is firmly established as the primary medium of digital sports marketing. But while the way people consume sports content on video is evolving, high-quality content remains a non-negotiable if you want to hold fans’ attention. Fans want to see more than just game action or final scores: They’re looking for visual storytelling that brings them closer to the athletes. To accomplish this, brands are livestreaming not just games, but behind-the-scenes access, press conferences, and training camp sessions. As mentioned, sports marketers are also using short-form video platforms, such as TikTok or YouTube Shorts, which are well-suited for reaching younger audiences. If you’re creating video content for different digital channels, such as short-form, make sure you optimize each piece for the respective platform. For example, use captions where applicable to enable silent viewing, along with quick, attention-grabbing intros on social. In addition to livestreaming, interactive video is also on the rise in sports marketing. Fans now have the option of choose-your-own-view cameras and real-time polls, while clickable stats overlays are often available during livestreams. - According to Adwave, approximately 70% of fans now watch sports via streaming platforms. - Amazon is the most used platform for streaming sports, garnering 65% of respondents in a 2025 Performance Research survey (reported by the Sports Business Journal). - 48% of fans say that streaming makes them feel more connected to their favorite teams and sports, per the same survey. ## Trend 6: The Integration of E-commerce and Digital Experiences in Sports In 2026, sports fans are connecting with their favorite teams online more than ever. In fact, the licensed sports merchandise market is expected to grow at a compound annual growth rate (CAGR) of 6.1% until 2033. Professional sports teams and leagues are responding to the demand by doubling down on e-commerce to boost revenues from licensed sports merchandise. (Source: Grand View Research) AI is playing a key role in this evolution. Teams can now deliver personalized product recommendations and trigger merchandise drops based on real-time moments, like a major trade or a championship win. Digital platforms are enabling fans to have virtual experiences that bring them closer to the action. For example, NBA fans who own a Meta Quest VR headset can watch NBA games in virtual reality. When it comes to getting into live games, mobile ticketing is now the standard. These digital passes can be bundled with personalized perks, in-seat ordering, and even in-game entertainment, like polls and trivia. - The licensed sports merchandise market size was over $37 billion USD in 2025. - The market size is projected to grow to $59.4 billion by 2033, per the same report. While digital collectibles, including NFTs, and limited-edition content still exist as niche engagement tools, the bigger shift is toward seamless, end-to-end digital ecosystems that connect content, commerce, and live experiences. In 2026, the most successful sports organizations aren’t just selling merchandise, they’re building integrated digital experiences that turn moments of fandom into measurable revenue. ## Trend 7: The Surge of Women’s Sports Fandom Women’s sports are no longer a niche category in 2026. In fact, they’ve become one of the fastest-growing segments in the entire sports landscape. Recent data from an L.E.K. Consulting Sports Survey shows that engagement is accelerating across key demographics. Overall, women saw an 8% increase in sports engagement, signaling a broader shift in who participates in sports fandom. At the league level, the growth is even more striking. The WNBA led all major leagues with a 65% year-over-year increase in avid fandom, outperforming traditional powerhouses and highlighting a meaningful change in fan behavior. This momentum is being driven by a combination of factors, including the quality of play, the rise of high-profile athletes like Caitlin Clark, and a growing cultural emphasis on representation and equity in sports. Star players are also playing a major role in expanding visibility and engagement, with breakout moments translating directly into increased viewership and media value. Importantly, this isn’t a short-term spike. Fans expect their engagement with women’s sports to continue growing, with a net 14% increase in anticipated viewing time over the next 12 months. For marketers and advertisers, this presents a major opportunity. Women’s sports audiences tend to be younger, more diverse, and highly engaged, making them especially valuable for brands looking to build long-term relationships with emerging fan segments. In 2026, the only question around women’s sports is how quickly brands can adapt to keep up with their growth. ## Key Takeaways Fan expectations have never been higher. As a result, sports marketing campaigns need to drive community engagement, deliver personalized content, leverage influencer partnerships, feature immersive video, and deliver seamless e-commerce experiences to be successful in 2026. Not surprisingly, data plays an important role. At the same time, new formats and livestream takeovers are providing sports organizations with new revenue streams and an opportunity to increase fan loyalty. Marketers who can make fans feel like they’re a part of something bigger will increase their chances of success. ## Frequently Asked Questions (FAQs) ### What are the most effective digital advertising channels for sports marketing in 2026? Short-form video engagement is becoming more critical in 2026. As a result, sports marketers must include platforms like TikTok and Instagram in their marketing strategies. YouTube works well for long-form content and livestreams, and emerging platforms like Discord and Threads are becoming prime locations for community-building. ### How can sports brands measure the ROI of their digital marketing efforts? Sports brands should track key KPIs, including CTR, video view rates, branded search lift, social engagement, merchandise conversion rates, and fan lifetime value (LTV). ### How is technology (e.g., VR, AR) impacting sports marketing? Sports marketers are using AR/VR technology to enhance fan engagement in 2026. This includes producing virtual events, gamified content, and fully immersive in-stadium experiences. For example, fans can use VR headsets like Meta Quest to watch games courtside from the comfort of their own homes, while teams are starting to offer in-app AR features that let players try on jerseys or interact with polls and trivia during games. ### What are the key metrics for measuring success in sports digital marketing? Sports marketers can measure success through several key metrics, including social media engagement rates, video views, CTRs, and conversion rates for various campaigns, including ticket and merchandise sales, or app downloads. --- ### AI-Powered Holiday Ads: Tips for Writing Effective Prompts URL: https://www.taboola.com/marketing-hub/ai-powered-holiday-ads/ Last Modified: 2026-05-28 13:01:26 Busy marketers can get a lot of benefit from using AI to help create ads, especially during a fiercely competitive time like the holiday season. AI can help advertisers keep up with demands without compromising creativity. Brands taking advantage of generative AI can produce high volumes of quality ad content across a range of formats, while also personalizing messaging to specific audiences or product lines. SMBs and mid-sized e-commerce brands, in particular, benefit tremendously from using AI. These advertisers often don’t have massive creative teams, yet still need to produce platform-specific ads at scale. AI helps close that gap, producing work like product carousels, native ad headlines, landing page copy, and more. During the holiday season, the stakes are higher, and the timelines are tighter, so AI can give marketers a head start when they need it most, delivering rapid ideation, first-draft copy, and campaign variations in minutes. The essential skill to master when you’re using AI in creative ways for advertising is the prompt. Here’s what to know to use this tool for holiday season success. ## First Things First: Aligning Brand With Your Audience As with other marketing and advertising tools, it’s essential to understand your audience when using AI. As you’re planning for the holiday season, use all the available data to anticipate who you’re targeting and what their shopping and buying needs are. From there, you can tailor and align your brand messaging with your audience (or various audience segments) to create the ideal set of content and messages that people will engage with. AI excels at structure, speed, and ideation. You bring the strategy, empathy, and edge that makes it convert. ## How to Make Prompts Work for You Maintaining consistency across all your ads and formats during the holiday season starts with the prompt. Treat your prompt like a creative brief: include tone, audience, product focus, and desired action. AI may not always deliver a polished, final ad, but it can get you about 70% of the way there — a huge time savings, especially for lean teams. AI doesn't think like a human, but it does learn from us. When you prompt a generative AI tool, you're not giving it instructions in the way you would a colleague: Instead, you’re feeding it patterns to predict likely responses based on its training. So, think of prompting like programming a conversation. The AI tool scans your input for structure, intent, keywords, and tone. It draws on billions of data points — past writing styles, ad language, customer behavior — to generate the most statistically relevant output it can. But, it doesn’t necessarily know what your brand sounds like or what will convert best for your specific audience unless you tell it. With that in mind, here’s what you’ll likely get back from AI once you’ve entered a solid prompt: - A strong first draft. - A range of viable headline or call to action (CTA) options. - Consistent tone (if prompted clearly). - A solid base for further human editing. These outputs are a significant head start for marketing and advertising teams creating ads at scale, especially during peak moments like the holiday season. The trick is knowing how to shape the final 30% with your brand insight, emotional nuance, and performance data. ## The 5 Core Principles of Effective AI Prompt Writing Whatever the season, follow these principles when prompting to get the best outcomes. ### 1. Be Specific The more context you provide in your prompt, the better that AI can tailor the content. Don’t use, “Write an ad:” Instead, try something like, “Create a native ad headline for a holiday gift guide featuring budget-friendly beauty products.” ### 2. Define the Audience or Persona Clarifying who the end user is can help the AI strike the right emotional chord. Instead of, “Write something for shoppers,” use “Target busy moms shopping last-minute for stocking stuffers under $25.” ### 3. Set the Format and Length These constraints help guide the AI and save you time on editing later. Don’t write, “Give me ideas.” Try something like, “Generate three ad headlines, each under 60 characters.” ### 4. Specify Tone and Style Include adjectives that match your campaigns and goals, such as bold, premium, casual, witty, nostalgic, or informative. Don’t tell the AI to “Write it nicely” — instead, say something like, “Use a playful and upbeat tone with a holiday spirit.” ### 5. Mention the Goal Telling the AI tool up front what success looks like will improve the clarity of the CTA. Don’t prompt, “Write a landing page.” Try, “Write a landing page that encourages sign-ups for a limited-time holiday offer.” ## How to Prompt for Different Holiday Ad Formats Getting specific on formats when you’re prompting AI will bring you better up-front ideas and possibilities. Remember that you can reuse key input elements like product descriptions, benefits, or audience personas across formats. You can even prompt AI to do something like, “adapt this native ad headline into a display ad CTA,” to maintain alignment while adjusting to format constraints. Holiday ads tend to follow emotional and thematic patterns (urgency, gifting, nostalgia, savings), which generative AI models are particularly good at replicating if prompted well. If you're writing for multiple formats (display, native, landing pages), keep the brand voice constant by specifying tone adjectives (e.g., “premium,” “playful,” “no-nonsense”), and refer back to shared value propositions across outputs. Here’s what else to know about writing AI prompts for different ad formats. ### Display Ads What matters: Display ads need to communicate value instantly. You have limited space, so every word — and every pixel — counts. A strong image, bold headline, and focused CTA are essential, especially during the noisy holiday season. Prompting tip: Keep your prompt tightly structured. Include details like product type, value prop, tone, urgency, and any visual elements you want highlighted (e.g., product shot, holiday theme). Prompt example: “Write a high-impact display ad headline (max 30 characters) and supporting text for a holiday flash sale on premium skincare gift sets. Emphasize luxury, limited-time urgency, and gift appeal. Include a three-word CTA. Suggest a visual direction for the image.” Example output: - Headline: “Glow for the Holidays” - Subtext: “Limited-Time Gift Sets – 40% Off” - CTA: “Shop the Sale” - Visual direction: A soft-focus image of elegantly wrapped skincare products on a marble vanity with warm golden lighting, subtle snowflakes overlay, and a red ribbon tied around the hero product. Refinement tip: If the tone feels generic, revise the prompt to include adjectives like “premium,” “soothing,” or “gift-ready.” ### Native Ads What matters: Curiosity, storytelling, and contextual alignment with the surrounding content. Prompting tip: Frame your prompt like you’re briefing a copywriter. Add the target persona, emotional hook, and platform context (e.g., Taboola, mobile feed). Prompt example: “Create three native ad headlines (max 60 characters each) for a holiday campaign targeting busy parents shopping for educational toys. Use a warm and helpful tone.” Example output: - “Savvy Moms Are Buying These Toys” - “5 Top-Rated Gifts for Curious Little Minds” - “Smart Toys Parents Are Rushing to Buy This Holiday” Refinement tip: Prompt the AI to focus on a specific feature (e.g. STEM, sensory play, age range) to align better with campaign targeting. ### Landing Page Content What matters: Clear structure, persuasive copy, and aligned tone throughout the page. Prompting tip: Specify the structure you want — headline, subheadline, product description, CTA — and the goal of the page (e.g., conversion, education, email sign-up). Prompt example: “Write landing page copy for a holiday deal on noise-canceling headphones. Structure it with a headline, subheadline, short product description, and CTA. Use a confident, tech-savvy tone.” Refinement tip: Ask the AI to rewrite for different personas (e.g., gift giver vs. self-purchaser) to match different funnel stages. ## 5 Best Practices for Prompt Refinement As discussed, a good AI prompt will get you about 70% of the way there. But, refining your prompt with human insight can take you the rest of the way. The most common mistake in writing ad prompts for AI is being too vague. Prompts like, “write an ad for our product” give AI very little to work with and typically result in generic, uninspired copy. Try these tips to refine your holiday ad prompts: ### 1. Start Broad, Then Narrow Down Begin with a general prompt to test how the AI interprets your request. Then revise by adding clarity around tone, audience, and format. ### 2. Change One Variable at a Time When refining, avoid rewriting the entire prompt each time. Isolate one aspect, such as tone, CTA, persona, etc., so you can learn what changes actually improve performance. ### 3. Add Context Without Overloading AI performs better with context, but too much information in one prompt can confuse it. Use clear, plain language and only include what’s essential. ### 4. Use Examples as Guardrails Guide the AI with samples of what “good” looks like. You can include one or two reference headlines or tone descriptors to nudge the model in the right direction. ### 5. Review and Re-Prompt With Feedback AI outputs are not final drafts. Treat them like a first round of copy: If something feels off, prompt again, this time with explicit feedback. ## Key Takeaways Using generative AI in holiday ad campaigns can save marketers a ton of time and lead to better outcomes, helping a small team scale faster and far beyond what they could do alone. The key to success is writing great prompts: Be specific, include goals, ensure consistency, and include tone when you’re writing AI prompts. Then, refine and re-prompt to get the best results. ## Frequently Asked Questions (FAQs) ### How can I use AI to write high-converting holiday ad copy for a Dyson vacuum cleaner? Combining the capabilities of AI with insight into current market trends can help you create the right ad copy for the product you’re marketing. Use the prompt advice above to input what you know about the audience, campaign goals, and tone, and ask for several versions of headlines and copy. You can then perform A/B testing. You can also explore current headline trends in Realize’s real-time Trends dashboard, and enter several options to see what’s likely to get the highest CTR. For a Dyson vacuum cleaner ad in particular, include holiday themes and language, and highlight Dyson’s unique selling points and giftworthiness. ### How can I use AI to create multiple holiday ad variations for A/B testing? With the right prompts, AI can generate multiple holiday ad variations that you can then A/B test ahead of the busy holiday season. Prompt AI to create several variations that are distinct from one another, making sure to specify which element should be emphasized in each variation. You can also ask for copy optimization if you have some options created already. Realize’s Trends dashboard can help you see which ad variations have the best chance of high click-through rate (CTR), as well as show data-driven insights into what’s trending in your particular market or industry. ### How can I adapt evergreen ads for the holiday season? The busy holiday season requires reuse and adaptation to succeed without starting from scratch. Evergreen ads can often be adapted with just a few tweaks, with the benefit of those ads already being on-brand visually and on-message in terms of copy and language. For an even more efficient approach, Realize’s Social Importer tool helps users recycle successful existing assets into new display or native ads to scale much faster. --- ### Realize Trends: Write Headlines That Attract and Engage Readers URL: https://www.taboola.com/marketing-hub/how-to-write-engaging-headlines/ Last Modified: 2026-01-20 12:21:45 In today’s competitive search marketplace, headlines are everything. They don’t just get attention — they win the click, which is exactly what you’re working so hard to achieve. The holiday season brings even more pressure for content marketers. As sales surge, competition intensifies, leaving you challenged to cut through the noise. Artificial intelligence (AI) can give you that edge you need, whether you’re blogging on your own website or setting up ads and social media posts. Realize Trends is designed to help you improve your holiday headline game, whether you are optimizing for back-to-school or Cyber 5 week, with built-in insights that show you, in real time, the headlines that are getting results. Maayan Leshem, director of creative shop and AI strategies at Taboola, shows us how to use those insights to create scroll-stopping, conversion-driving headlines. ## Why Should You Care About Headlines? First impressions are important, and your headline serves as your introduction to a potential customer. Whether it’s a blog title, native ad, or social media post, your headline tells users what to expect — and, more importantly, whether they should engage or keep scrolling. That said, the holiday season brings a shift in online behaviors. Online shopping can become more frantic as users scramble to buy gifts without going broke. Consumer searches may also be driven by nostalgia for holidays past, giving marketers a valuable opportunity to connect with new audiences. “The Headline Trends section in Realize identifies weekly-updated performance patterns in top-performing headlines across Taboola’s platform,” says Maayan Leshem. “This lets marketers align their headlines with real-time consumer engagement trends, instead of relying solely on intuition or past-year data.” ## 4 Tips to Write Engaging Headlines with Realize Trends Creating effective headlines combines creativity with hard data and timing. This is especially true during the holiday season, when brands have a short timeframe to engage potential customers. Here are some tips that can help you boost engagement during the final weeks of the year. ### 1. Spot the Trend The Headline Trends feature makes it easy to view trending topics to inform your content strategy. Because the results are in real time, you can quickly gather information on what’s working during the current holiday season. The key is to find the right verbiage for your industry segment. “For holiday creatives, marketers can spot trending adjectives, keywords, and verbs tailored to seasonal sentiment — like ‘festive,’ ‘save now,’ and ‘limited edition,’” Leshem says. She adds that headline themes like “urgency,” “self-gifting,” “nostalgia,” and “celebration” pack a solid emotional punch during the holiday season. ### 2. Use the Right Filters Headline Trends lets you pull top-performing keywords, phrases, and emotional triggers across multiple channels, but to get those results, you’ll need to choose the right filters. During the holiday season, the most useful filters are: - Vertical: See what’s working for your category, whether it’s fashion, pets, electronics, or beauty. - Country and language: Localize your language and references for cultural relevance, which is crucial for holiday-season campaigns. - Device: Mobile and desktop shoppers tend to engage differently. For instance, mobile users might prefer shorter, action-driven headlines and concise, action-driven copy. ### 3. Choose the Right Words Words matter. According to Taboola data, the following words, when used in a headline, can be highly impactful, particularly during the holiday rush: - Adjectives: Cozy, perfect, limited, exclusive, and best-selling. - Verbs: Grab, shop, unwrap, gift, save, and treat. - Keywords: Holiday, deal, last chance, for her/him, stocking stuffer, and free shipping. “These words evoke emotion, action, and timeliness, all of which are critical in holiday shopping decisions,” Leshem says. ### 4. Balance Urgency With Authenticity Urgency incites action, and that’s especially true during the holiday season. Phrases like “Time Is Running Out” and “Last Chance for Christmas Delivery” can increase conversions. Overuse of these terms may undermine your credibility, though, so that’s where balancing it with authenticity can help. “Balance comes from pairing urgency phrases with authentic emotional or value-based appeals,” Leshem says. “Instead of just, ‘Last Chance!’, try, ‘Last Chance to Gift Joy’ or ‘Ends Tonight: Wrap Up the Perfect Surprise.’” You can also create urgency using specifics like “Only 200 Left” and “Ships in 24 Hours.” These factual statements are straightforward and simple. ## Examples of Effective Headlines (Holiday Edition) Seeing is believing! These examples of hot headlines, straight from the Realize tool, demonstrate how an eye-catching headline can stop the scroll and encourage clicks. ### Cozy Must-Haves to Gift Yourself This Winter Why it works: Shoppers might be focused on buying for others, but that doesn’t mean they can’t grab a few things for themselves. “Cozy must-haves” paints a comforting picture of soft blankets, fuzzy socks, and warm beverages, while “gift yourself” tags onto the “treat yourself” trend. The message of prioritizing joy while shopping for others can resonate well just after Cyber Week, when the stress of searching for the right gifts leaves customers ready to seek some relaxation. ### Last Chance to Unwrap Holiday Savings Why it works: This headline delivers a one-two punch — both urgency and seasonal relevance. The phrase “last chance” evokes fear of missing out (FOMO), prompting shoppers to take action before the offer is gone. “Unwrap holiday savings” brings that emotional connection, reminding shoppers of the joyful experience of opening presents, especially when those presents come at a cost savings. This headline is perfect for later in the season, when shoppers are rushing to check off those last few gifts on their list. ### Best-Selling Deals That Ship by Christmas Eve Why it works: This headline hits on three fronts. First, there’s the social proof that comes with the term “best-selling.” It showcases items that are popular, serving as a shortcut for last-minute shoppers who simply don’t have time to research. As always, “deals” appeals to the bargain hunter. It’s the phrase “that ship by Christmas Eve” that seals the deal, though. It reassures time-crunched shoppers that their purchase will arrive when it needs to. It’s a great end-of-season marketing push that can earn you extra revenue. ### New Year, New Skin: Refresh With These Top Picks Why it works: Once all the presents are unwrapped, attention turns to another big milestone — the end of the year. Words like “refresh” capture the spirit of the “clean slate” theme that comes with the start of a new year. “New year, new skin” is a clever play on the “new year, new you” phrasing, making it clear what you’re offering. Lastly, the term “top picks” lets customers know that the products have been curated to make things easier for those shopping for new beauty products. This is a great way to extend your holiday sales beyond Christmas. If you want to make sure your headline hits the mark, it’s well worth trying the Headline Analyzer within Realize Trends. “The analyzer scores your headlines using CTR scores relative to others in your vertical,” Leshem explains. “It also highlights weak spots, such as low-impact words. Users should treat this as a creative tuning tool, making small headline changes to push performance before launching at scale.” ## Pairing Visuals and Headlines: A Holiday Must While a compelling headline gets results, pairing it with an eye-catching visual can amplify its effect. That’s where Image Trends can help. This AI-powered tool reveals the visuals showing the most engagement. “From past insights, we know that indoor imagery often outperforms outdoor shots during the holidays,” Leshem says. “Warmth, comfort, and authenticity win.” Top-performing images for the holidays include the following traits: - Warm indoor settings (e.g., kitchens and fireplaces). - Bright reds, golds, and greens. - Hands holding gifts, decorations, or people looking at the camera. - Real-life photos over polished studio shots. For best results, Leshem recommends combining Image Trends with headline insights. “Pull top-performing images from Image Trends, then match them with equally strong headline variants from Headline Trends and the Analyzer,” she says. ## When to Refresh Creatives Normally, trends may refresh every month or so, but that isn’t the case during the holiday season. Things move fast, and trends can change from one week to another. “Check Realize Trends weekly for shifts in tone, keywords, and imagery,” Leshem advises. “Refresh creatives every 10-15 days to remain aligned with trend data.” Leshem finds it best to start monitoring trends early, suggesting that you start in late September or early October for any Black Friday content, then use October to analyze holiday trends and begin early testing. From there, you should revisit trends weekly starting in early November and continuing until the holiday season is complete. “For New Year’s and January deals, shift your tone to ‘refresh,’ ‘declutter,’ and ‘start the year right,’” she adds. ## Key Takeaways During the holiday season, headlines matter more than ever. Tools like Realize Trends offer timely insights into user engagement, helping you adjust your approach to get immediate results. Using emotional triggers and seasonal language like “cozy,” “unwrap,” and “gift,” you can connect with busy, time-sensitive shoppers. You can also combine the Headline Analyzer with Image Trends to fine-tune your creatives before launch. ## Frequently Asked Questions (FAQs) ### How much time should you spend writing headlines? More time than you think! While headlines seem like a quick task, they’re the first impression you make on a new customer. In fact, one study found that 75% of those who share an article on social media don’t read beyond the headline before sharing. This can be especially true during the chaotic holiday season, when ad fatigue sets in and audiences are bombarded with promotions. Allocate extra time to test headline variations and optimize based on your results. ### How do you start writing a headline? Every headline starts by capturing the message in its corresponding content. You should also consider the emotion or action you want to inspire. Are you creating urgency in the hopes of getting clicks, or is your goal to build your brand by promoting comfort? Once you have your topic and goal in mind, consult the Headline Trends tool to look for keywords and terms that are currently resonating with readers in your industry and location. Using that information, come up with three to five options and test them using the Headline Analyzer. ### What is the format for headlines? Headline formats tend to vary by vertical and audience, but for SEO purposes, it’s best to keep your headlines to 60 characters or less. For social sharing, posts with shorter headlines (8-12 words) get more shares on X, while longer headlines (12-14 words) get more Facebook likes. Start with a strong verb or an attention-grabbing phrase, then clearly convey value or emotion. Also consider adding a number to your headline when relevant. Headlines like “5 Holiday Must-Haves” can capture attention and get clicks. --- ### Ideal Customer Profile: Know Exactly Who You’re Marketing To URL: https://www.taboola.com/marketing-hub/ideal-customer-profile/ Last Modified: 2025-07-24 10:07:06 Say you owned a company that exclusively made jackets for men with a 42-inch chest size and a 34-inch waist, who stood six-foot-two-inches tall and who had arms measuring from shoulder to fingertip exactly 37 inches. Now, for starters, many people would rightly advise you that you were probably a bit too specific in your product inventory. But, no one could say it would not be very easy for you to come up with what marketers call an “ideal customer profile.” For you, it would be a man of the exact dimensions laid out above. As it happens, an ideal customer profile can be quite useful for advertisers to use even when it’s not quite so specific. ## What Is an Ideal Customer Profile? An ideal customer profile, often referred to as ICP, is a detailed, well-reasoned, and researched description of a hypothetical customer who would best benefit from, and most appreciate, your products or services, and who would theoretically be the most valuable customer for your business. An ICP is a way to think about and focus your marketing efforts, seeking the most promising leads. It incorporates that old wisdom that you can’t be all things to all people; rather than trying to appeal to everyone, you try especially hard to appeal to your perfect customer. In so doing, you will also appeal to people who are adjacent to the ICP in many ways. ## Importance of an ICP in Your Marketing Strategy An ideal customer profile is highly important for a successful marketing strategy because it helps advertisers identify and then target their most valuable potential customers. Use of an ICP as you create marketing campaigns can lead to increased efficiency, better return on ad spend (ROAS), higher conversion rates, and ultimately, better business performance and less wasted time, effort, and money. By defining who your ideal customers would be, you can better focus your marketing efforts, personalize your messaging, and allocate resources more effectively. ## Key Benefits of Building an ICP Once you have a solid ideal customer profile created (or a few ICPs, if you offer a wider range of products or services), you can see how ICPs help in the creation of marketing materials and whole campaigns. They will assist with: ### Targeted Marketing When you have a real sense of who your best customer would be, you can go out and find them, meeting them where they are with your ads. Would your ideal customer frequent this or that social media platform, this or that type of YouTube channel, or this or that news site, for example? Then spend money to place your ads appropriately. Would your ideal customer be more likely to respond to a video ad or to open an email? Would they be more likely to appreciate humor or informative content? The more you know about them, the better you can reach them. ### Improved Products and Services The more you know about your ideal customer, the better you can make products, or the better services you can offer, that they would genuinely enjoy and use. It becomes almost a chicken-and-egg proposition: Are they your ideal customer because of what you already offer, or because of how you tweak things to meet them? The answer to that rhetorical question does not matter — what matters is that you continually refine not only your advertising, but also the products and services you're hoping to sell to your ideal customer. ### Better Customer Retention Every marketer knows that it’s easier — and cheaper! — to retain existing customers than it is to attract new ones. The better you know your customers, the better you can hold onto them by offering messaging and products they will truly appreciate. ## ICP vs Buyer Persona ICP Buyer Persona Type of party you seek. Specific party you target. Ways customer consumes media. Actual media consumed (TikTok, e.g.). Examples of habits or pastimes. Specifics (prefers Orange Theory gym, e.g.). Goals ideal customer may have. Actual work towards success. Potential customer pain points. Specific examples of barriers to conversion. ## How to Build an ICP ### Define What Would Make a Best Customer Before you worry about defining your best customer, think through what would make them so. Are you hoping for subscribers? Are you simply looking to drive sales? Are you looking for repeat buyers? Do you want to attract people who will in turn spread the word about your business? Rather than starting with who they are, think of what you want them to do. ### Gather the Data The more you know about your ideal customer, the better. You need to gather all sorts of data, from demographics like age and income levels, to the way they likely use social media, to their usual shopping habits with other brands and so on. You can never have too much data about your customers; even if you don't use some of it, it's good to have it. ### Engage in A/B Testing Even if you’re quite confident that you’ve successfully created an ideal customer profile, you need to allow that you might be a bit off in some areas. It's a good idea to create at least two ICPs, even if they’re relatively similar. Use each of them to engage in different marketing campaigns and see which one performs better. ## What Data and Metrics Should I Use to Build My ICP? The short answer is to use as much data as you can gather, and to use as many metrics as you can. A slightly longer answer is to be sure you use these three considerations: ### Demographics You need to know, as accurately as possible, about your ideal potential customer’s age, gender, location, occupation, and income level, education, and other such basic (and often findable) pieces of information. ### Psychographics This is a knowledge of a person’s values, personality traits, lifestyle, and interests. Do they seem particularly religious? Highly patriotic? Spiritual? And so on. ### Buying Habits The more you can learn about how your would-be customers already spend their money on products and services, the better you can meet them where they are with your marketing efforts. ## How to Use an ICP to Improve Lead Generation and Targeting ### Targeted Outreach An ICP can help you identify specific characteristics of your would-be ideal customers, allowing you to focus all lead-generation efforts on those who will be most likely to convert. ### Refined Content You know your audience, so make sure your marketing efforts will resonate with them. Tailor your ads, emails, and other marketing content to be designed for your people, not just for people in general. ### Website and App Optimization Your marketing efforts capture leads, but where they land is what matters when it comes to conversions. Make sure your site or your app (or both) will be a pleasure for your ideal customer to navigate so that they move through to conversion. ## What Are Common Mistakes to Avoid When Defining and Using an ICP? ### Not Using A/B Testing Even the best marketers in the business can sometimes misunderstand their potential ideal customers. As mentioned, then, it's a good idea to create at least two ICPs and to run marketing campaigns based off of each of them. You may find one is notably more successful than the other. If both are a success, consider a third ICP and ad campaign, as you might have a varied base of ideal customers — not running multiple campaigns may leave many of them off the table. ### Not Using Hard Data Your intuition and your understanding of your customer base can and should play a large role in the development of an ICP, but you need to also use hard data, letting information you collect about your customers inform the ICP. Remember, this is not who you would want your ideal customer to be — this is who your ideal customer actually is. ### Not Updating Your ICP People change. It happens to individuals and it happens generationally. You need to rethink and update your ICP now and then to make sure you’re still appealing to your actual potential customer base, not to someone who would have been a customer five years ago ## Key Takeaways An ideal customer profile (ICP) is a detailed description of a hypothetical customer who would benefit most from your product or service. Marketers create ICPs so that they can customize their advertising to people who would genuinely be likely to buy what they are trying to sell, if only they could be made aware of it. ICPs are created using market research about demographics, online behavior, past shopping behavior, and more. The more information a marketer can get, the more accurate the ICP will be. ICPs need to be reevaluated and updated now and then, with a yearly review likely sufficient. This helps ensure you still accurately understand the customer base. ## Frequently asked questions (FAQs) ### What role does an ICP play in account-based marketing (ABM) strategies? In account-based marketing strategies, the ICP serves as something of a blueprint for identifying and targeting high-potential accounts. It defines the characteristics of people that are most likely to become valuable repeat customers, allowing marketing teams to focus their efforts on these most promising leads. ### Which qualitative and quantitative factors are most important in defining an ICP? It gets back to those demographic considerations. You need to think about things like income level, location, education level, and other tangible factors, but you also need to consider your potential customers in a more holistic way. Think about their interests, how they spend their time, and so on. ### How do I tailor my messaging and content to resonate with my ideal customer? To tailor your messaging and content to truly resonate with ideal customers, start by deeply understanding who they are, what they need, and how they communicate. Then, create that ICP, segment your audience if you find you have several different ideal customers, and craft targeted content that addresses as many of their specific pain points and desires as possible. Over time, continuously analyze and refine your approach based on performance data, audience feedback, and continued research. --- ### Data-Driven Marketing: How to Leverage Data for Smarter Campaigns URL: https://www.taboola.com/marketing-hub/data-driven-marketing/ Last Modified: 2025-09-11 10:05:57 In today’s marketing landscape, data is no longer just a supporting player, it’s the star: For marketers, data-driven campaigns are essential to remaining competitive. But, what is data-driven marketing, exactly? Simply put, it’s the strategic use of data to shape, execute, and measure your efforts. From targeting to real-time ad bids, data can help you make smarter decisions and maximize return on your investment. Effective use of data starts with knowing how to collect it and how best to use it. This guide will take you through the process of leveraging data to get better results from your campaigns. ## What Is Data-Driven Marketing? At its core, data-driven marketing is simple — you gather information and put it to use in creating and launching marketing campaigns. Instead of running your campaigns on instinct, you have real, verifiable facts that can inform your strategies moving forward. The key components of data-driven marketing are: - Data collection: Marketing success starts with gathering the right data. Typically, that’s a combination of information on consumer behavior, demographics, and previous transactions. - Analysis: No group of data comes in fully formed. For best results, you need to thoroughly analyze the information you’ve collected, whether it’s who’s clicking on your ads, which products people buy, or when they abandon their cart. - Activation: Once you’ve reviewed the data, it’s time to put it to use. You can use your collected data while you’re coming up with new strategies, rejuvenating sagging campaigns, or personalizing your marketing to specific audience segments. - Measurement: The key to success in data-driven marketing comes from constantly measuring your results, then using the information to refine and optimize. ## What Are the Benefits of Data-Driven Marketing? Yes, data helps you understand how customers are interacting with your brand, but that isn’t the only benefit. When businesses embrace data, here are a few ways they excel. ### Improved Personalization Today’s customers aren’t just comfortable with personalization — they expect it. If you’re still blasting out generic marketing messages, you’re likely to fall behind. Data is the engine behind all that personalization. It empowers marketers to target messaging to individual consumers based on who they are, what they’ve done, and what they’re likely to do next. With the right data, you can display personalized product recommendations, dynamic landing pages, and customized offers. ### Better Decision-Making Data can eliminate the guesswork that has long plagued marketing departments. You don’t have to wait months to determine whether a particular campaign brought a return on investment: Instead, you can gather information and make adjustments before you’ve sunk months of effort into a campaign. ### Higher Efficiency Are you maximizing every dollar you spend? Data can help with that. Marketers are increasingly turning to AI-powered data for their ads for one reason: It can boost return on ad spend (ROAS) by 400%. Not only can data better inform your decisions on the creatives you use, but it can also optimize bids in real time to ensure you get the best results. Data can also help identify untapped potential in your marketing campaigns. Instead of continuing to pour money into oversaturated audiences, you can pinpoint new outreach opportunities. As a result, you can cut spend where it isn’t working and double down on your most effective efforts, increasing your efficiency. ## 4 Types of Data Utilized in Marketing Not all data is created equal. Successful data-driven marketers know how to grab the right kind of information from the right places. By combining the following data types, you can craft more relevant experiences. ### Demographic Different products draw different audiences. A line of anti-aging creams might be targeted toward women over 50, for instance, while a new sneaker line might be targeted to men ages 18 to 30. To narrow your targeting to the audience most likely to buy, you’ll need data like: - Age. - Gender. - Income level. - Education. - Marital status. - Location. Using demographics for targeting is nothing new, but today’s marketers face new challenges when it comes to gathering information on consumers, thanks to updated privacy laws. Even with those complications, though, demographics remain the starting point for broader campaign strategies. ### Behavioral How do customers interact with your brand? Monitoring that data can be one of the best ways to gain insights on your customers. Keeping an eye on metrics like website visits, page views, time on site, and abandoned carts can help you develop more informed campaigns. For example, if data suggests a high exit rate on a certain product page, you might test shorter copy or change your page layout. You can also use behavioral data to identify and target users who looked at a product but didn’t buy. Today’s tools let marketers gather behavioral data in real time and personalize interactions for better results. ### Transactional This type of data focuses on a customer’s past transactions with your brand. It looks at what a customer has bought, how often a customer purchases from you, and how much they spent. Data points include order history, purchase frequency, and average order value. This type of data is crucial when you’re tracking loyalty and retention rates, or when you’re interested in cross-selling or upselling existing customers. As an example, a coffee retailer might identify customers who order a fresh supply of coffee every 30 days. Using that information, the retailer can send a reminder as the 30-day mark approaches, or offer a discount for setting up a recurring purchase. Transactional data is also useful when trying to forecast demand, as you can understand which products are attracting the most interest and adjust your inventory accordingly. ### Engagement Engagement goes beyond your website and product pages, it tracks user interactions across platforms, with metrics including: - Emails opened. - Links clicked. - Social shares. - Video watch time. - Live chat interactions. This data helps you determine which types of content resonate most with your audience. High engagement signals that your content is performing well, encouraging you to use similar messaging on future campaigns. With social media interactions, you can monitor to see which types of posts resonate with your customers. Over time, the information creates a feedback loop that helps you generate better content strategies. ## 4 Technologies (and Tools) for Data-Driven Marketing Modern tools can pull everything together, including collecting data, analyzing it, executing on the information, and analyzing the results. Here are four of the most useful types of tools used by high-performance marketing teams. ### Tracking and Reporting Platforms The first step in data-driven marketing is gathering the data. For that, you’ll need a tracking and reporting platform. These tools help marketers monitor campaign performance, understand audience behavior and get the information they need to optimize their campaigns. While you’ll find no shortage of tracking and reporting platforms, here are a few of the top options: - Google Analytics 4 (GA4): A longtime leader in the market, Google Analytics stands out for its integration with Google Ads. GA4 brings cross-device and cross-platform journey insights, custom event tracking, and the ability to build predictive audiences based on purchase or churn probability. - Adobe Analytics: Another tool offering cross-platform tracking, Adobe Analytics is known for its deep data customization and robust visualizations. You’ll also get AI-powered predictive analytics to identify trends and forecast performance, making it a great choice for enterprise teams who want control over their reporting environment. - Mixpanel: For product-based analytics, Mixpanel is a powerful tool. It tracks event-based interactions in real time and supports cohort analysis to help you get a better feel for your audience. - Looker Studio: Formerly known as Google Data Studio, this free tool excels at bringing multiple reports together in one handy visual report. It’s ideal for marketers who need to regularly share reports with stakeholders across teams. ### Performance Marketing Platforms Once you have data in hand, the next step is putting it to use. Performance marketing platforms help with that, giving you the tools you need to achieve your goal, whether that goal is lead generation, increasing online purchases, or generating app installs. Here are a few leading performance marketing tools, along with information on what helps them stand out: - Realize: This set of tools is designed for the consideration and action stages of the funnel, bringing real-time data, AI-powered optimization, and a wide variety of high-visibility ad formats. These tools not only help with creation of different ad types, but you also get a built-in social importer and landing page builder to help make the most of your creative. - Skai: Formerly Kenshoo, Skai specializes in omnichannel optimization. It integrates more than 100 different publishers and retail media networks to help marketers unify bidding, budgeting, and targeting strategies, making it a great tool for businesses that manage large-scale media investments. - Marin Software: If you market heavily through Google and/or Meta, Marin Software can pull everything together to help you maximize your efforts. You’ll get budget forecasting, cross-channel attribution, and keyword-level optimization. - The Trade Desk: For help with your ad buys, The Trade Desk provides the tools necessary to run programmatic ad campaigns across multiple channels. You’ll get audience segmentation, real-time bidding, and metrics that help you optimize performance. ### Customer Relationship Management Systems Customer relationship management (CRM) systems are designed to collect and store customer data. These systems can be indispensable for leveraging first-party data. The right CRM stores a detailed interaction history for every customer, including email opens, sales calls, purchase history, and customer support tickets. Here’s how three of the top CRM platforms stand out: - Salesforce: One of the most popular and customizable CRMs on the market, Salesforce is known for its scalability and deep integrations. - HubSpot: Small businesses on a budget can’t beat HubSpot’s free tier, which lets you get started, then move to a premium plan as your business grows. - Zoho: Another cost-effective option is Zoho, which offers plenty of features starting at $14 a month. A CRM can act as the connective tissue between marketing and sales. Your marketing team can create customer lists, with the sales side of your business logging calls and deals. CRMs typically integrate with analytics platforms and marketing tools to give you a holistic approach to your data-driven marketing efforts. ### Marketing Automation Platforms Marketing automation platforms eliminate manual processes, letting you create behavior-triggered workflows that target customers at various points along their journey. Using marketing automation, you can send an email when someone abandons a cart or wish a returning customer a happy birthday based on CRM data. Here are a few marketing automation platforms, along with what sets them apart: - Marketo: This AI-powered tool from Adobe excels when it comes to lead scoring, multitouch attribution, and custom workflows. - ActiveCampaign: With a user-friendly interface, ActiveCampaign combines email marketing, CRM, and machine learning. Its predictive segmentation feature makes it easy to quickly act on customer data. - Klaviyo: E-commerce brands gravitate toward Klaviyo for its integration with popular platforms like Shopify and BigCommerce. You’ll get real-time tracking of customer behavior and a solid analytics dashboard. - Mailchimp: With a drag-and-drop automation builder that makes the process a breeze, Mailchimp can handle your A/B testing and integrate with your CRM. ## Creating Smarter Campaigns With Data The smartest campaigns don’t start with a concept. They start with data. Today’s marketing platforms provide easy access to both quantitative and qualitative insights, which means that you get both numbers and data (quantitative) and observations of user behavior (qualitative). This lets you build your campaigns on facts, not assumptions, but it isn’t just what you collect, it’s also what you do with that information once you have it. ### Using Data to Build Smarter Campaigns Data can help you at every phase of your campaign, from strategy to execution. Here’s a step-by-step guide to leveraging data to boost your return on investment. - Step 1: Define your goals and key performance indicators (KPIs). This is essential to setting up a campaign with measurable results. - Step 2: Gather quantitative data like click-through rate (CTR), conversion rate, bounce rate, and scroll depth. Make sure you measure this across various creative formats, including carousel, vertical, and display. - Step 3: Layer in qualitative data by analyzing behavior patterns. Look at the typical user journey and identify areas where users seem to drop off while also paying attention to which creatives get the most clicks. - Step 4: Use predictive intent signals, focusing on what users are researching or considering rather than just their demographics. - Step 5: Use a social importer or creative assistant to repurpose high-performing assets, using the data you’ve gathered to tweak the format and placement. - Step 6: Build landing pages that match user expectations. Tools with integrated page builders can help you test different landing page versions to find the one that best resonates with your target audience. - Step 7: Data can help you constantly test, analyze, and optimize your campaigns. Launch A/B tests, track results in real time, and utilize AI to optimize bidding, ad rotation, and budget allocation based on performance. ### Turning Insights Into Action Collecting data is only the first step. As you learn more about how users interact with your campaigns, you can use that information to target customers, personalize your messaging, and create assets that perform. Let’s say you see that your vertical video gets high engagement rates for first-time visitors, but carousel ads tend to perform better with return visitors. This information might prompt you to create separate campaigns where you target newcomers with short-form video, but use personalized carousels for those who’ve visited your site before. Then, you can create landing page copy to match the assets you’re displaying. ### Optimizing Campaigns Based on Data: Best Practices Effective data-driven campaigns aren’t “set it and forget it.” You’ll need to continuously monitor and optimize. Here are some tips to help: - Test one variable at a time: For best results, you should always isolate one item, such as a headline or an image, and test it. - Avoid testing too early: You’ll need a decent sample size to get an accurate picture of a campaign’s effectiveness. - Run tests across segments: What works for new website visitors might not work as well for repeat visitors and regular customers. Test segments separately. - Let AI optimize in real time: AI has empowered marketers to adjust everything from creative to bids on the fly. Use tools that make it easy to optimize your campaign to get the best results. ## Measuring Success in Data-Driven Marketing Data-driven marketing is only as good as your ability to track your efforts. It’s important to have tools in place to gather the right metrics on the most important KPIs. The top indicators to monitor include: - CTR: Measures how compelling your creative and messaging are. - Cost per acquisition (CPA): Tracks how efficiently you’re converting users into customers. - Conversion rate: Indicates how well your landing page and offers are performing. - Customer lifetime value: Projects the long-term financial impact of a campaign. - ROAS: Measures the overall financial effectiveness of your efforts. The best tools go beyond reporting numbers: They’ll help you see patterns, identify underperforming segments, and find opportunities for optimization. Here are a few top tools marketers are relying on in 2025: - GA4: As mentioned above, Google Analytics’ latest version can track behaviors across devices and events, giving marketers an even better understanding of user behavior across the entire customer journey. - Looker Studio: Formerly Google Data Studio, this tool can help you visualize your campaigns in real time. It allows you to combine data sources, including Google Ads, YouTube, and CRMs. - CRM-integrated reporting: If you use a customer relationship management (CRM) tool like HubSpot or Salesforce, you can connect your marketing efforts to your sales tracking, letting you see how your campaigns lead to closed deals and incoming revenue. ## Overcoming Challenges in Data-Driven Marketing While the benefits of data-driven marketing are clear, you can face a few hurdles during the process. With the right strategies, though, these challenges are easy to overcome. Here are some of the most common roadblocks marketers face, along with some practical solutions to help you overcome them. ### Data Silos Are all your teams on the same page when it comes to your data-driven marketing efforts? If not, you aren’t alone. Marketers often find their teams are disconnected, making it tough to combine everything. The same goes for your data collection efforts. If your ad platform doesn’t sync with your CRM, for instance, you’ll struggle to track your results. Sure, you may be able to manually upload the information, but this not only adds to your daily workload, it also inhibits your ability to scale in real time. Breaking down data silos should be a priority for organizations. Invest in centralized platforms that can combine your data and bring your teams together. If you have tools that don’t integrate natively, a middleware platform like Zapier or Segment can help bridge the gaps. ### Poor Data Quality Not all data is created equally. If you’re pulling in outdated or inaccurate information, you’ll likely see disappointing results. Often, when marketers find errors in segmentation, personalization, or targeting, it’s a byproduct of bad data inputs. In fact, Gartner estimates that bad data costs businesses $12.9 million every year on average. So what can you, as a marketer, do to keep your data collection efforts solid? In addition to vetting your data collection sources, you should regularly conduct audits on the information you’ve gathered. The right tools can automate the process, constantly looking for duplicates and outdated information. Marketers should also work with their IT teams to set up quality assurance processes at every point where data is collected. ### Talent Shortage Although data collection is nothing new, it has now infiltrated every phase of the marketing process. Many marketing teams face a skills gap when it comes to interpreting data, running analyses, and using advanced platforms. Unfortunately, this means that valuable data can go unused or, worse, be misinterpreted. While user-friendly platforms can help, in the end, marketers need to have at least one person on hand to ensure nothing falls through the cracks. For this, marketers can invest in internal upskilling, typically through online courses. Organizations can also bring in a data analytics specialist, either as a payrolled team member or a contractor. ### Privacy Regulations One of the most pressing challenges faced by today’s marketing teams is privacy. Not only are consumers concerned about it, but governments are getting involved. New regulations set restrictions on how organizations can collect, store, and use customer information. These rules can limit your access to third-party data and restrict your targeting capabilities. To remain proactive, marketers need to pay attention to the European Union’s General Data Protection Regulation and the California Consumer Privacy Act, as well as any new legislation that emerges. That said, many marketers have now made the move toward first-party data collection. This refers to data you’ve collected from your own website, apps, and emails, always getting user consent. You can also focus on intent-based targeting, relying on behavioral and contextual signals rather than data you’ve collected through third-party cookies. ## Case Studies and Success Stories Across all industries, organizations are leaning into data-driven marketing not just to inform their efforts, but also to improve outcomes. By leveraging advanced audience insights, AI-powered optimization, and performance-focused creatives, organizations have boosted their success rates. Below are three real-world examples that highlight how data is transforming the marketing landscape. ### Meitav Meitav, a top investment management company located in Israel, wanted to promote a new Bitcoin price index fund while also significantly scaling its performance marketing. The company succeeded at both, achieving 78x return on ad spend, which far surpassed its 20x goal. How did Meitav achieve such impressive results? It partnered with a performance-focused ad platform to launch a multi-format ad campaign that included motion, image, and video ads. Using a combination of predictive bidding tools and strategic optimization, the campaign showed the power of a focused, data-driven strategy. ### Chery Automotive brand Chery also implemented image, motion, and video ads, but in this case, their campaign sourced successful existing assets — most notably video content that featured electric vehicle driving tips. Video was used to boost engagement while static ads encouraged form completions. The campaign produced impressive results: Chery saw a 12% increase in conversion rates and a 35% reduction in CPA while also lowering its cost per click by 30%. The mix of video and static ads showed the power of combining creative formats. ### AIDA Cruises This leading German cruise line showed the power of using data at scale. The company launched 120 campaigns that combined image ads, powerful bidding tools, and audience segmentation. The campaigns incorporated first- and third-party audience data, dynamic retargeting, and ongoing peer benchmarking to optimize their strategy over time. All the effort paid off. AIDA achieved 5x average ROAS and 26% lower cost per opportunity compared to peer benchmarks. Overall, conversions increased by 82% while cutting CPO in half. The success was attributed to a combination of AI-powered optimization, intent-based targeting, and continuous testing, showing how a long-term, data-driven strategy can drive scale and efficiency at the same time. ## Key Takeaways Data-driven marketing empowers organizations to move beyond guesswork to create smarter, performance-focused campaigns rooted in real insights. By combining quantitative metrics like CTR and ROAS with qualitative behavioral signals, marketers can personalize content, optimize targeting, and improve efficiency throughout the funnel. But, it’s important to have both the tools and the personnel to ensure the accuracy and relevance of the data being generated, while also measuring results over time. ## Frequently Asked Questions (FAQs) ### How can data improve the personalization of marketing campaigns? Data allows marketers to tailor their efforts to specific audiences based on user behavior, preferences, and demographics. This improves relevance, which can boost engagement and increase conversions. ### What challenges might marketers face when adopting data-driven approaches, and how can they be resolved? As powerful as data can be, marketers find it brings some challenges. One of the biggest challenges relates to finding the right platforms and personnel to extract data and ensure it’s solid. Siloed systems and teams can also fragment data use, which can lead to ongoing issues. Marketers can solve those issues by finding the right platforms, hiring skilled professionals, and regularly auditing their data collection efforts. ### How often should marketing data be reviewed and strategies adjusted accordingly? The exact frequency depends on your resources and goals, but generally speaking, marketers should review their data at least once a week. If you’re going hard on a particular campaign or promotion, daily checks are recommended, particularly if you’re investing significant funds. It’s important to keep an eye on your campaigns to ensure they remain aligned with audience behavior and market trends. --- ### B2B Marketing Trends Digital Performance Advertising Should Know in 2026 URL: https://www.taboola.com/marketing-hub/b2b-marketing-trends/ Last Modified: 2026-02-24 12:30:40 Marketing teams in 2026 are faced with tons of opportunity, navigating a crowded field of jaded users and creating the right mix of tactics, across channels, to find high-intent prospects and convert them to buyers. The B2B customer journey matters for both the prospect and the brand: Users want personalization done well, easy access to the right information, and clear pricing, while marketers have to use budget wisely while engaging these users to capture leads and drive conversions. As AI’s potential becomes clearer and the technology proves its value, B2B marketers can find endless ways to save time and money with it. Marketing automation and data-driven decisions, paired with video, social media, SEO and AEO, and thoughtful content, can all help marketers succeed, even if budgets and resources stay the same. Here’s a quick look, by the numbers, at the state of B2B marketing and where it’s headed next: - 39% of B2B marketers say resource constraints — time, people, and budget — are a challenge. - B2B customers use an average of 10 interaction channels in their buying journey, up from five channels in 2016. - More than 50% of B2B survey respondents want a true omnichannel experience — interacting and buying while moving seamlessly across channels. ## Trend 1: AI Becomes Infused Throughout Marketing B2B marketers can access AI-powered applications for more and more uses in 2026, whether automating processes, doing account-based marketing (ABM), distributing content, or analyzing data. AI has made an impact on the generative front for copy, imagery, and video, and marketers now also have to tweak SEO strategies and implement more answer engine optimization (AEO) tactics as well. To do AI well — building agents, automating workflows, creating high-performing ads — marketing teams have to work across their organizations to ensure high-quality data, which is key to AI’s success. And marketers will have to learn how to create content that’s easily found by AI tools to show up in answers — the new discipline of AEO. Here’s what else to know: - 95% of B2B marketers say their organizations use AI-powered applications. - 74% of B2B marketing teams will use AI marketing analytics to gain competitive advantage this year. - Data quality issues cause 60% of AI projects to fail or underperform. - Nearly 30% of marketers reported decreased search traffic due to AI tools. - Nearly 24% of marketers are working to update their SEO strategy for generative AI in search. ## Trend 2: The Continued Rise of Account-Based Marketing (ABM) and Personalization at Scale Account-based marketing (ABM) and personalization at scale address the pressing needs of digital advertisers and marketers today — namely, user fatigue and intense competition for attention. ABM brings together marketing and sales teams’ efforts to create better customer experiences and drive more conversions and sales. This discipline identifies high-value accounts and tailors messaging and offers to those accounts specifically. It’s easier to measure the success of ABM campaigns, so teams can see exactly what ROI they drove. This efficiency can lead to bigger deal sizes, higher conversion rates, and shorter sales cycles. ABM tactics are suited for fragmented buyer journeys, as they can extend consistent messaging across multiple channels. With the inception of AI, ABM platforms can also now layer in deep data analysis to find patterns in customer intent. Digital marketers can get data-driven recommendations for personalizing content, tactics, and messaging to specific accounts within the ABM process. Predictive analytics tools may be embedded in other technologies, or teams may build them: Either way, B2B businesses need a single source of enterprise truth to offer better intelligence to sales and marketing teams, avoid repetition in contacting leads, and craft personalized messaging and campaigns. Here’s what else to know about these trends: - B2B companies using ABM report a 38% higher sales win rate and 91% larger deal sizes for 24% faster revenue growth. - 71% of email marketers who have embraced personalization said their return on campaigns was good or excellent. - Faster-growing companies are driving 40% more of their revenue from personalization versus slower-growing companies. - AI-powered ABM drives a 15-25% increase in marketing ROI. ## Trend 3: The Power of Content Marketing and Thought Leadership in B2B Seasoned web users have become quite savvy at navigating the multiple channels, offers, and companies vying for their attention, whether in B2B or B2C spaces. Content marketing is essential in building an always-on program that educates and builds trust with prospects and customers. This includes thought leadership, which can help brands stand out with a strong point of view, interesting or new data points, or deep education around a new or emerging technology or other topics. Prospects want to be engaged, and they’ll quickly navigate away from fluff or inauthentic content. Personalization also plays a role — a B2B company’s content can easily be segmented and tailored to different audiences, for example, and brands can stand out with fresh, engaging content, particularly for decision-maker and C-suite executive audiences. Generally, B2B brands these days have to tell interesting, SEO-informed stories to attract the attention of buyers. The range of formats and channels has expanded, and experimentation and testing can help digital marketers figure out which is best for their brand and audience, whether that’s LinkedIn articles, carousels, email newsletters, white papers and e-books, webinars, in-person events, case studies, or other options. Thought leadership has also expanded beyond print into video and events formats. B2B marketers working on content and thought leadership have to consider authenticity and voice in their work with the rise of generative AI, but also consider scale and speed to stay ahead. Here’s what else to know about content marketing and thought leadership for B2B this year: - 65% of buyers said a piece of thought leadership content changed their perception of the company for the better. - 30% of decision-makers rate most thought leadership as mediocre, poor, or very poor. - Just 43% of organizations consider their content strategy to be data-driven, though 80% say they rely on content analytics to decide what to create. - 40% of B2B marketers say one of their biggest challenges is creating content that prompts a desired action. ## Trend 4: Leveraging LinkedIn and Professional Networking for B2B Marketing With more than a billion users, LinkedIn has proved its staying power for B2B companies over the last 20 years. Its focus on professional networking, ad buying options, and options for company pages means it’s the primary social network for business-to-business marketing. LinkedIn has expanded its formats and channels for both paid and organic marketing efforts, so B2B digital marketers can use targeted paid ads and messages alongside organic content types (such as posts, articles, carousels, live events, groups, and polls) to engage prospects. LinkedIn users can engage directly with their colleagues and former coworkers, as well as follow particular companies and leaders. Executives can use LinkedIn to build their own name recognition alongside their company, and brands can build their reputation there, too. The specificity and engagement levels on LinkedIn are good for lead gen: Brands see a 2x to 3x lift in their brand attributes when they advertise on LinkedIn, and marketers see a conversion rate up to 2x higher there. Here’s what else to know about LinkedIn and professional networking trends: - 42% of marketers reported using LinkedIn as part of their marketing strategy in 2025, an 11% increase from 2024. - Four out of five LinkedIn members drive business decisions, and the audience has twice the buying power of the average web audience. - 89% of B2B marketers use LinkedIn for lead generation, and 62% say it produces leads. ## Trend 5: Incorporating Video Throughout B2B Marketing Strategies Social media platforms and a proliferation of self-serve video platforms mean that video marketing is more popular than ever for B2B brands. Nearly all marketers (96%) agree that videos have helped users understand their product or service better. That number means that just about every marketer working today is incorporating video into their marketing mix, whether it’s quick mobile videos or more professionally produced short films. The trend toward short, snappy video has also reached B2B marketing, where 30-second and 15-second videos are popular. For digital advertisers, engaging videos can help cut through creative fatigue on the right channels at the right time. LinkedIn, Facebook, and Instagram are all effective channels for B2B video, along with YouTube and TikTok as appropriate. Videos tailored to B2B audiences should show off a company’s expertise and explain or demonstrate their solutions and how they can help, educating as well as building trust and brand recognition or authority throughout the funnel, including thought leadership videos, product demos, customer quotes or testimonials, or the occasional fun video that picks up on social media trends. Here’s what else is rising to the surface when it comes to B2B video marketing: - More than 80% of B2B buyers rely on video during vendor evaluations. - 73% of consumers like to watch short-form videos to learn about a product or service. - 89% of people say they’ve been swayed to purchase a particular product by an explainer video. - 49% of marketers said that short-form video earned the most ROI last year, more than long-form or live-streaming videos. ## Trend 6: More Strategic Use of Marketing Automation and CRM Integration in B2B The tried-and-true CRM platform is still the foundation for many B2B companies in 2026. Taking full advantage of a CRM platform’s capabilities, alongside AI-enabled marketing automation, can save lots of time and resources, removing repetitive tasks for marketers and bringing rich data-driven insights. A digital B2B marketing team should also be focused now on integrating any disparate systems, including the CRM, marketing automation tools, and other platforms like the customer data platform (CDP). That single source of enterprise truth, with data available to all teams in the business, will play a big role in success. AI-enabled tech can help generate more qualified leads, then nurture them in a targeted way, with a buyer’s journey that tailors messaging and content appropriately. AI’s massive capacity for scale means you can capture rich data from buyer journeys to iterate and inform future campaigns. Consider these other numbers for using CRM, automation, and AI in 2026: - The CRM market is estimated to have grown by 12-14% CAGR in the past few years. - 98% of B2B marketers see automation as critical. - Businesses using B2B marketing automation generate an average return of $5.44 for every $1 invested and see ROI in under six months. ## Key Takeaways In 2026, B2B marketers are cutting through user fatigue and oversaturated digital channels to reach the right prospects where they are. AI capabilities will continue maturing as marketing teams explore new use cases. Other new and emerging technology, like hyper-personalization and predictive analytics, combines with foundational strategies like SEO, AEO, and thought leadership to help B2B digital marketers find and convert users. ## Frequently Asked Questions (FAQs) ### What are the most effective digital advertising channels for B2B in 2026? In 2026, B2B marketers have a lot of digital advertising channels to choose from. 73% of B2B marketers use social media advertising and promoted posts in their marketing strategy, while email marketing, SEO, AEO, paid search, video, webinars, open web, and influencer marketing also play a role in a digital marketing mix. ### How can B2B companies measure the ROI of their digital marketing efforts? B2B marketers can build a dashboard of relevant metrics, then continually track them and tweak the data gathered accordingly. Measuring the ROI of digital marketing efforts includes tracking lead generation rates, sales conversion rates, customer acquisition costs, and customer lifetime value, then using an ROI formula to understand the big picture. Integrated marketing technology tools and a single source of data will help build a data-driven culture. ### What are some common mistakes to avoid in B2B digital marketing? In B2B digital marketing, new trends and technology are emerging all the time. While digital marketers may make occasional missteps or overemphasize one channel or tactic over another, there are some common mistakes to avoid. These include neglecting SEO and AEO, which need regular attention and consistency for long-term success. Other common mistakes include overusing tech jargon in content, failing to include customer proof or connect with customers overall, using automation or generative AI poorly or too much, and not integrating channels and platforms, which can lead to wasted resources and duplicated efforts. ### How can B2B brands build trust and credibility online? B2B brands can build credibility by sharing useful information, educating prospects, and explaining how customers have found their product useful. This may take place through whitepapers, blog posts, webinars, events, and social media content, among others as appropriate. ### What are the key metrics for measuring success in B2B digital marketing? Key B2B digital marketing metrics are usually foundational metrics that all marketers understand. These include web data like traffic, bounce rate, engagement, conversions, and more, along with increases in leads generated, cost per lead, marketing qualified leads (MQLs), sales qualified leads (SQLs), and sales conversion rate. --- ### Anchor Text: How to Improve It for Better SEO URL: https://www.taboola.com/marketing-hub/anchor-text/ Last Modified: 2026-06-22 08:37:59 Marketing content doesn’t magically market itself. Instead, specific strategies such as optimizing anchor text on your site are vital, helping control the user experience and journey through content, and optimizing SEO. Here’s what to know about anchor text as you execute your content strategy plan, and ensure you’re getting the most from your content. ## What Is Anchor Text? Anchor text is the highlighted (sometimes underlined) word or phrase that includes the hyperlink that sends readers to another destination. It gives readers key information about where they’d be headed next if they click that hyperlink, and looks much better than including a long URL. ## Why Is Anchor Text Important in SEO? “Anchor text is important in SEO and digital marketing first and foremost because it provides a transparent understanding of where the user will land (i.e. what information will be delivered) if the item is clicked,” says Ilana Del Core, growth marketing SEO lead at Taboola. “Secondly, it helps search engines understand what web pages are about. Because of this, it’s critical to ensure relevance between the anchor text and the destination page.” ## What Does Anchor Text Look Like? Anchor text looks like a regular word or phrase seamlessly integrated into content. The only indicator to the reader that it is, in fact, a link, is that it’s a different color, often blue, e.g., Taboola Marketing Hub. However, behind the scenes, it looks a little different. As Del Core explains, anchor texts are made up of: - A hyperlink. - The visible text, which embeds the link. - The HTML (HyperText Markup Language) anchor tag and attribute. Example: <a href="https://www.taboola.com/marketing-hub/">Taboola Marketing Hub</a> “Creating a clickable link on a webpage is done using the <a> (anchor) tag in HTML,” she says. “The visible text that becomes the hyperlink is placed between the opening <a> tag and the closing </a> tag. The href attribute within the opening tag specifies the destination URL.” ## Types of Anchor Text Not all anchor text serves the same function. Here’s how to understand the differences, per Del Core: ### Exact Match Anchor text is ‘exact match’ if it includes the exact match of the keyword you’re targeting. For example, ‘performance marketing’ links to a guide on ‘performance marketing.’ ### Partial Match Partial match is anchor text that includes a variation of the keyword on the linked-to page. For example, ‘how to optimize your marketing budget’ links to a guide on ‘marketing budget.’ ### Branded Branded is the use of a brand name in the anchor text, for example, ‘Realize’ linking to its own product page. ### Generic Generic is the use of a generic keyword link, such as ‘click here,’ which should lead to the relevant page given the context of the sentence. Note that if you use an image, Google will use the text in the image ‘alt attribute’ as the anchor text. Here’s an example: <img alt="performance marketing metrics" class="XXX" src="https://www.taboola.com/XXX/". ### Naked ‘Naked’ means that the URL of the page the user will land on is used as the anchor itself. For example, ‘https://www.taboola.com/marketing-hub/’ is a naked link anchor. ### LSI This stands for latent semantic indexing, and is sometimes also referred to as ‘related.’ In this case, the keyword is closely related to your targeted keyword, for example, ‘performance marketing software’ or ‘performance marketing tools’ can be used as anchor text instead of the targeted keyword ‘performance marketing platforms.’ ### Long-tail By contrast, long-tail anchor texts are specific, multi-word phrases. Instead of using broad, generic keywords, they incorporate longer, more conversational queries that your target audience is interested in — for example, instead of ‘best performance marketing platforms,’ the long-tail anchor text would be ‘best performance marketing platform for small businesses in the USA with CRM integration.’ ## Internal Links and Anchor Texts Once users arrive at your website, you don’t want to send them to outside resources and materials. Instead, you want to keep them moving through your website. Internal links from anchor texts help guide users to other pages of your website. “Internal links are fundamental to website structure, user experience, and search engine optimization,” says Del Core. Internal links contribute to your overall website architecture in a few ways. You can link your pages via anchor texts, based on logical and intuitive structure or hierarchy to help users find what they need on your site. You can also link pages to help search engines understand and properly index your website’s pages. “Since Google uses web crawlers (also known as spiders or bots), which follow links from one page to another to detect, index, and rank the pages on the World-Wide Web, internal links help search engines find, index, and understand the pages of your website,” says Del Core. “Orphan pages, for example, are pages which are not linked from any other source — internal or external URLs or sitemap — and have close to no chance of being found and therefore indexed on search engine results pages (SERPs).” ## Anchor Text Best Practices ### Balance Internal and External Links Finding the right mix of internal and external links is a delicate balance. “Closely tied with site architecture and user experience, internal linking through anchor texts helps optimize your website to rank for the terms that contribute to your bottom line,” says Del Core. “Since the anchor text gives the search engine an indication of what information the landing page delivers, strategic internal linking can help search engines understand the topical configuration of your website, i.e. what your business specializes in. Together with high-quality content, this supports rankings for keywords which translate into qualified leads.” “Conversely, external links can affect the SEO rankings of your page, depending on the pagerank of the linking page and the anchor text it uses to link to your page,” Del Core adds. “Usually, this comes as an aggregated result of multiple pages linking to your page in this fashion. However, it’s possible that a high-quality, relevant external link will boost a page's ranking, and vice-versa.” ### Keep User Experience Top of Mind Anchor text also helps create a better user experience for those searching. “By clicking on relevant, clear, and concise anchor text, users know where they are navigating to and can get to relevant pages with ease,” says Del Core. To ensure good user experience, balance different types of anchor texts and don’t over-focus on certain types for the sake of SEO. Instead, make sure user-friendliness is the top priority of each decision. ### Diversify Types of Anchor Text “Best practices include having natural, concise, diverse, and visible anchor text that links to a landing page. Anchor text distribution includes a mix of all types of anchor text types,” says Del Core. A quick audit of recent content might show you that you’re leaning too heavily on one type of anchor text over others. ## How to Find Anchor Text Issues ### Study the Competition If your content isn’t performing as well as you want, study and mimic your competitors’ anchor distribution, at least for internal linking, Del Core recommends. ### Use a Technology Tool If you aren’t sure where to start, you can use a diagnostic or problem-solving tool to identify where your anchor text issues are impeding your SEO success. That said, a longer-term strategy for finding and fixing SEO problems, including anchor text, will be more successful. “Third-party tools like Ahrefs, Semrush, and Screaming Frog can run a crawl and let you export in bulk all of your internal links, including anchor text and landing page,” says Del Core. “It's important to see the extent to which your backlinks (external links) are healthy, as opposed to spammy, and ensure you have a healthy anchor text distribution.” ### Hire Help Using an SEO writer or editor to go through past copy and update anchor text can help improve your SEO. Ask them to find places where text is too generic, and locate broken and outdated links. Make sure you aren’t using too many generic anchor texts, like “click here.” You can also train writers and editors moving forward on the best practices of creating new anchor text. ## How to Fix Anchor Text Issues The same tools and experts that identify the issues can often fix them as well, though sometimes you need more expertise. “Third-party tools such as Semrush, Ahrefs, and Screaming Frog will crawl your site/subfolder/subdomain and report back all internal link issues,” says Del Core, but adds that a webmaster or SEO expert will need to manually correct them. “Sometimes, with the help of your web developer (using find-replace functions or scripts), certain common patterns can be fixed in bulk. However, from my experience, an SEO audit on a website that has been overlooked is a manual endeavor and requires careful and meticulous work to improve your performance on search engines.” ### Don’t Use the Same Keyword on Multiple Links Unnatural and spammy use of anchor texts and using the same keyword on multiple links will most likely devalue your links, says Del Core. If you continue to use bad practices, your site could even face a penalty. Manipulating backlinks and keyword-stuffing constitute anchor-text spam, and can hurt your SEO strategy in the long run. ### Skip the Black Hat SEO It can be tempting to try to play the system by using “black hat SEO,” in which you buy backlinks, keyword-stuff, use clickbait, or sometimes even present different versions of text to readers than to the algorithm. Recovering from a black hat SEO penalty isn’t easy, though. “These are important patterns to monitor and detect since they can cause penalties, devaluation of links, and poor user experience,” says Del Core, who recommends avoiding those strategies altogether. ### Build Genuine Authority and Trust In addition to the content itself, external links are still known to be the one of the most important ranking signals for pages. “You want your links to be valued for your rankings, and for the search engine and users to see you as a trusted expert and authority figure in your line of business, for optimal results,” says Del Core. Del Core adds that backlinks are important since they represent “endorsements” or “votes” as to how valuable your page is. “From a search engine's perspective, the more positive votes you have, the higher the ranking correlation,” she says. “From a user perspective, the more the traffic coming to your page sees content that’s relevant and high quality, the more likely these users can turn into qualified leads.” ### No-follow External Links Because no-follow external links instruct search engines not to crawl a web page, you may wonder what role they play, but they can be an effective way to link to another site without passing on any link authority. First, determine if you trust the destination: For example, bypass external links that are flagged because they’re a user comment or sponsored content. An SEO expert can make this decision one link at a time, or you can flag and change them in bulk with a tech tool. As you identify the above problems, follow a logical linking architecture. “Through the right anchor text, users can easily navigate to information they are seeking and know what to expect when they click on your anchor text,” says Del Core. “In addition, site architecture will distribute authority around your website via Google’s link analysis algorithm PageRank, which flows to pages and helps improve rankings on search engines.” ## Key Takeaways Anchor text is an important SEO practice used to build more visibility and trust, and to rank better on Google searches. There are multiple different types to explore, and with a little time and the right tools, you can identify any current issues with your anchor text, and resolve them for better SEO results. ## Frequently Asked Questions (FAQs) ### How do you check your backlink profile? To improve your own site’s SEO, it can help to check your backlink profile, which will tell you who else on the internet is linking to your site, what anchor text they are using to do so, and how high (or low) quality those links are. You can do this for free with Google Search Console using the “links” button, or you can use various checkers from the technology tools for SEO mentioned above, including Ahrefs, Semrush, and others. ### What is rich anchor text? Rich anchor text, as opposed to generic anchor text, is text that incorporates relevant, descriptive keywords that accurately reflect the destination page it links to. It’s important since it builds topical authority, indicating to the search engine who you are and what you are trusted for as a business; distributes PageRank in a more targeted way, promoting your link equity; provides context, which is crucial for users as it helps them navigate intuitively and ensures good user experience; and makes it easier for crawlers to navigate and understand the relationship between the different pages on your website, which helps indexing and, ultimately, rankings. ### What is anchor text spam? “Anchor text spam is the manipulative use of anchor text in backlinks and internal links to boost rankings for specific keywords,” says Del Core. “This happens through excessive or irrelevant use of what we used to call ‘money-making’ keywords — basically, exact match keywords which translate into money — keyword-stuffing, or high and repetitive amounts of the same generic keyword from external sites, usually spammy, low-quality sites.” --- ### Customer Acquisition: Funnel, Channels, Strategy URL: https://www.taboola.com/marketing-hub/customer-acquisition/ Last Modified: 2025-08-21 14:18:51 Getting a good grasp on customer acquisition is important, whether you’re running an established business or managing brand new marketing and advertising campaigns. In this article, I'll break down what customer acquisition is, the best channels to optimize the acquisition, how to measure its success, and best practices to increase conversion. ## What Is Customer Acquisition? In the simplest terms, customer acquisition is the process of getting new customers to buy your product or use your service. It includes everything involved in convincing prospects to make that leap from being interested onlookers to actual paying customers. It’s not just about making that initial sale, though — it’s about creating a relationship through top-of-funnel and bottom-of-funnel practices that can lead to those customers coming back again and again. ## What Is Customer Acquisition vs. Marketing? Marketing and customer acquisition are not the same, even if they’re integrated pretty closely. Marketing is the broader umbrella that includes all kinds of strategies to promote a business — think advertising, content creation, and social media buzz. Customer acquisition, on the other hand, is laser-focused on just gaining new customers. It’s a targeted mission within the marketing universe. You can think of marketing as the whole party, and customer acquisition as the dance floor where the magic happens. ## The Customer Acquisition Funnel Now, let’s talk about the customer acquisition funnel. This tool helps visualize and understand the journey potential customers go through, from first hearing about you to becoming a loyal fan. Here’s how it breaks down: ### Awareness This is where people first learn about your brand. They see your ads, read your blog posts, or hear about you through word-of-mouth. We call this stage "top-of-funnel" (or TOFU for short), as it's where you initially attract a wide range of potential customers to your business. This first phase of the customer journey has the primary objective of raising awareness, establishing trust, and generating leads. ### Interest Once your future customer has heard about you, they might get intrigued and look for more info, like reading articles on your website, watching review videos of your product on YouTube, or checking out your social media. ### Consideration At this point, customers are evaluating their options, comparing your offerings with competitors to find the best value. This is the middle-to-bottom-of-funnel stage, where your company aims to whittle down a wide pool of prospects into converting customers. In this stage, you might present case studies and testimonials to highlight real-life successes, offer product demonstrations and free trials for the customer to experience your product firsthand, provide detailed information, or present a limited-time discount to create urgency. Together, these elements effectively engage and convert leads heading toward the final stages of the buying journey. ### Conversion This is the moment! The conversion stage in customer acquisition is when potential leads turn into paying customers by taking specific actions, such as making a purchase, subscribing to a service, or booking an appointment. This phase emphasizes finalizing the sale and providing a seamless, positive transaction experience. ### Retention Post-purchase, the goal is to keep customers happy so they return for more and spread the word with others. This might involve good customer service, following up with your customer, and registering them in loyalty programs. ### Why Each Stage Matters Each part of the funnel is significant, and depending on where potential customers are in their decision-making process, you can adjust your strategies. For example, if they’re stuck in the interest phase, you might want to produce engaging content that answers their burning questions. By understanding the stages of this funnel, you can create targeted strategies to help effectively guide potential customers toward conversion. ## Digital Acquisition Channels Today's digital landscape has a smorgasbord of channels to help with customer acquisition, from social media and email to how-to vids. The most effective channels will depend on your target market, resources, and overall strategy. ### Social Media Places like Facebook, Instagram, YouTube, and TikTok are goldmines for connection. Create eye-catching posts, run ads, and collaborate with influencers to get the word out. Social media also creates brand awareness and deepens the connection with your current audience. ### Search Engine Marketing (SEM) Getting your business in front of potential customers when they search online is a biggie. Investing in pay-per-click (PPC) ads can put you at the top of search results, making it easier for customers to find you. ### Content Marketing Content marketing is all about creating and sharing helpful, relevant, and consistent content to grab the attention of a specific audience and keep them engaged. By providing informative and entertaining content, businesses can build trust and establish authority in their industry. Create valuable content that answers questions and provides solutions for your target audience. Think blogs, videos, or even how-to guides. It not only attracts attention but also builds trust in your brand. ### Organic Search/SEO The goal of search engine optimization (SEO) is to position your content at the top of the search results to grab attention. To effectively use organic search, you need to invest in SEO, correctly identifying and utilizing appropriate keywords and following on-page and off-page SEO best practices to help your content rank higher. Tools like Semrush and Ahrefs can help you optimize your content. ### Email Marketing Don’t underestimate the power of a well-crafted email. Building an email list makes it easier to convert leads into customers, and email marketing is still a highly effective way to engage with customers and promote valuable content, product updates, discounts, and events. Whether sending a birthday greeting or a special promotion, email creates a direct connection to their inbox, cutting through the noise of social media and search engines. Grow your list with an enticing freebie or discount, and send tailored messages to reconnect with prospects and nudge them toward conversion. ## How to Develop a Customer Acquisition Strategy ### Understand Your Target Audience The initial step is to define your ideal customer clearly. Consider their interests and the specific challenges they are seeking to address. Developing detailed buyer personas can significantly enhance the precision of your messaging. ### Select Appropriate Channels It’s essential to determine which digital channels are frequented by your target audience. Whether it’s Instagram, email, or TikTok, focusing on the right platforms will optimize your outreach and engagement efforts. ### Establish Clear Objectives Define what constitutes success for your organization. Your goals may include increasing website traffic by a specific percentage, or acquiring a predetermined number of new customers each month. Establishing measurable objectives will maintain your focus and direction. ### Engage in Continuous Testing and Refinement Ongoing testing is vital in developing an effective strategy. Experiment with various approaches and monitor performance closely. Studying your analytics will provide insights into which strategies are effective and which require modification, allowing for continuous improvement. ## How Do You Measure Customer Acquisition (Key Metrics)? Here are a few key metrics you’ll want to keep a close eye on: - Conversion rate: This tells you how many of your prospects actually turn into customers. Simple, but essential! - Customer acquisition cost (CAC): This is what you’re spending to acquire each new customer. - Return on investment (ROI): Look at how much revenue you're bringing in against what you’re spending on customer acquisition efforts. ## How to Calculate Customer Acquisition Cost (CAC) Calculating CAC is pretty straightforward: - Total Costs: Add up all your marketing and sales expenses over a given period (think advertising, salaries, software tools, etc.). - New Customers: Now divide those costs by the number of new customers you got in that same timeframe. For example, if you spent $10,000 and acquired 100 new customers, that means your CAC is $100. This helps you evaluate how efficiently you’re bringing in new business. ## Key Takeaways Customer acquisition is crucial for growing your business and gaining a competitive edge. By understanding the customer acquisition funnel, effective digital channels, and creating a solid strategy, you can successfully attract and retain customers. Don’t forget to measure your efforts and keep refining your approach for the best results. ## Frequently Asked Questions (FAQs) ### How does customer acquisition differ from customer retention? Customer acquisition is all about getting new customers, while customer retention is focused on keeping those existing customers happy and coming back for more. Both are important, but require different marketing strategies to meet the customer where they're at. ### What are the best customer acquisition strategies for startups? Startups can shine by using low-cost strategies like social media outreach, content marketing, and partnerships with other businesses. Building an email list early can also be a game-changer, so create an enticing freebie that requires email registration. ### Which channels drive the best customer acquisition results? Test out different channels to see what works best for you, as each industry has different acquisition channel strengths and weaknesses. Many businesses find success on social media, through SEO, and by using email marketing. ### What are the best tools for optimizing customer acquisition funnels? There are tons of helpful tools out there. Performance advertising platforms like Realize are especially helpful for optimizing the consideration and conversion stages of your customer acquisition journey, while Google Analytics will help you acquire and analyze the data from your web pages and content funnels, and Hootsuite can help you monitor your social media performance so you can adjust your strategies as needed. --- ### Top Fintech Marketing Trends to Watch in 2026 URL: https://www.taboola.com/marketing-hub/fintech-marketing-trends/ Last Modified: 2026-03-16 11:52:44 Fintech continues to grow more competitive in 2026, with fintech marketers navigating a more complex environment shaped by rising customer acquisition costs, stricter privacy expectations, and increasingly selective consumers. To stay in the game, fintech brands are rethinking their playbooks, moving away from broad, full-funnel approaches and focusing instead on relevance, trust, and measurable performance across every touchpoint. Below, I’ll run down the top five marketing trends shaping fintech marketing in 2026. What’s changed in our 2026 update: - All entries include updated and current information and advice. - All stats and figures updated with new and current information. - Information on how AI is impacting fintech marketing updated in FAQs. - New Fintech graph (benefits from AI for loyal consumers). ## Trend 1: Building Trust and Credibility in Fintech Marketing Fintech is more than software — you’re dealing with people’s money. That reality is even more pronounced in 2026, as consumers grow more cautious amid increased fraud, AI-driven scams, and heightened regulatory scrutiny. As a result, marketing today is all about building trust with the consumers interacting with your brand. Whether you’re offering advice or asking customers to provide payment details to use your platform, it’s important to lead with transparency, security, and complete respect for privacy. Here are some key statistics relating to consumer trust: - 74% of consumers say they trust their financial provider to protect them from fraud and security risks. - 84% of consumers expect proactive fraud alerts from their lenders. - Global money laundering is estimated to cost about $5.5 trillion per year, representing roughly 5% of worldwide gross domestic product. Clear communication is the first step. Financial topics can be complex for many, whether it’s cryptocurrency investments or finding the best annual percentage rate (APR) on a credit card. Simplifying those concepts with easy-to-read language can provide a solid foundation, while visual trust indicators like badges and certification logos can help reassure new customers that their information is safe with you. Many fintech marketers already incorporate testimonials and user reviews as social proof, but in 2026, brands are expanding beyond surface-level endorsements. By adding influencer quotes, expert endorsements, and success stories to your strategy, you can make your brand messaging even more persuasive. Remember, too, that compliance is always part of the conversation in a heavily regulated industry like fintech. It’s important that your marketing team bake compliance into your brand messaging to demonstrate that you’re a trustworthy brand. ## Trend 2: Personalization and Customer-Centricity in Fintech Marketing Generic marketing messages won’t suffice with today’s savvy customers, who have grown used to personalized messaging that speaks to their interests today, not a few months ago. Here are a few statistics demonstrating the power of personalization: - Recent data shows that consumers still highly value personalized experiences, but they’re increasingly selective about how their data is used. - 82% of customers are willing to share personal data in exchange for personalized experiences. - Willingness to share data declines with more sensitive data. Source : PWC With this in mind, how do you personalize your messaging for customers across all devices, platforms, and moments in the customer journey? Today’s successful personalization strategies include: - Segmented targeting based on factors like user behavior, financial goals, and demographics. For example, you’d likely target retirees differently than you’d target a recent college graduate. - Creatives that are unique to each channel. You might create unique images and messaging for email, app, social, and search. - Predictive analytics that suggest products to a user based on that user’s unique interests. - Behavioral retargeting that presents compelling display ads to users based on prior interactions with your brand, going beyond demographics and other traditional targeting criteria. Some marketers are shifting to innovative solutions that are built for performance, not just awareness. These tools can be the antidote to issues like soaring costs and creative fatigue, both of which are negatively impacting campaign effectiveness. Look for tools that offer cross-channel optimization, dynamic creative testing, and precision targeting. In short, fintech marketers in 2026 are focusing on building long-term relationships with customers, rather than chasing clicks. ## Trend 3: The Rise of Mobile-First and App-Based Fintech Marketing More than ever, consumers are picking up their mobile devices when they need to manage their finances. In fact, the global personal finance app market is expected to grow by 18.2% by 2033. Source: Market.us Here are some other notable stats related to mobile app use: - By 2027, global app payments are expected to top $1 trillion, driven by both consumer adoption and retailer integration. - In 2026, global mobile app downloads are expected to reach 299 billion. - 54% of bank customers say they use mobile apps on their phones for managing their bank accounts. Banking is only the beginning, though. From investing to paying rent, consumers are gravitating toward mobile. Not only does it make it easy to manage finances on the go, but for younger demographics, mobile is the primary gateway for everything from making purchases to researching products. In 2026, this shift has made mobile optimization a requirement. That said, a mobile-first approach means more than just responsive design: You’ll also need to optimize the app experience from start to finish, ensuring that each step of the customer journey is user-friendly. To satisfy consumer demand, innovative fintech marketers are currently investing in: - App store optimization: For fintech marketers, visibility means ensuring your app is optimized for all relevant app stores. Descriptions should also be optimized to ensure the app gets in front of the right users. - Push notifications: Once an app is installed, it can easily be forgotten. Push notifications can keep users engaged, but it’s essential to deliver timely, relevant messages without annoying your audience. - In-app messaging: Personalized nudges like, “Don’t forget to track your expenses” and, “You’re halfway to your savings goal” can provide the individual coaching many fintech customers seek. - Mobile-specific creatives: Are your images and videos optimized for mobile? Make sure you’re also creating interactive content that engages your customers. Once your mobile app is optimized, it’s time to get the word out about it. In this area, fintech marketers are looking at a variety of approaches. Mobile app SOCIOPAL, for instance, increased downloads by 30% after creating a few relevant blog posts for the app’s target audience, which included a strong call to action with a free download, then distributing them across premium publisher networks. ## Trend 4: Leveraging Content Marketing and Financial Literacy in Fintech The complicated nature of financial topics can serve as a roadblock for the average consumer. That’s why educational content will always convert well. In 2026, fintech marketers are focusing on content that builds confidence and reduces friction throughout the customer journey. The key is to not only educate, but boost a customer’s confidence in your brand. Here are some examples of standout content you can add to your marketing strategy in 2026: - Explainers on APRs, budgeting, and investment basics. - Interactive tools like retirement calculators or credit score simulators. - Search engine-optimized blog posts answering specific financial questions. - Video tutorials embedded within onboarding sequences. Thought leaders can also give your brand a boost. Chances are, members of your team have specialized expertise on various topics, so have them create blog posts or videos educating consumers on niche financial subjects. Consider the following: - Financial knowledge has remained consistently low over the past decade, with average correct scores on standardized financial literacy questions hovering around 50%. - Younger generations, particularly Gen Z, often score lower than older generations in objective financial literacy measures. - Most U.S. adults believe teaching financial literacy in schools could improve quality of life for younger generations. The key is to position your educational content as informational, not sales-y. This trust-first approach is especially crucial in 2026, as consumers grow more skeptical of overly promotional messaging. By doing that, consumers will see your brand as a partner, not a company trying to sell services. This approach attracts users who will become long-time brand loyalists rather than clicking, making a purchase, and moving on to another platform. ## Trend 5: The Strategic Use of Social Media and Influencer Marketing in Fintech Financial marketers always need to keep an eye on compliance when posting on social media, but while that might once have held marketers back, in 2026, social platforms remain a powerful discovery channel, though they’re rarely the final conversion point. Platforms like Instagram and YouTube continue to play a role in reaching today’s consumers. Here are some thought-provoking statistics regarding social media and influencer marketing in the finance sector: - Brands can earn about $5.78 for every dollar spent on influencer marketing, with top campaigns delivering much higher returns. - Influencer marketing can achieve up to 11x the return on investment (ROI) of banner ads and other static digital placements. - 61% of consumers trust influencer recommendations over traditional ads. - A majority of customers trust influencer recommendations more than a brand’s own social channels. Top social media approaches in 2026 include: - Short-form video content: Fintech marketers are using video to break down complicated financial topics like compound interest and investing. You can also create videos offering personal finance hacks such as paying off student loans or saving for retirement. - Real-time brand engagement: Finance routinely makes the news. Responding to the headlines or industry trends can help you reach users who are following those hashtags. - Authentic influencer partnerships: Collaborating with personal finance creators with solid followings can help get your brand in front of consumers. However, it’s important to ensure those partnerships are authentic, and that the creator uses the product and provides an honest review. - Data-backed performance tracking: In 2026, focus has shifted away from tracking impressions and clicks. Instead, fintech marketers are paying attention to app installs, conversions, and customer retention. Frustrated with diminishing returns from their social media efforts, most marketers are testing out other formats. For many marketers, this often means supplementing social with performance-focused channels across the open web. ## Key Takeaways The fintech space in 2026 is both fast-moving and competitive, but top marketers are rising to the challenge. Building trust and informing customers is still a top priority, but marketers are finding that personalization is necessary to grab attention. As consumers continue to prioritize mobile device use, apps are no longer nice-to-have products: They’ve become a crucial part of a fintech marketer’s strategies. Social media is still important, but content like short-form videos and influencer-led messaging dominates. ## Frequently Asked Questions (FAQs) ### What are the most effective digital advertising channels for fintech in 2026? Search and social are still staples of most fintech marketing strategies, but multitouch strategies lead the way in 2026. It’s important to follow consumers throughout their buying journey using tools like influencer partnerships and performance-focused ad networks that take advantage of the open web. ### How can fintech companies acquire and retain customers cost-effectively? It starts with winning over a new customer, and that takes precision targeting and mobile-first marketing campaigns. From there, educational onboarding can help build trust and boost customer retention. Fintech marketers who invest in lifecycle marketing can reduce churn, so it’s important to have marketing strategies for each stage of the customer journey. ### What are some successful examples of fintech digital marketing campaigns? AIG Israel was aware of the challenges it faced in educating users about mortgage insurance, so they leaned into video to help drive conversions. As a result, they saw a 50% uplift in purchase intent in the first 30 days. ### How is AI impacting the fintech marketing landscape? AI’s impact is being felt in almost every stage of marketing, from creative testing and dynamic ad placement to churn prediction and customer support. In 2026, fintech marketers are also prioritizing transparency in AI-driven decisions. ### What are the key metrics for measuring success in fintech digital marketing? Launching a marketing campaign is only the beginning. Fintech marketers need to keep an eye on results to better inform future efforts. Some key performance indicators to watch in 2026 are: - Cost per acquisition. - Customer lifetime value. - Conversion rate. - Cost per app install. - Engagement-to-conversion ratio. - Return on ad spend. - Loan approval rates. - Churn rate. --- ### Qualified Leads: Definition, Frameworks, Optimization Process URL: https://www.taboola.com/marketing-hub/qualified-lead/ Last Modified: 2026-03-19 13:24:11 While some leads turn out to be your ideal customer, not every lead is right for your product or service. To determine which leads have the most potential, you need to learn about qualifying leads. A qualified lead is one that is most likely to convert, demanding more attention and resources than other leads. ## What Is a Qualified Lead? Qualified leads are those individuals who are most likely to convert. The average cost of a lead varies by industry, but based on a recent survey from FirstPageSage, it can be anywhere from $91 (e-commerce) to $982 (higher education). Either way, when you multiply that by thousands of leads, it adds up fast, and it can be hard to know where to spend the money most effectively. Qualifying a lead ranks them by likelihood of conversion, letting you know where to focus your efforts. ## Difference Between a Lead and a Qualified Lead A lead is someone who has engaged with your product or service offer through a touchpoint, therefore showing some interest in your company. A qualified lead is one you’ve vetted further and determined is likely a true fit for what you’re offering. Think of it like a customer shopping for fruit at the grocery — they’ve not just looked at the apple, they’ve picked it up and inspected it to determine it’s ripe. ## Common Types of Qualified Leads ### Marketing Qualified Lead (MQL) - Top of funnel. - Aware of product but hasn’t taken the next step. - May have downloaded an e-book or attended a webinar. ### Sales Accepted Lead (SAL) - Still considered top of the funnel. - Has potential to be a qualified lead based on certain definitions, such as budget, location, need. - Sales team may be able to move the SAL to an SQL. ### Sales Qualified Lead (SQL) - Intent has been made. - May have asked for a demo or quote. - Further along in the sales funnel. ## What Is Lead Qualification? Lead qualification, also sometimes called lead scoring, is an analysis of a lead to determine how much potential they have to become a buyer, and also how well the product or service matches their needs. This process involves multiple considerations, including: - Does it match the need they think they have? - Does it match the need they actually have? - Is it within their budget (or, in B2B cases, do they have the power to make decisions over budget-related decisions on their team)? - Are they just considering their options, or very close to making a decision after comparing multiple options? By lead scoring, then focusing resources on those qualified leads, you will enjoy higher conversion rates and shorter sales cycles. ## The Importance of Lead Qualification Lead qualification is a critical aspect in determining where to allocate time and resources as part of a comprehensive marketing and sales strategy. Guessing which leads will convert is a waste of time and money, and will result in a more difficult and expensive process, with lower conversion rates and longer cycles. ### Loyal Customers If you vet potential customers thoroughly up front, you can identify those who may be long-term and returning customers. ### Shorter Sales Cycles Reap the benefits of shorter sales cycles with more targeted resource allocation as you get better at lead qualification. ### Organization You can target specific customers and use lead scoring to determine which leads are best for specific offers, further breaking down your lead scoring as needed. ## Lead Qualifying Frameworks ### GPCTPBA/C&I Goals, Plans, Challenges, Triggers, Barriers, and Actions/Context and Insights A marketer’s main goal is to truly understand customer behavior. This model has the customer journey and their choices along the way at heart, through identifying and understanding these key aspects. Goals: What are your leads’ main goals? What do they want to accomplish? What specific needs are they looking to fulfill with your product? Plans: What plans do your leads have to reach those goals? This might include a combination of hiring experts and purchasing tools or products, or metrics and data measurement over time to determine progress. Challenges: What issues, concerns, or obstacles might your lead face in achieving their goals, whether they predict having these issues, or if you’re the one suspecting they might arise? Triggers: What event prompted them to take action and investigate solutions? Barriers: What’s preventing them from simply buying now? Do they really have the funds and authority? Are there other hoops they need to jump through first? Actions: What are they doing along their customer journey? Are they downloading free resources, but hesitant to explore paid options? Context and Insights: This is a related framework that asks the marketing team to consider more deeply what the context is around the lead considering your product or service. What are their experiences in the past? How have their demographics, social interactions, background and training, and other aspects affected the decision? This leads to insights that can help you better tailor the customer’s journey. ### MEDDIC Metrics, Economic Buyer, Decision Criteria, Decision Process, Identify Pain, Champion Often seen as best for those selling to an enterprise business audience, MEDDIC uses data-driven decision making to help create strong deals. Metrics: What metrics can you share to demonstrate the financial or economical impact of the product or solution? Economic Buyer: Who is making the financial decisions at the lead’s company? Are you dealing with the decision maker directly? Decision Criteria: What metrics or rubrics are they using to determine if you and others are a fit? This can help you to position your own solution in an ideal light, demonstrating the necessary metrics to make the decision, phrased in a way that makes the most sense to their team. Decision Process: How long does decision making usually take? Identify Pain: What solution are you providing that solves their problem, and what happens if they don’t solve that problem? Champion: Who is a fan of your company, and will be a “champion” for your cause internally? Focus on that person. ### BANT Budget, Authority, Need, Timeline It’s a framework initially developed by IBM over 70 years ago, and for some, it still works best today. Others criticize its relevance and efficacy, pointing to newer and more helpful methods, but each marketing team needs to decide for themselves. Budget: Can they afford your product or service? Do they really have the budget for it? Authority: Is your contact the person who actually has decision-making power, or do they work with a team who does? Need: Do they need your product or service, or is it just an extra or “nice to have” that they’re considering? Time: What is their timeline, and how soon will they need what you have to offer? Are they shopping with intent to purchase soon, or on a longer timeline? ### CHAMP Challenges, Authority, Money, Prioritization If you’re really sick of hard selling tactics and want to get back to truly helping your leads meet their goals, you can consider CHAMP, a nickname for a four-step process in which hard selling takes a back burner to making sure the customers get the help they need. Challenges: What difficulty is the lead solving for? Authority: Who has the power to decide the solution? Money: What is the budget? Prioritization: How soon do they need to solve this problem, and how high/low priority is it in their business? ### FAINT Funds, Authority, Interest, Need, Timing Are you hoping to cultivate interest from very early stage buyers? This might be the most helpful framework for those who aren’t even shopping much yet, or are very early in the shopping process and just entering the funnel. Funds: Is there a budget in the near future that allows this shopper to consider engaging with your services? Authority: Does your contact have the ability and position within their company to make shopping and buying decisions? Interest: Has the potential lead engaged a little with the brand, and showing signs of interest in learning more? Need: Do they need the product or service you’re selling, and/or are you well-positioned to understand their need and communicate how you can help solve it? Timing: Will they be shopping at a leisurely pace, or are they working with an upcoming trigger event that makes it more pressing? ## A Step-by-Step Guide to Lead Qualification ### Revisit Your Ideal Customer Profile Align your lead qualification strategy with your existing buyer personas. These are detailed profiles of your best-fit customers: Think of them as a guide for spotting the kinds of leads most likely to convert. Use these profiles to set up a consistent scoring system that ranks prospects based on how well they align with your ideal audience. ### Build a Scoring Model That Reflects Real Value Once you know which customer attributes matter, assign point values to each. For instance, someone at a large company in your target audience who you’re positive has budget, purchasing authority, and leadership status over their project direction would hold much more weight than a junior marketing team member at a smaller company. Collaborate with your sales team to research and determine which data points should carry the most weight. Often, they’ve got firsthand knowledge of what makes a lead worth pursuing, and which aren’t worth as much of your time and money. ### Use Your Data You’ll gather data for lead scoring from a variety of sources: form submissions, site activity, email engagement, newsletter open rates, social media interactions, and more. These insights can help you identify not just who the lead is, but how they’re behaving. Maybe prospects who attend a certain webinar or download a particular guide are consistently closer to making a purchase. Next, set a threshold score that determines when a lead is ready to be handed off to sales, and keep adjusting it as you learn more. Tools like automated lead scoring software can help scale this process efficiently, making it easier for marketing and sales teams to focus on the leads most likely to convert. Lead generation software can help you use this data more efficiently and improve performance marketing at scale. ## How to Optimize the Lead Qualification Process ### Technology and Automation Use technology to automate as much as possible, to prevent wasting time manually qualifying leads. Use AI tools to educate the system on who your ideal customer is, then let it do the work going forward. Lead qualifying and generation software can also help match ideal customers to your preferred customer profile, and track how well they’re engaging with your content. You can also ensure your lead capturing strategies and initial interactions gather the correct information to quickly and accurately qualify leads through your tech tool. ### Get Ready for Meaningful Smarketing What happens when your sales and marketing team actually collaborate efficiently? Quite a bit! Smarketing involves the two departments working together, sharing data, helping each other with content creation, and making both jobs easier to ensure lead generation and qualification are going well. ### Access to First-Party Data First party data will clue you into real customer behavior, such as what they’re searching or downloading, allowing you to better customize the customer journey to their needs and actions. Lead qualification is much easier with this data, as you can teach your technology tools to watch for specific types of searches, downloads, and other high interest data to help push a lead over the qualifying threshold, and ensure they move to the next step of the funnel. ## Key Takeaways A qualified lead is one that has been vetted for their potential to convert, ensuring that they’re a great candidate for your company’s product or services. Creating and automating a lead scoring system and staying in frequent communication with the sales team can help marketing experts optimize this process. Streamlining the lead qualification process can lead to quicker converting leads and more loyal, long-term customers. ## Frequently Asked Questions (FAQs) ### What is the role of an MQL in the lead generation process? An MQL is a “marketing qualified lead,” meaning they've passed through your scoring system and have been qualified as an optimal lead with high potential of converting. The role of MQLs in lead generation is to help sales and marketing teams identify who to focus on, rather than spreading resources too thin, such as on unqualified leads that have a lower chance of converting. ### How do marketing teams hand off MQLs to sales to become SQLs? Once a qualified lead is ready to purchase, they hand the lead off to the sales team, and they become a Sales Qualified Lead, or SQL. Once they’ve accepted them into their section of the CRM or other marketing platform, the rest of the sales process can move forward to complete the deal. ### What is lead scoring? Lead scoring is assigning a numerical value to a lead, based on multiple features and data points that communicate how high of a priority they should be. Marketing companies can score leads based on the size of a company, industry, budget, timeline, and other factors, and set a threshold for what would officially qualify a lead. ### Intent vs. identity-based targeting: How does it affect qualification of leads? Intent-based means a lead is qualified based on their intent to buy, including them signalling they are close to being ready, e.g., they might have downloaded a price sheet or set up a call. Identity-based targeting is lead generation and qualification based on who someone is, such as their position in a prominent company that would make them an ideal client, rather than based on their behavior. --- ### Featured Snippets: Boost SEO and Win Position Zero URL: https://www.taboola.com/marketing-hub/featured-snippet/ Last Modified: 2026-06-22 08:37:40 When you type a question into Google and get a brief answer at the top, you’ve encountered Google’s “featured snippet.” This content block aims to answer a user’s question quickly and efficiently, pulling information from a webpage. How do you ensure Google includes your information in its featured snippet, though? This guide will help you sort out how featured snippets work and how you can take advantage of the visibility they bring. ## What Is a Featured Snippet? Also known as “position zero” to advertisers, featured snippets are the ideal place to be in search. Google extracts them from a website and features them prominently at the top of a search results page. It can be a block of text, list, table, or even a video, depending on the search query it’s answering. “I like to think of featured snippets as premium real estate on the search engine results page (SERP),” says Brianna Strouse, SEO manager at LevLane Advertising. “These SERP elements are a hot commodity in the search engine landscape, and acquiring them should definitely be a part of your SEO strategy.” ## Importance for SEO A featured snippet can take your visibility to the next level, especially for mobile and voice search, where users may only get the top result. That visibility can bring the following benefits: - Increased click-through rate (CTR): Featured snippets can quickly answer a question, but they often entice users to click over for more. Even if your website appears in the first organic position, the snippet can further increase visibility. - Improved authority: If Google selects your content for a featured snippet, that means the algorithms see your content as authoritative and relevant, and that can be good for your overall search performance. - Voice search dominance: Smart devices like Google Assistant will typically read the featured snippet in response to spoken search queries. This puts you ahead of competitors in an environment that’s increasingly shifting toward voice. ## How to Optimize for Featured Snippets ### 1. Target Question-Based Keywords Instead of optimizing for keyphrases, move toward a strategy that answers questions. Look for popular, relevant search queries that begin with “what,” “how,” “why” or “who.” Innovative tools that use artificial intelligence (AI) to predict search queries can help you come up with the right type of content. These tools are built on extensive engagement data and can pinpoint the exact phrasing you need. ### 2. Use Clear and Concise Answers Google’s featured snippets typically range from 40 to 60 words, which highlights the importance of being concise. Make sure your content answers the question quickly but thoroughly so that the algorithms will spot it and extract it. ### 3. Apply Structured Headers Big blocks of texts don’t appeal to the algorithms. Use H2s and H3s to break content into logical segments. The algorithms scan content for this type of structure. As a bonus, this hierarchy also makes it easier to, say, repurpose your content into carousel ads. ### 4. Create Lists and Tables Google’s algorithms tend to favor how-to lists and product comparisons when choosing content for snippets. To make it clear that’s what you’re offering, use bullet points and tables in your content rather than relying solely on paragraphs. ## Types of Featured Snippets and How to Win Them ### Paragraph Snippets These snippets answer queries using a few concise sentences. To boost your odds of winning the featured snippet, make the subheading a question, then follow it with a short answer. ### List Snippets If your content is a listicle (i.e. “5 Simple Ways to Clean Your Washer”), it’s important to use bullet points or numbers under a clearly labeled header. These snippets tend to come from step-by-step guides or rankings. ### Table Snippets Consumers often compare multiple products, and table snippets can be a quick way to get that information. Tables give users a quick way to compare prices, features, and statistics, making them ideal for featured snippet status. ### Video Snippets Since Google owns YouTube, it’s no surprise Google often displays YouTube videos answering a query. The best way to shoot for a featured snippet with your video is to optimize the video title, description, and timestamps. Tools with social import functionality can help you repurpose your assets into snippet-ready content. ## Tools for Featured Snippet Optimization ### 1. Semrush Semrush has a featured snippet option inside its Position Tracking report that lets you track where your content is appearing. You can easily view which target keywords have featured snippets and see opportunities for future featured snippets. ### 2. Ahrefs Ahrefs offers a feature called Site Explorer that helps you identify which areas of your website have featured snippet potential. It looks at the formatting of the content, as well as its current ranking position. ### 3. Google Search Console You can use the Google Search Console to determine whether your content is already appearing as a featured snippet. The performance tab will also show keyword data and offer click-through insights. If you’re seeing high impressions but low click-through rates on a page, that page might be a candidate for a featured snippet. ### 4. Predictive Audience Tools Platforms with deep learning technology using historical performance data can accurately predict the content most likely to win snippets. Thanks to AI, these tools can be great for streamlining everything from content creation to landing page builds, making it easier for you to compete for featured snippet positioning. ## Key Takeaways Featured snippets appear at the top of a search results page, providing quick, direct answers to question-based search queries. Optimizing for featured snippets can bring more customers to your website, especially in an era where users are shifting toward voice and mobile search. Advanced tools that use AI technology can help predict, format, and deploy snippet-winning content. ## Frequently Asked Questions (FAQs) ### How do you get a featured snippet on Google? While you can’t control whether your content appears as a featured snippet, you can boost your chances by creating content that answers relevant search queries in a concise, structured format. “Google typically scans the first few ranking articles as its source for a featured snippet, so you'll need a webpage that is ranking at least in the top 10 (ideally top 5),” says CJ Miller, CEO at Dypto Crypto. “From there, snippets are selected based on their quality and how well they answer the specific query. Google wants something succinct and to the point, usually referencing the question within the answer itself.” ### What is the ideal length for a featured snippet? If your goal is to earn a featured snippet, aim for a word count of between 40 and 60 words. But, length isn’t as important as structure. “The ideal length is roughly 50 words for paragraph snippets, and for list-based snippets, aim for four to seven concise bullet points so that Google can display the entire list without truncation,” says Linda Orr, fractional CMO at Orr Consulting. ### Can you lose a featured snippet? The short answer to this question is yes. Google’s algorithms are constantly hard at work, ensuring the most useful content appears in that section. “If another page presents a more relevant, up-to-date, and well-organized response, it's possible to lose one's earned featured snippet,” says Brandon Schroth, founder at Reporter Outreach. “Google's algorithms perpetually evaluate the content used for snippets. This is why it's crucial to monitor featured snippets for highly relevant keywords and phrases and update your content accordingly, to improve your chances of your content being picked.” ### Why are featured snippets important for voice search? In an era of smartphones and smart speakers, consumers increasingly use voice assistants to get answers to everyday questions. Voice assistants often pull directly from featured snippets, giving featured brands a voice-search advantage. “Featured snippets are incredibly important for voice search because of how people talk to their phones and smart speakers,” says Alex Smith, manager and co-owner of Render3DQuick.com. “In voice search, there is no second-place finish. Either you are the snippet, or you are invisible. Getting into that position puts your content at the center of the conversation people are having with their devices, and that naturally pulls more visitors to your site.” ### What type of content is most likely to appear in a featured snippet? Concise, informative content that directly answers a question is a prime candidate for a featured snippet. The following content is most likely to qualify: - How-to guides. - Definitions. - Comparisons. - FAQs. “I always advise businesses, ‘Think about what a person actually wants to know, then tell them that without any unnecessary fluff,’” says Abigail Wright, marketing specialist and senior business advisor at ChamberofCommerce.org. ### How do you format lists and tables for featured snippets? To format lists and tables, you’ll need to know a little HTML. Google responds to well-structured posts, so make sure you use relevant headers to improve both crawlability and readability. "Use clean headers, short bullet points, and properly labeled tables,” says Jensen Savage, CEO of Savage Growth Partners. “Google loves structure it can easily understand." ### What is the "People Also Ask" box and how does it relate to featured snippets? You’ve probably noticed the “people also ask” box hanging out among search results. Content that ranks in people also ask (PAA) can also show up as featured snippets. That further expands the visibility you’ll get by formatting your content for featured snippet eligibility. “A tip to rank for this type of content is to have an optimized FAQ on your site with FAQ schema markup,” Strouse says. ### Do featured snippets increase website traffic? Featured snippets can drive traffic toward your website, particularly if they relate to high-volume keywords. Yes, they can answer the question directly, causing some users to never click over to your site, but more often than not, consumers will click over for a deeper dive. “For some questions, the snippet gives users everything they need, and they might not click,” says Rodrigo Cesar, CEO and founder at SSInvent. “But, for more complex answers, winning a featured snippet can dramatically boost click-through rates and drive highly qualified traffic to your site.” ### How do you track your featured snippet rankings? You’ll find a variety of tools for tracking featured snippet rankings. Some of the most popular are Semrush, Ahrefs, and Moz. If your chosen marketing platform allows, set alerts to monitor for changes. “I use a combination of tools to track featured snippets,” Cesar says. “Tools like Semrush, Ahrefs, and Moz can show whether you're ranking for a snippet. I also manually check critical keywords by searching them incognito to see how my pages appear visually on the results page.” ### What are some tools for featured snippet research? In addition to SEMrush, Ahrefs, and Moz, look for tools that aggregate user intent data, automate headline testing, and enable quick content deployment. Bonus points if you can find solutions that periodically refresh your creative and provide predictive targeting capabilities. --- ### Holiday Shopping Trends 2025: What Our Experts Predict for the Peak Season URL: https://www.taboola.com/marketing-hub/holiday-shopping-trends/ Last Modified: 2025-08-11 06:43:13 The holiday season is always the biggest moment for brands to capture demand. This means that in 2025, digital advertisers will be under a lot of pressure to deliver results in a marketing landscape that has become increasingly complex. Consumer expectations are evolving, and competition continues to increase along with advertising costs. To help you stay ahead during the peak shopping season, here, industry experts at Taboola have shared their informed predictions and strategies for this year’s holiday shopping trends. ## Follow These 10 Holiday Season Trends by Our Taboola Experts ### 1. Keep Testing Your Creative This holiday season, advertisers should resist the temptation to rely solely on what’s worked before. As Katia Gelfman, senior growth marketing manager at Taboola, explains, “A new year is a great opportunity to challenge your creative tactics, whether it's trying a new format that you didn't try before, or thinking about new ways to have a better understanding of your creative's results.” Successful campaigns are the result of constant iteration and a willingness to try new things. You can uncover winning combinations by testing new formats, such as user-generated content (UGC), motion graphics, or AI-generated visuals. That said, the creative testing process should go beyond just changing visuals. As Gelfman notes, “You can try to find the strength of your creative — is it a high CTR? High CVR? Then, think about how you combine these two strengths together, and iterate again. If a woman creative worked for you, try a man. If a specific hook worked for you, try it with a different angle. Remember, your customers are relying on your advertisements when they’re considering whether to buy your product/service or not, so remember to be innovative, and don't just stick with the ones that are already working. Maybe there's another creative iteration that could have worked even better!” In short, continuous experimentation could be the secret to discovering your next top performing ad. ### 2. Maintain Precise Tracking With increased competition and the shift toward AI-powered performance marketing, precise tracking of your campaigns has never been more critical, especially during the peak holiday season. “As the holiday shopping season approaches, brands face increased pressure to maximize performance across every touchpoint,” says Shani Zagron, an expert campaign manager at Taboola. “With Google’s shift toward first-party data and similar changes across platforms, precise and privacy-compliant tracking is now essential. Strong data infrastructure enables real-time optimization, accurate attribution, and smarter budget allocation — all critical for capturing demand during the most competitive retail period of the year.” As we head toward holiday season, then, now is the time to ensure your tools and processes are set up to provide you with a clear view of the factors driving results. This way, you can adjust your ad spend dynamically as consumer behavior shifts throughout the season. ### 3. Navigate Higher Digital Marketing Costs Not surprisingly, the holiday season brings increased competition, but there are ways to keep your costs manageable. Taboola’s demand generation director, Rima Sherman, says that, “Marketers need to think outside the box when it comes to campaign timing — success during the holiday season isn’t just about showing up, it’s about showing up early and strategically.” By ramping up campaigns in Q3, Sherman suggests, marketers can establish their brand voice, refine their messaging, and stay top-of-mind for consumers before peak shopping season begins. “This early momentum reduces acquisition costs and ensures brand familiarity when shoppers are ready to act. You can then also double down in Q5 to capture last-minute buyers and extend the season’s impact beyond the traditional peak.” Rather than relying on traditional channels, Sherman adds, try to think strategically about when and where you spend. “Diversifying channels and identifying new growth engines — whether it’s emerging platforms, new placements within existing ones, or untapped audience segments — can give your brand a competitive edge,” she says. “Early action, creative timing, and smart platform choices are key to winning the season.” ### 4. Consider Your Analytics Holiday marketing campaigns can’t succeed without clear, actionable insights at every stage of the funnel. As Yoav Shaham, data and measurement team lead at Taboola, puts it, “When it comes to reporting and analyzing your holiday campaign performance, the conventional wisdom is that if it's not measurable, it's not worth doing.” But, he insists, this doesn’t mean you should focus solely on bottom-of-the-funnel metrics, which are traditionally easier to track and more directly tied to business results. “Ideally, you should determine your most important KPIs across the entire funnel, top to bottom, and make sure to set the right goals for every phase.” By defining and measuring KPIs across the entire customer journey, from awareness right through to purchase, marketers can aim for sustainable growth, rather than just achieving quick wins. This broader perspective is critical for optimizing your campaigns throughout the funnel. ### 5. Be Flexible With Your Campaign Budget The days of fixed budgets per channel are over, especially during the holiday rush. “In 2025, the changing landscape of online marketing demands flexibility more than ever,” says Taboola’s growth marketing director, Eyal Shnaides. “It’s no longer just Google and Meta: Most advertisers now run campaigns on at least two or more additional platforms, like retail media, TikTok, programmatic, or content discovery platforms.” It’s a shift, Shnaides adds, that has created more competition and made both performance and cost less predictable. “The brands that succeed will be the ones that stay agile, constantly reallocating budgets based on real-time signals like performance trends, creative fatigue, and shifting demand. Fixed budgets by channel are becoming outdated. Flexibility is not just important — it’s essential for maximizing results during the peak season.” If you can adopt this flexible mindset, you’ll be much better positioned to make the most of the holiday season. ### 6. Focus On Content Personalization During the holidays, consumers are bombarded with an overwhelming number of ads, promotions, and competing offers, so generic messaging doesn’t cut it. You need to produce creatives that are not only relevant, but also highly personalized. “In 2025, successful brands will be ones that deliver custom experiences at every touchpoint,” says Maayan Leshem, director of creative shop and AI strategies at Taboola. “With AI-powered personalization, you can go way beyond adding someone’s first name to an email: You can serve content that feels custom-made for each shopper’s interests and behaviors.” You could, for example, use browsing and purchase history data to create gift guides tailored to each user. Let’s say a shopper recently browsed hiking gear and outdoor apparel: You might highlight a “Top Gifts for Adventurers” collection, featuring best-selling camping equipment, winter jackets, or hiking boots. These could be delivered in an ad or an email at the ideal moment. By adding urgency triggers, such as “limited stock” or “final day for free shipping,” you can also make the offer feel personal and timely. The goal is to make every interaction, whether it’s an ad, email, or a landing page, feel like it was made just for them. If you do it right, you’ll boost engagement and drive higher conversion rates throughout the peak shopping season. ### 7. Use Authentic-Feeling Design Great holiday ad creative needs to feel festive, authentic and engaging. “During the holiday season, people crave ads that feel real and timely, not overly staged or generic,” says Taboola creative strategist Tyler Sorensen. “Ads should display energy, with movement, organic footage, and a clear seasonal vibe, as it grabs attention and blends naturally into the content your audience is already consuming.” The numbers prove it: Holiday-themed creatives deliver a conversion rate 4.4x that of generic ones. This might include showing your product placed in a warm holiday setting, with a Christmas tree or flickering candles in the background. Perhaps it’s a gift basket displayed next to a cup of hot cocoa. You could even share a quick “holiday hack” using your product, e.g., you could film a 10-second clip of someone wrapping a scarf as a gift. ### 8. Prepare Omnichannel Strategies Shoppers expect to switch seamlessly between channels, so brands need to deliver a smooth, cohesive experience. Shaked Afek Lifshitz, senior marketing automation expert at Taboola, recommends “focusing on a seamless omnichannel strategy to stay ahead, starting by using AI-driven personalization to tailor content and offers across all customer touchpoints.” She adds that you have to ensure that the customer’s experience is smooth from discovery to conversion, whichever path it takes. Lifshitz also recommends incorporating live commerce experiences, such as livestream product demos, and following up with smart retargeting and personalized emails to keep audiences engaged throughout their journey. As the season heats up, Lifshitz advises setting up proactive tools like “live chat, AI bots, and self-service options” to minimize friction, since these tools can prevent bottlenecks, helping to maintain that smooth customer experience. ### 9. Practice Good Data Hygiene More holiday traffic means more potential revenue, but it can also have a down side, warns Taboola marketing automation expert, Igor Shwarts. “During the holiday shopping season, the surge in traffic might be a double-edged sword: While it brings more potential customers, it also attracts a flood of fake sign-ups, bots, and low-quality leads,” he says. “Left unchecked, these distort campaign performance, inflate costs, and send optimization models chasing the wrong signals.” To combat this, marketers need to focus on smart segmentation and real-time lead enrichment. “By detecting and filtering out gibberish inputs, disposable emails, and mismatched intent at the point of entry, advertisers can ensure their campaigns focus only on high-quality, purchase-ready users,” says Shwarts. “The result? Better ROAS, cleaner signals, and more effective algorithmic optimization.” In other words, don’t just go broad — go smarter this holiday season. Shwarts truly believes that “brands that prioritize data hygiene will be the ones converting intent into revenue.” ### 10. Let Data Guide Your Holiday Engagement Plan While you want to gear up for the holiday season, don’t forget about the testing you’ve been doing all year long. Per Einat Wishnevski, user engagement team lead at Taboola, “This is the time to leverage everything you've tested throughout the year and put those insights into action, to maximize impact during the busiest and most competitive period.” Wishnevski says that to do this effectively, you must answer the four core questions of engagement: - Who are your customers — new, existing, or casual? - When is the optimal time to reach them — early for consideration, or later, with FOMO-driven urgency? - Where do they prefer to be contacted — email, SMS, or in-app? - What is the right message? As Wishnevski puts it, this last question is “the ultimate variable. The message cannot be static: It must be personalized, evolve as the holidays approach, and be relentlessly validated through A/B testing, allowing you to analyze and optimize on the fly.” It’s essential to engage customers early, creating a sense of urgency closer to key shopping dates. Use your audience’s preferred channels and, as Wishnevski advises, never stop testing your messaging — always refine to keep engagement high throughout the season. Remember, a flexible, data-driven engagement plan is crucial if you want to convert customer interest into dollars and cents. ## Key Takeaways This holiday season, the most successful brands will be the ones that are flexible, data-driven, and willing to innovate. From creative testing and tracking to budget flexibility and omnichannel engagement, the strategies highlighted by our industry experts demonstrate the importance of adapting to the evolving landscape. By embracing these expert predictions, you can benefit from the massive opportunities that the holiday season brings. ## Frequently Asked Questions (FAQs) ### What is the trend in holiday spending? Is holiday shopping up or down? Not surprisingly, consumer spending tends to increase during the holiday season. According to a recent survey by Salsify , a product experience management platform, more than half of consumers (53%) plan to maintain their 2024 holiday spending level in 2025. Additionally, 22% of holiday shoppers plan to spend more money this year than they did last year. The National Retail Federation has forecasted retail sales growth in 2025 between 2.7% and 3.7% compared to 2024. While it remains to be seen how much shoppers will spend during the 2025 holiday season, these projections indicate solid spending numbers. ### What do people buy the most during the holidays? According to National Retail Federation data, gift cards were the most desired product on people’s wish lists for 2024, at 53%. Other popular holiday purchases include clothing and accessories, books and other media, and personal care items. --- ### Beyond Last-Click — Measuring Engagement on the Open Web: Xevio eComm Insights (Part 3) URL: https://www.taboola.com/marketing-hub/beyond-last-click-engagement-measurement-open-web/ Last Modified: 2025-11-13 10:44:03 As the marketing funnel continues to get more complex, last click attribution is becoming a less reliable metric. I talked to Xevio co-founder and CEO Nadim Kuttab to get his thoughts on the value of softer metrics when measuring performance, as well as their limitations, and the future of performance measurement as a whole. Be sure to also check out our previous two installments, in which Nadim looks at driving direct response sales on the open web, and building high-converting e-commerce funnels with native content. ### Why is measuring engagement and intent crucial for e-commerce performance, even when the primary focus is direct sales? Because in the e-commerce space, it’s always important to look at several events in the buyer journey. Most affiliates start with last click tracking, because that's the easiest way of tracking what the user buys in one session. It's the dream of every marketer: There’s no better case for a marketer to say they’re providing value than if they’re spending a dollar, then making $2 back within 10 minutes. It's a great case! The truth is, though, that it’s becoming harder to rely on last click in marketing. People like to take their time in buying — they have options, they want to research it or look at TikTok reviews. There are so many things people do nowadays before buying. I know that when I had more time when I was younger, if I bought a laptop, that was a huge investment, so I’d spend hours reading about processors and graphics cards and battery life, going down these rabbit holes to inform myself before making a purchase. Most people are no different — maybe they don't spend 20 hours on choosing what shoes to buy, but they'll at least inform themselves on what alternatives there are, whether there's a better priced alternative from a competing brand elsewhere, etc. Maybe you’ll lose them in that process, or maybe they might come back later and buy via another channel — either way, if you're only looking at last click, another traffic source will be taking the credit for that sale. First click is also not an ideal way of looking at purchases, though. Both first click and last click have their place, but I recommend a mix of both, which is essentially distribution modeling. Using a mix of first and last click will give you a good idea of where the users are getting the first contact with your brand, and where they're buying. Often, though, it's going to be a mix of a lot of different things. We've seen with our native ads on Taboola that if you do it correctly, there should be about 2-4x as many first click conversions as last click, because Taboola is an educational platform, it's content-based — people are coming, they're reading, they're jumping away, buying it elsewhere. If you measure this correctly, it gives Taboola a much better standing for budgets and for scale, because you're not just looking at last click, which will only take into account 20-30% of all conversions Taboola actually helps generate — you're looking at the whole pie. That’s why it’s important to use one attribution tool across your whole business. ### How can metrics like "Time on Site" and "Session Depth" (as offered by Taboola’s performance platform, Realize) provide valuable insights into e-commerce buyer intent and campaign effectiveness? The issue that a lot of brands have is that there are a lot of different elements to test on Taboola. You have thousands of sites, you have dozens of ad formats, you have potentially dozens of landing pages and product pages behind those ads. So, if you really wanted to collect enough data to make decisions about everything on a purchase level, you would need to spend thousands a day. Some brands don't want to do that, so they need to make educated decisions quickly, without spending a ton of money, which is where these top of funnel conversion events come in. We use Voluum as a tracker, which tracks landing page CTR, and we always push that data back to Realize. We always have add-to-cart in our e-comm campaigns with Taboola, too. Then, if we collect enough data for one type of event, that's the event that we use. Obviously, the goal is to get enough purchases to optimize based exclusively on purchase data, but in lieu of that data, you need to look at pre-purchase events, which you then optimize until you have a campaign that's stable enough to generate those purchases. Think of it more as a means to an end, like stepping stones towards having a campaign that's scaling and spending enough to optimize exclusively on purchases. ### How do you identify and track "micro-conversions" or engagement signals (like product views, add-to-carts, or wishlist additions) that indicate an e-commerce user is progressing through the purchase funnel? The same way we track every other event — either with server to server, or we just fire that event back to Taboola. (If you're using a tracker in between, that's the same thing, just with an additional step.) At the end of the day, you want the data in Taboola so that you can look at it and understand, okay, this is an add-to-cart, this is a purchase, this is what I should be working towards. We're moving away from micromanagement of media buyers in campaigns. When I started media buying, every hour I'd go in and, like, bid up 5% or 3% on a site. It was a massive amount of work! It gave us a huge edge because we were doing that and others weren't, but the future of media buying on Taboola is very much automated. The biggest lever will be the creative. ### What strategies can e-commerce advertisers employ to leverage engagement data (such as creating engaged audience groups for retargeting) to improve overall campaign ROI? You need to put the Taboola pixel on your page — ideally on the checkout page or the product page — to retarget. Period. There's no exception for this, it's just a no-brainer. It's very easy, and it's free money! It can be a very, very easy ad where you just say, “Hey, here's a 25% discount on the product. Buy it here.” Our recommendation is to pixel everything, because why not? Taboola is also rolling out smarter features for building audiences to target, based on first party publisher data. That can be used to create pockets of audiences to target. Again, without data, Taboola can do nothing, so just feed it everything! We recommend feeding as much data as possible into the campaigns, because nowadays, the campaigns are very much on autopilot when it comes to media buying. ### So it’s essentially a classic case of, the more you put into it, the more you get out of it? Exactly. ### How do you balance optimizing for immediate direct sales with building long-term engagement and brand affinity for e-commerce on the open web? Xevio is a performance agency, so we’re measured on a dollar in, a dollar out. There are other agencies that have other targets, but that’s what most of our clients will look at and say, “We gave you a dollar, you made $3. Well done.” Or, indeed, “We gave you a dollar, you made $1.20, not so good — we need to improve that.” Obviously we consider lifetime value, which we try to increase by creating bundles or upsells or downsells that people can buy while they're in the purchasing process. But, the true long-term brand value is not something that you and I can properly assess by any measure. I don't think it’s our job as performance marketers to measure the long-term impact of brand equity — I think it falls squarely on the brand owners. As a performance marketer, if it's not our brand, the best we can do is educate the brand on the impact that we can have. ### What are the limitations of relying solely on engagement metrics for e-commerce performance measurement, and how do you combine those metrics with direct sales data? Engagement metrics are a means to an end, so relying on them isn't smart. It's more of a necessity — I don't think anybody with enough purchase or sales data would rely solely on soft metrics. Engagement metrics are really good at the beginning, when you have to get a campaign moving. If you don't want to use engagement metrics, you need to spend more money to buy data. Think of it like getting a plane into the air: You need to have engagement metrics to be able to see some data without burning money. Now, you could also just strap a rocket to it and it'll go up, but it'll be incredibly inefficient — you'll buy a lot of data very quickly and therefore be able to make a lot of decisions, and there are brands that do that: They just set, like, a $5k daily budget, spend it for a month and say, “Hey, we've learned something!” But, most of those brands are enterprise or very large growth brands. Engagement metrics are needed by, I'm guessing, the majority of the people that are going to read this, because they don't want to burn $100 grand on learning. Really, you have a choice: You can be inefficient with your money, but efficient with time, or you can be efficient with your money, but it’ll just take longer. You can save time or money, you can’t do both. Essentially, engagement metrics let us make decisions in a shorter period of time, with less budget. Are they going to be 100% accurate? Hell, no. But, at least they give us an indicator of what works, and if we see a trend between engagement metrics and sales, we can use that ratio to guesstimate what sales data we'll see. Will it be accurate? Again, no! But, it'll give us an idea of what to optimize towards. ### What's your vision for the future of e-commerce performance measurement on the open web, moving beyond simplistic last click models to a more holistic view of customer value? I think this is a great question for everybody that does online marketing, because every single advertiser on every channel will be forced to use a more holistic method, since people’s purchasing behavior is changing. It’s taking longer to get a sale, people are buying over more channels, people are taking their time when buying. Even with spontaneous purchases, they might look at it on their mobile phone, but buy it on their tablet or desktop. How do you track that? It's extremely difficult. There are ways to do it, but it requires unified attribution methods that are applied by brands universally, and that's what we're seeing. The very big enterprise brands are building these models themselves — good for them! It might cost a ton of money, but it's good for them because they can use that to make better decisions on their $100 million or $200 million budgets. It’s absolutely worth it if you're that big. For everybody else, we'll just use off-the-shelf solutions that'll cost a couple hundred or a couple thousand bucks a month, and that will help us make those informed decisions as best as possible. I don't think there’s a perfect way to do this, but if you’re applying one method to everything, you won’t be over-reporting revenue or attribution. I think that's the future. We’re moving into an area where I think, in five years from now, there won't be a successful e-commerce brand without a proper attribution model. First and last click do not cut it, even now! --- ### Social Media Marketing Trends 2026: Win With Video and AI URL: https://www.taboola.com/marketing-hub/social-media-marketing-trends/ Last Modified: 2026-03-16 11:44:45 The only thing guaranteed when working with social media marketing is that you’ll have to constantly adjust your approach to keep up. The landscape remains complex and challenging, so here, I’ll run down five important trends shaping social media marketing in 2026. What’s changed in our 2026 update: - All entries include updated and current information and advice. - All stats and figures updated with new and current information. - Information on Metaverse and Web3 updated in FAQs. - Grandview Research.png graph. ## Trend 1: The Dominance of Short-Form Video and Engaging Visual Content Short-form video isn’t just a trend: Platforms like TikTok, Instagram Reels, and YouTube Shorts continue to dominate consumer attention, including the desirable 18-to-34 age group. The appeal is clear, with short-form videos being immersive, easily digestible, and mobile friendly. But how long should a short-form video be? Rather than focusing on a fixed time range, marketers in 2026 are prioritizing strong hooks in the first few seconds, pacing, and completion rate over total length. Previously, many marketers recommended staying between 31 and 60 seconds, which can still work for certain use cases, but performance now varies more by format, platform, and intent rather than duration. Source: Yagauro.co That said, 2026 has brought a new level of sophistication to short-form video. Brands can no longer earn virality with dance videos and funny clips: To compete in an overcrowded marketing space, marketers need to grab attention and inspire action. Some key statistics related to video in 2026 include: - 63% of consumers state a preference for short-form video when researching a product or service. - Approximately 91% of marketers are expected to use short-form videos in their strategy in 2026. - 66% of marketers consider short-form content the most engaging format. - More than 50% of adults and 75% of Millennials/Gen Z consumers use captions even when not hearing-impaired, and brands that use captions see higher brand recall. Top strategies for short-form video success include: - Keep it vertical: Mobile has taken the lead in internet traffic, and that’s even more relevant when it comes to short-form video. To match that behavior, top short-form video platforms recommend a 9:16 aspect ratio. - Use captions and text overlays: Users aren’t always in an environment where they can listen to content. Captions and text overlays make your videos more accessible to viewers. - Prioritize strong hooks: In 2026, algorithms favor early engagement, making the first 1 to 3 seconds critical. - Leverage creator-style formats: Authentic, creator-driven visuals often outperform polished brand ads. - Incorporate trending sounds and challenges: In 2026, this remains relevant for some organic and creator-led content, but it’s been less central to brand performance than in past years. - Focus on storytelling: Today’s consumer tends to tune out obvious marketing pitches. Instead of using these platforms to sell your products or services, put the emphasis on storytelling, and sales will follow. - Test interactive tools: Polls, augmented reality filters, and clickable overlays can be a great way to engage your audience. - Go live: Live video remains effective for specific use cases like shopping, education, and customer Q&A, but it’s no longer a universal growth tactic. Performance marketing platforms that focus on results, not reach, are key. You don’t just want to put your content in front of users — you want to inspire action. In 2026, advanced performance platforms increasingly use predictive signals and cross-channel data to help advertisers repurpose top-performing creative and deliver it at moments most likely to convert. ## Trend 2: The Rise of Niche Communities and Authentic Engagement Algorithms have taken over social media feeds, particularly with the rise of artificial intelligence (AI). As a result, consumers have sought out smaller, more niche spaces where they feel seen and valued. Whether they’re on Discord, Reddit forums, or Facebook groups, social media users seek community and authenticity. - Smaller, niche, human-centric communities are growing in popularity, with groups of less than 100 taking priority. - Influencers with 1,000-10,000 on platforms like TikTok now average ~10.3% engagement rates, which is significantly higher than many larger accounts. - About 61% of brands report working with 10 or more influencers. - 58% of adults report buying products because of influencer recommendations. Source: Influencer MarketingHub For brands, this shift brings more work. It’s no longer sufficient to blast a message to the masses: Instead, marketers have to put time into building and participating in communities. You can reach users with behind-the-scenes videos and posts that elevate users. Even sharing your customers’ posts can show your support for brand loyalists. Engaging with these communities requires a genuine presence, though: Members can tell if a brand is simply engaging to market to them. Brands should participate in conversations and provide value without selling. In 2026, many brands are also shifting from traditional influencer campaigns to longer-term creator partnerships that emphasize trust, consistency, and shared values. ## Trend 3: The Evolution of Social Commerce and Direct-to-Consumer Sales Social media has moved from networking to shopping. In 2026, social platforms play a central role in discovery and influence, even when final conversions happen elsewhere. Source: Grandview Research Previously, in-app purchases, shoppable posts, and seamless checkout purchases meant businesses could grab sales directly on social media sites. In 2026, brands are more selective, using native checkout where it delivers frictionless results, while supporting hybrid journeys that move users to owned e-commerce experiences. Here are a few notable statistics related to social commerce: - More than half of Gen Z and millennials have purchased through social media recently, outpacing older cohorts. - TikTok Shop is expected to exceed $20 billion in global sales in 2026, highlighting rapid social commerce traction. - China remains one of the strongest markets for social commerce, with conversion rates at about 30%, driven by livestream shopping, influencer marketing, and integrated social shopping experiences. Take note, though, that as convenient as these integrated shopping features can be, brands still need to build a shop that attracts and engages customers. Live shopping events and real-time product launches can help turn content into conversion points. You can also tap influencers for product/service walkthroughs to grab the interest of top-of-funnel customers. The challenge is in making every touchpoint shoppable and engaging to boost the chance of a conversion. In 2026, data-driven timing and content sequencing have become just as important as D2C shoppable formats themselves. In today’s competitive market, timing is everything. Analyzing customer behavior can help you pinpoint peak engagement times and preferred content formats. By gathering this data, you can create content your target customers want to see and post it at the time they’re most likely to engage with it. ## Trend 4: Leveraging Data and AI for Personalized Social Media Marketing Social media is vital in reaching customers, but audience saturation, rising costs, and ad fatigue have led to diminishing returns. In fact, 75% of marketers say their efforts are seeing declining results. This has led marketers to seek out better ways to grab attention using data and the latest AI-powered technology. In 2026, personalized marketing is more important than ever. Don’t believe us? Here are some statistics that could change your mind: - 81% of Gen Z customers and 57% of millennials like personalized ads. - Nearly 69% of shoppers say personalized product recommendations feel relevant and helpful. - Almost 50% of consumer data collected by companies is used for personalized or targeted ads. - 63% of marketers plan to increase their budgets for hyper-personalization in 2026. Here are some ways you can stand out in an increasingly competitive market: - Audience segmentation: Privacy concerns have made it tougher for marketers to target audiences by personal details like gender and past browsing habits. AI-driven platforms now focus on behavioral signals, intent, and predictive modeling rather than demographics. - Creative variation testing: Creatives can make a big difference in your ad’s performance. Today’s tools let you try different text and visuals, then monitor results to identify the ads that convert best. - Automated bidding strategies: The paid advertising space is more competitive than ever, making it tough for marketers to stay on top of their bids. Tools that continuously optimize spend can help advertisers gain an advantage on the most popular social media platforms. - Conversational AI: AI-powered messaging and moderation tools allow brands to maintain engagement and support communities at scale, without sacrificing responsiveness. Traditional social media marketers are also finding value in expanding into digital channels outside of social media. PortAventura World, for example — a tourist resort in Spain — achieved a 44% increase in return on ad spend (ROAS) by placing ads across top publisher sites. By combining your social media efforts with other advertising efforts, you can create an omnichannel marketing strategy that works well into the future. ## Trend 5: Navigating Platform Changes and Emerging Social Media Trends The only constant in social media is that it will change. For the past few years, social media has been on a trajectory toward more privacy and niche user spaces, so the best thing brands can do is get ahead of future shifts, and that requires agility and a willingness to experiment. Some current shifts I’m seeing in 2026 include: - Instagram’s focus on AI-suggested video feeds. - TikTok’s new specialized rewards program. - Ephemeral content like Instagram Stories driving engagement. Emerging platforms BeReal and Lemon8 previously drew Gen Z attention, but growth has since stabilized, underscoring the risk of chasing novelty without a clear performance strategy. Here are some statistics to keep in mind as we tackle the rest of 2026: - More than 58% of the world’s population uses social media. - Globally, the average daily time spent on social media is 2 hours and 21 minutes per user. - The most popular social media platforms by active monthly users are: Facebook (3.2 billion), YouTube (2.8 billion), Instagram (2.2 billion), TikTok (1.7 billion), and LinkedIn (1 billion). Staying updated on platform changes and user behavior trends is more crucial than ever. You can do this by regularly reviewing industry reports while also keeping an eye on platform analytics. This will allow you to adjust your strategies proactively, helping you stay ahead of the curve. ## Key Takeaways When it comes to social media marketing, short-form video and visual content dominate in 2026. It’s also crucial for brands to build community, but authenticity is essential. If brands can find audiences in smaller, more niche communities, they may have an easier time reaching consumers. But, constant platform changes require brands to be flexible. Staying on top of industry shifts and your own analytics can help you future-proof your marketing campaigns. ## Frequently Asked Questions (FAQs) ### What are the most effective social media platforms for advertising in 2026? The most effective platforms are those that help you meet your goals and reach your intended audience. For visual content, TikTok, Instagram, and YouTube dominate, while Pinterest and Facebook rule when it comes to social commerce. Demographics are also a consideration. Facebook still skews older, while TikTok and YouTube remain great ways to reach Generation Z. ### How can brands measure the ROI of their social media marketing efforts? To measure ROI (return on investment), look beyond likes and follows. Metrics like return on ad spend, cost per acquisition, click-through rates (CTRs), and conversion rates offer better insights into how your ads are paying off. The right performance platform will offer this information at a glance, along with AI-powered insights that can pinpoint areas where you can boost ROI. ### What are some common mistakes to avoid in social media marketing? Failing to pivot remains one of the biggest mistakes a brand can make in 2026. It’s also important to spread your efforts across multiple platforms rather than relying on one, particularly if you’ve focused on the same platform for years. Neglecting to engage with your audience and ignoring user feedback can sabotage your social media campaigns. ### How is the metaverse and Web3 impacting social media strategies? Interest in the metaverse and Web3 has cooled, but some brands continue experimenting. In 2026, practical AI-driven personalization and performance optimization have taken precedence over speculative virtual experiences. ### What are the key metrics for measuring success in social media marketing? In this area, 2026 isn’t all that different from previous years. CTRs, conversion rates, and customer lifetime value are all essential metrics. You can also keep an eye on engagement rates, including likes, shares and comments, to get a feel for audience sentiment. --- ### 5 Influencer Marketing Trends for 2026: What to know URL: https://www.taboola.com/marketing-hub/influencer-marketing-trends/ Last Modified: 2026-03-08 13:49:18 Over the past ten years, influencer marketing has gone from being an experimental novelty to a core revenue channel for brands of all sizes. As trust in traditional digital advertising continues to fragment, creators remain one of the most effective ways to reach audiences in environments where attention is voluntary, contextual, and relationship-driven. However, influencer marketing in 2026 looks markedly different from even a year or two ago. It is no longer defined by follower counts, likes, or one-off sponsorships: Instead, it’s shaped by performance accountability, creator-led commerce, tighter compliance standards, and full-funnel integration. Brands are asking harder questions about incremental value, while creators are prioritizing sustainability, ownership, and long-term partnerships. The result is a more measurable and more strategic influencer ecosystem. ### What’s changed in our 2026 update: - All entries include updated and current information, figures, and stats. ## Trend 1: The Maturation of Influencer Marketing and the Focus on ROI Influencer marketing has fully crossed the threshold from experimental brand outgoing to a performance-oriented growth lever. In 2026, brands increasingly plan influencer programs with the same intensity as paid search or paid social, with defined KPIs, structured testing, and budget accountability tied to outcomes rather than exposure. What has changed most is how ROI is evaluated. While affiliate links, promo codes, and UTMs remain foundational, many teams are now layering in incrementality testing, lift studies, and blended attribution models to understand the true contribution of creators across the funnel. This shift has been accelerated by platform behavior that keeps users in-app longer, reducing clean click-through signals and forcing brands to think beyond last-touch attribution. Another notable evolution is the normalization of content licensing and usage rights as part of influencer contracts. Brands are no longer paying only for organic distribution, but also for reusable creative assets that can be deployed across paid media, ecommerce, and owned channels. Performance-based compensation models, combining a base fee with conversion or revenue, are also becoming more common as both sides seek aligned incentives. At the same time, legal and compliance maturity continues to rise. Contracts increasingly include performance benchmarks, content ownership clauses, exclusivity windows, and clearly defined disclosure requirements aligned with FTC regulations. Stats you should know about influencer marketing in 2026: - The global influencer marketing market was valued at approximately $32.55 billion in 2025, up from $24 billion in 2024. - 86% of U.S. marketers reported working with influencers in 2025, reinforcing influencer marketing as a mainstream channel. - 63% of businesses now include ROI-specific targets in influencer contracts. ## Trend 2: The Continued Rise of Micro and Nano Influencers for Targeted Engagement Bigger doesn’t always mean better, particularly for small to medium businesses with modest budgets and specific target audiences. Micro influencers (those with under 100,000 followers) and nano influencers (those with under 10,000 followers) are continuing to outperform even the most famous influencers when it comes to tangible performance metrics, including the mid-tier segment in between. This is especially the case in the beauty, fashion, food, and wellness industries. These smaller creators often have hyper-engaged communities that brands can benefit from. Their followers trust their recommendations, and due to their smaller audience, influencers are able to engage one-on-one with many of their audience to create an authentic, two-way relationship. With such a personal connection to their followers, brand advertising still feels personal, rather than promotional. As they operate on a smaller scale, for brands on a tight budget, working with micro and nano influencers can often be more cost-effective. This means that you can work with multiple creators for the cost of a single, larger influencer. What’s changed in 2026 is how brands activate these creators. Rather than running one-off collaborations, many teams now build structured creator portfolios, working with dozens or even hundreds of smaller creators simultaneously. This approach diversifies creative angles, reduces dependency on any single partner, and produces a steady flow of authentic content that can be tested and repurposed across channels. Tools like Upfluence and GRIN can help streamline your influencer management, from outreach to content approvals and reporting. This is especially useful for micro and nano influencers who likely aren’t working with their own management companies. Working directly with influencers also means that you can start to build mutually-beneficial relationships with them, encouraging long-term professional connection. Micro-influencer stats for 2026: - 45% of marketers find that small influencers have more trust with their followers than macro influencers. - Micro influencers generate up to 60% more engagement than macro influencers. - The majority of Instagram influencers globally are nano-influencers with fewer than 10,000 followers. ## Trend 3: The Importance of Authenticity and Transparency in Influencer Partnerships With the increasing use of AI, consumers have become both savvy and skeptical when it comes to online content. Fake recommendations are easy to spot, while others are quick to call out influencers or brands that don’t disclose their partnerships correctly. Regulators are also taking note. The FTC updated its guidelines in 2023 to require influencers to clearly communicate when posts are sponsored or feature gifted content. Brands are also being held accountable for these partnerships, with both the influencer and brand facing fines for non-compliance. What’s new for 2026 is the growing concern around synthetic media. AI-generated endorsements, manipulated testimonials, deepfakes, and virtual creators introduce new layers of brand risk if audiences are misled or disclosures are unclear. As a result, influencer vetting now extends beyond follower fraud to include content authenticity, creator history, and transparency around AI usage. Authenticity in influencer content goes so much deeper than FTC compliance, though: It’s about relevant storytelling and real-world relatability. User-generated content (UGC) has seen a significant rise in recent years and influencers are now creating content in this same style, even for paid promotions. Behind-the-scenes glimpses and honest reviews now resonate more than a polished ad — it’s no longer about looking perfect. Instead, consumers want honesty and transparency. Digital advertisers play an important role here, too. Vetting influencer history, looking at their previous campaign metrics and engagement quality through PR auditing tools, can all help brands find out if an influencer is who they say they are. This is an essential step for ensuring brand integrity before a campaign goes live. Transparency in influencer marketing stats that you need to know: - One in three consumers say too many sponsored social media posts cause them to lose trust in influencers. - 77% of consumers trust content from people similar to themselves. - FTC fines for deceptive practices can be up to $50,120 per violation. ## Trend 4: The Evolution of Platforms and Formats for Influencer Content As the world of influencer marketing has continued to expand, so too have platforms to support this type of content. Short-form videos are continuing to dominate in this form of marketing, especially those under one minute, but other formats, including long-form content on YouTube, are also seeing increased engagement. The fastest-growing formats are those native to shopping behavior. TikTok Shop, livestream shopping, and integrated affiliate tools have pushed influencer marketing closer to performance affiliate models, particularly for DTC, beauty, fashion, and lifestyle brands. Virtual influencers and AI-generated creators are also starting to enter mainstream influencer marketing campaigns. There remains controversy around these, though, with brands looking for consistent messaging 24 hours a day choosing to make use of this new technology, but consumers remaining wary of supposed “influencers” who, by definition, have no actual lived experience. Podcasts and newsletters are also growing influencer partnership channels thanks to the rapid growth of platforms like Beehiiv and Substack. Influencers are able to build their community without algorithmic interference, an approach that brands can capitalize on with strategic sponsorships and partnerships. Stats around the evolution of influencer marketing in 2026: - TikTok Shop generated over $500 million in U.S. sales during the 2025 Black Friday-Cyber Monday sale. - 45% of influencer marketers now work with podcast hosts as brand ambassadors. - Threads surpassed X in daily mobile users by early 2026, signaling shifting attention patterns. - The global influencer market is expected to reach $84.89 billion by 2028. ## Trend 5: Influencer Marketing Becoming a Cross-Channel Growth Asset Influencer marketing in 2026 performs best when deeply integrated into broader marketing strategies, rather than run as a standalone channel. Brands are increasingly treating creator content as a media asset that can be deployed across paid, owned, and earned channels. Creator content is now commonly licensed and repurposed for paid social ads, ecommerce product pages, email campaigns, SMS flows, and even retail media placements. Whitelisting and creator-led paid amplification has become a standard tactic for blending authenticity with targeting and scale. This shift requires more sophisticated measurement. When influencer content fuels both organic trust and paid performance, teams must align on shared success metrics across social, site analytics, CRM, and media platforms. As a result, influencer marketing is increasingly planned alongside SEO, paid media, and lifecycle marketing. Influencer marketing stats you need to know in 2026: - 71% of marketers use influencer content in paid ads to boost performance. - Brands see an average 20% increase in click-through rate when using UGC in email campaigns. - 52% of marketers repurpose influencer content across three or more channels. - U.S. creator ad spend rose from $29.5 billion in 2024 to a projected $37 billion in 2025, accelerating integration with broader media plans. ## Key Takeaways For direct-to-consumer brands, influencer marketing can be a powerful tool for growth beyond traditional digital marketing channels. By focusing on ROI and nurturing relationships with micro and nano influencers, brands can create authentic marketing campaigns that remain transparent and enticing to a growing audience. ## Frequently Asked Questions (FAQs) ### What are the most effective platforms for influencer marketing in 2026? TikTok, YouTube, and Instagram are some of the most popular and effective platforms for influencer marketing in 2026. ### How much should brands budget for influencer marketing? Small to medium businesses typically spend around 10-25% of their marketing budget on influencer campaigns, with many influencers in the micro and nano markets charging $500-$1,000 per campaign. ### What are some common mistakes to avoid in influencer marketing? Lack of clear goals, poor influencer vetting, ignoring FTC compliance, and failing to repurpose content are some of the biggest mistakes brands can make when it comes to influencer marketing. ### How can brands protect themselves from influencer fraud? Tools like HypeAuditor are some of the best ways to detect fake followers on an influencer’s profile, while also managing multiple influencer relationships and campaigns at one time. ### What are the key trends in influencer marketing measurement and analytics? ROAS tracking, affiliate link attribution, sentiment analysis, and cross-channel performance are all some of the best ways to determine how successful an influencer marketing campaign is. --- ### What Makes a Great Taboola Advertiser? Here’s Who to Refer (and Why You’ll Benefit) URL: https://www.taboola.com/marketing-hub/taboola-referee-characteristics/ Last Modified: 2026-03-19 13:37:42 Modern performance marketers tend to be creative and resourceful, always looking for new ways to engage prospects and stand out from the crowd. Taboola’s referral program helps tap into that spirit, rewarding referrers even if they aren’t themselves Taboola customers, and connecting new users to all the benefits Taboola’s platform provides. A great Taboola advertiser is data-driven, willing to try new things, and shift direction as needed in real time. Typical Taboola users are marketers managing ad campaigns, whether individuals or agencies; publishers, such as websites and platforms that host content; content creators; and marketing and advertising strategy consultants, among others. They’re all seeking to grow their business, attracting new audiences with compelling ads. ## 3 Qualities to Look for in Your Taboola Referee Here’s what to keep in mind when you’re considering colleagues and other connections to refer to Taboola. ### 1. An Understanding of Modern Marketing Performance marketing, unlike full-funnel marketing or high-level brand awareness, aims to target users with laser precision for better conversion rates, lower customer acquisition costs, and improved return on investment (ROI). Performance marketers track data points carefully and optimize campaigns continually as they learn more about their audiences. Beyond identity data, modern marketing requires careful attention to user intent to pinpoint the right users, who can then benefit from what a brand is offering. And, while search and social channels have been mainstays of advertising for many years, modern performance marketers know they have to move beyond that to get more for their money and engage their target prospects. ### 2. A Need for Speed With oversaturated digital marketing channels, higher ad costs, and pressure for advertisers to do more with less, performance marketers can’t afford to test programs over a quarter or longer. Smaller businesses in particular have a fixed budget, so it’s essential to change course as quickly as possible when a strategy or tactic isn’t working. Artificial intelligence (AI) and machine learning (ML) capabilities have improved targeting abilities, as they learn from how campaigns are performing, allowing marketers to revise their plan or optimize in close to real time. Many of Taboola’s successful referrals are already using platforms like Meta or Google and are looking to diversify and scale with performance advertising. Companies working to scale their business significantly, or reach new, high-quality audiences, are a good fit for the Taboola referral program. ### 3. A Clear Set of Goals A modern buyer’s journey is often complex and nonlinear. A prospect might see several social media posts, engage with a brand, then click on an ad weeks or months later. Another might see search ads repeatedly but never click on one. These behaviors can make digital marketing much more challenging, and marketers must look for creative ways to reach users and meet their goals. While marketing generally aims to find and convert new users, modern performance marketers have more specific goals. That could be to increase new users, gather more qualified leads, reduce acquisition costs, and more. The specific goals and relevant metrics might change quarter-over-quarter or month-over-month depending on company growth goals and other factors. Whatever the case, performance marketers using a platform like Taboola can articulate what they’re trying to achieve and in what timeframe. Taboola’s Realize platform is often the natural next step for marketers who are familiar with Google and Meta and looking to take their efforts further. ## Key Takeaways Taboola’s referral program is designed to be accessible to a range of advertisers and marketers in businesses of all sizes. Referrers don’t have to be Taboola customers themselves, and can refer as many relevant connections as they like. Performance marketers use Taboola to cut through user and creative fatigue and target high-intent users with fresh, continually updated data. ## Frequently Asked Questions (FAQs) ### Who can refer an advertiser to Taboola? Anyone, whether a Taboola customer or not, can refer an advertiser to the Taboola referral program. Clients, partners, and any industry contacts who could benefit from Taboola are eligible. You can invite a potential customer on the Advertiser Referral Program page if they haven’t been in active discussion with Taboola in the past 180 days. If the lead is approved and signs an insertion order (IO), puts the Taboola pixel or server-to-server (S2S) tracking on its landing pages, and spends a minimum of $5,000 within the first 45 days, the referrer is eligible for the $1,000 payment. ### When and how is payment received for a successful referral? Taboola’s referral program offers $1,000 to qualified referrals, which means that the advertiser referred was first qualified as a lead, then spent $5,000 within 45 days. The person referring the new advertiser can then invoice Taboola, and expect to get payment by bank transfer within 30 days. ### What are the goals of performance marketing? Performance marketing captures the nuances of modern advertising and marketing practices more than traditional full-funnel marketing can do in a digital environment. Performance marketing aims to target the right audiences at various points in their buying journey, rather than casting too wide a net and wasting money. Performance marketing platforms carefully track campaign performance and optimize ad spend, and are designed to deliver measurable results and show ROI. What metrics should I track for performance advertising? It’s all about the metrics for performance marketing. Common metrics include cost-effectiveness to ensure there isn’t wasted ad spend; ROI calculations; more and higher-quality leads captured; and metrics like click-through rate (CTR), conversion rate, and cost per acquisition (CPA). These data points will be key for optimizing campaigns, and making data-driven decisions in the moment. --- ### How to Create A Black Friday Landing Page That Converts with Realize URL: https://www.taboola.com/marketing-hub/black-friday-landing-page/ Last Modified: 2025-07-27 08:41:27 Love it or dread it, Black Friday has changed the game for marketers and shoppers alike, both online and in real life. It's a chaotic, exhilarating sprint to the finish line (which seems to get longer every year), where every click, scroll, and second counts. As an advertising copywriter, I’ve seen more Black Friday campaigns than I can remember at this point. I’ve also learned that a high-converting landing page isn't just nice to have — it's a non-negotiable these days. So, how do you cut through the noise of it all and build a Black Friday page that does more than just look flashy? How do you create something that truly converts, turning casual browsers into buyers? Let's dive into some tried-and-true approaches, straight from the trenches of holiday marketing. I asked Taboola’s director of creative shop and AI strategies, Mayaan Leshem, to delve into details with me and share some helpful tips. ## 5 Tips to Create the Best Black Friday Landing Page ### 1. Think Mobile, Act Early Something obvious that needs to be reinforced anyway: Your customers are glued to their phones. At this point, it’s officially not just a trend, it's the dominant shopping experience. Mobile accounts for 55% of holiday e-commerce, and mobile holiday revenue now outpaces desktop, and is expected to grow. This isn't just about having a responsive design anymore, it's about building a mobile-first experience from the ground up. “Make sure your content is thumb-stopping, fast-loading, and easy to browse on small screens,” says Leshem. In addition, she recommends promoting exclusives for mobile shoppers to encourage early action. Timing is another big part of launching a landing page, and the best time to start the process, quite honestly, is now. Nearly half of holiday shoppers plan to shop before November: People, like myself, want to avoid the last-minute holiday rush and check off their list as early as possible. So, launch your holiday-specific, mobile-optimized page by October to reach those eager shoppers. This means getting your content, exclusive mobile bundles, and limited drops live well in advance, with plenty of time to deal with problems that come up. As a creative copywriter, my favorite part of the job is to write funny, clever headlines, but that’s not always what’s needed. For Black Friday landing pages, the headlines should communicate a simple message of urgency. Let potential customers know your shop is live and ready for mobile shoppers, and promote any special offers you have. ### 2. Make Your Creative Festive, Authentic, and Kinetic The landing page itself is where you can lean more heavily into creativity. Generic is boring, especially during the holidays: Think about how many ads you see every June that use “Dads and Grads” or something equally bland and emotionless. It all blends together as a blur in a reader’s mind. Remember, a memorable site doesn’t just look pretty — it can drive a higher conversion rate as well. “During the holidays, emotions run high — tap into that energy and align your creative story with the spirit of the season to build trust and resonance,” says Leshem. “Feature warm, candid imagery of real families, cozy settings, or shared holiday moments.” Regarding that last part, it doesn’t mean just slapping some snowman art on it. Your visuals need to feel real and organic: Think unfiltered, user-generated content (UGC) and style authenticity, since users can often sense when a site was hastily thrown together. DIY imagery builds trust and is often more relatable than polished stock photos, so try to show products within genuine holiday moments that fit your audience. “Stock-looking visuals feel flat,” confirms Leshem. “Authentic holiday footage drives deeper connection.” You also want to avoid images that look too static. “Use kinetic visuals — movement, gestures, and smiles — to create emotional pull,” advises Leshem. To that end, looping, short-form HD videos that are human-driven capture attention and truly bring products to life. Don't just tell them about your new decorative cake mold, show them how it fits into their holidays with a "SEE IT IN ACTION" button. It can really make a huge difference. ### 3. Personalize and Gamify the Experience In a sea of Black Friday deals, how are you supposed to stand out? “Generic gifting doesn’t cut it anymore,” says Leshem. “Today’s shoppers respond to content that feels custom and fun. Try adding interactive elements, like quizzes or gift finders, to increase engagement. Use gamified calls-to-action (CTAs) such as ‘Play to Shop’ or ‘Take the Quiz’ to make the experience feel rewarding.” Interactive elements like these speak directly to specific audiences, and can end with CTAs that turn browsing into an engaging experience. It doesn’t have to be cutting-edge web technology, either: A "Find Your Gifting Style" quiz with copywriting that’s funny, informative, and leads to a personalized gift list can work wonders. Even a 60-second quiz to help shoppers find gifts for a family member who is tough to buy for could be effective. Leshem also recommends segmenting by shopper behavior (e.g., last-minute buyers, early planners, or gift givers for specific demographics). Making your page personalized goes a long way, especially for a burnt-out shopper who’s been browsing endless other sites. Shoppers love to feel understood and seen: “Personalization boosts relevance,” Lesham adds, “and relevance drives clicks and conversions.” ### 4. Highlight Discounts and Ensure a Frictionless Checkout This one might seem like a no-brainer, but it's worth emphasizing: Black Friday is all about the deals, so don’t lose sight of that — it’s what shoppers are looking for. Your landing page should communicate the savings clearly from the moment someone lands on it. You may have seen the page in layout a million times, but to them, it’s fresh and new. “Shoppers are deal-hungry and pressed for time: Lean into urgency and simplicity,” insists Leshem. But, a great discount is only half the battle. If your checkout process is clunky, customers may give up halfway through, which means you're leaving money on the table. “Match your ad message with a frictionless checkout experience to keep the momentum going,” advises Leshem, adding these helpful tips: - Highlight deep discounts prominently (e.g., “75% Off. Zero Hassle.”). - Use phrases like “Ends Soon” or “Limited-Time Offer” to spark FOMO. - Emphasize speed: Promote features like one-click checkout, no logins, and instant shipping. - Drive conversions with direct CTAs such as “Shop & Save Now” or “Buy Now.” Basically, the fewer the hurdles, the higher your conversions. Keep your landing pages simple and straightforward. ### 5. Strategize Your Campaign Timeline Timing is everything, and your Black Friday strategy shouldn't just kick off in November — it's a multi-phase operation. Here’s a general timeline of how a plan could look: #### October (Early Shopping and Research) “Holiday shopping starts earlier each year, and mostly happens on mobile,” reiterates Leshem. “Launch mobile-optimized landing pages early (ideally by October) to catch planners and early birds. ‘Early access’ and ‘mobile-only drops’ can give mobile-first campaigns a powerful edge.” #### November (Consideration) Holiday shopping is ramping up, so nurture intent with gift guides, comparisons, and personalized messaging. #### Cyber Five (Discount Purchasing) The five days from Thanksgiving through Cyber Monday is the main event. Lead with urgency, highlighting your biggest deals and fastest checkout options. #### December (Last-Minute Purchasing) Be sure to emphasize speed, shipping deadlines, and last-chance offers. #### "Q5" (Christmas through New Year’s: Return and Refresh) Almost at the finish line, so don’t stop just yet! Capture return traffic with self-gifting, clearance sales, and New Year promotions. This is a great time for industries like finance to position products as the "first smart money move" of the new year, or health brands to offer a "post-holiday reset." ## Leveraging Realize for Black Friday Success There’s a ton of possibilities, but platforms like Realize can help make your vision a reality. As a performance engine designed to deliver prospecting and conversion outcomes on a cost-per-click (CPC) basis, Realize brings some serious power to your Black Friday efforts from multiple angles: ### Unique Data and AI that Drive Results at Scale This means faster creative variations and further optimizations, all driven by artificial intelligence (AI). It’s a game-changer for quickly testing and iterating on those festive, authentic, and kinetic creatives mentioned above. ### Creative Formats and Placements Beyond Native  With Realize, you’ll get creative formats and placements beyond traditional native ads, entirely designed for performance. This broad reach helps you captivate audiences with timely holiday footage that feels real and organic, not just emotionless stock. “Some of the most engaging holiday campaigns are brought to life through a variety of creative formats that are available in Realize,” says Leshem. “With Display, Rich Media, Native, and Carousel formats at your fingertips, you can build immersive, high-performance experiences tailored to your performance goals. Whether you're driving urgency with a bold Display banner, gamifying a gift guide with Rich Media, storytelling through Native ads, or showcasing variety with Carousel, Realize empowers you to meet shoppers wherever they are — at every scroll, tap, and click.” ### Control and Transparency Realize offers marketers greater control and transparency, which is crucial for monitoring your campaign performance and ensuring you’re getting the most out of your budget. ### Seamless Integration and Tracking Realize integrates quickly with platforms like Shopify and WordPress for seamless pixel setup and automatic conversion tracking, eliminating manual coding or theme limitations. This is a big win, as implementing tracking four to six weeks in advance is vital to collect data on users interacting with your brand before peak season. ### Reaching High-Value Shoppers Realize's Predictive Audiences allow you to go beyond your core audience, tapping into high-intent users who are likely to convert during the peak season. This means reaching new customers who are primed to buy. ### Stability During Peak A critical piece of advice: Avoid major changes to targeting, bidding, or other settings during peak days. This prevents re-triggering the learning phase and keeps your campaigns stable when it matters most. ## Key Takeaways When it comes to Black Friday landing pages, mobile-first is mandatory: Design for phones first, since mobile dominates holiday e-commerce. Launch your holiday content and offers by October to capture early shoppers and use festive, real, and kinetic (short-form video) visuals to boost engagement. Engage shoppers with quizzes and tailored content and clearly highlight deals. ## Frequently Asked Questions (FAQs) ### What are the best tools to create a landing page? Platforms like WordPress are incredibly helpful due to their flexibility and extensive plugins, allowing for custom designs and integrations, and user-friendly interfaces. Other popular options include Unbounce, Leadpages, Instapage, and even website builders like Shopify or Squarespace, which have lots of landing page functionalities. ### Should I redirect a Black Friday landing page after the sale is over? Yes! It’s generally a good idea to redirect your Black Friday landing page after the sale ends. This prevents visitors from seeing outdated offers and helps maintain a positive user experience. You can redirect it to your homepage, a general sales page, a "current offers" page, or even a relevant category page. A 301 redirect (permanent redirect) is usually recommended for search engine optimization (SEO) purposes. ### Why do I have no conversions if I have hundreds of visitors to my landing page? A high number of visitors with low conversions can be one of the more frustrating experiences, but it points to a disconnect between your traffic and your landing page's ability to persuade people. Here are some common reasons why: - Mismatch between Ad and Landing Page: Your ad might promise one thing, but your landing page delivers another. Ensure consistent messaging and clear offers. - Poor User Experience: Is your page slow to load, hard to navigate on mobile, or cluttered? Remember that mobile is 55% of holiday e-commerce, so mobile optimization is key. - Unclear Call-to-Action (CTA): Is it obvious what you want visitors to do next? Is the button prominent and compelling? - Lack of Urgency/Value: Are your deals clear and enticing enough? Do visitors feel a reason to buy right now? - Missing Trust Signals: Do you have customer reviews, security badges, or clear return policies? - Too Much Hassle: Is the checkout process too long or complicated? Prioritize a fast checkout so buyers don’t bail midway through. - Audience Mismatch: Are you driving the right audience to the page? It’s possible that your targeting may need some refinement. --- ### Unwrap a High ROI: Holiday Creative Strategies That Drive Results URL: https://www.taboola.com/marketing-hub/maximize-roi-creative-practices/ Last Modified: 2025-10-28 10:05:36 The holiday season is a major investment for advertisers, and for many businesses, it can make or break the year’s advertising and sales goals. A solid creative strategy can help marketers ensure a great return on investment (ROI), since eye-catching, unique, on-brand creative can make all the difference in marketing ROI. “Creative is the single biggest lever we control,” says Anthony Blatner, founder and CMO of Speedwork. “Based on LinkedIn’s own data, ads built on a clear story and strong visual identity generate 6-7x the ROI of ‘safe’ creative, because they lift recall and consideration at the same time.” Creative development and measurement for holiday ad campaigns should start early, taking into account the relatively condensed time period and the opportunity to capture new audiences. “Seasonal creative often performs better because it’s timely,” adds Blatner. Businesses of all sizes and industries can craft their holiday creative carefully with an eye toward great returns. Check out these seven tips on how marketers can strike the right balance of maximizing ROI and thoughtfully, creatively engaging prospects during the holiday season, all based on data insights and trends from Taboola’s creative team. ## 7 Ways to Maximize ROI This Holiday Season With Creative Practices ### 1. Understand the Link Between Creative and ROI Creative best practices — including visual imagery, copy, interactive components, and more — aren’t a siloed effort from the rigor of holiday marketing campaigns. Putting time and effort into creative advertising can have a direct impact on ad click-through rates (CTR) and conversion rates (CVR), ultimately leading to improved ROI. Taboola’s creative team found, for example, that holiday-themed ad visuals can boost CVR by 4x compared to generic creatives. The cost of ineffective creative, whether it’s bland, repetitive, or low-quality, shows itself in wasted spend, reduced conversions, higher customer acquisition costs, and lost opportunity cost. Marketing teams can’t do holiday season over with better creative — it’s essential to spend time on it well ahead of the season starting. ### 2. Be Authentic and Stay on Theme Unfiltered, UGC-style creative can help build trust and is generally more relatable than stock photos. Candid, DIY imagery is a straightforward way to capture the holiday spirit and stand out to users and prospects in a busy market. Vibrant colors and festive themes can also help you stand out, as can emotional triggers and themes. For B2B companies, holiday themes in creative can help stand out and connect to the season. This often extends throughout December, folding in end-of-year and New Year’s creative opportunities, and taking advantage of holiday slowdowns as periods of reflection and goal-setting. “Give before you ask,” says Blatner of businesses running holiday campaigns. “Insight reports, ROI calculators, or year‑in‑review benchmarks are often great introductory pieces that can help draw in interested prospects.” ### 3. Know Exactly What to Measure Advertisers and marketers have to get as specific as they possibly can on capturing metrics that show how well creative is performing. Again, timely creative like holiday ads often perform better, according to Blatner, and he recommends examining performance at various levels. That might start with micro, “thumb-stop” metrics like scroll depth and carousel swipes to see if the creative is being consumed. Next, mid-funnel metrics like CTR, cost per lead (CPL), and the qualified lead rate can show whether the story resonates with the right people as planned. At a macro level, Blatner recommends measuring pipeline dollars, win rate, CAC-to-LTV (customer acquisition cost to customer lifetime value), and close/won revenue that connects creative to the CFO’s scorecard. “For LinkedIn, we stitch all of that together with conversion tracking and offline CRM uploads,” says Blatner. “For bigger accounts, we do brand lift studies to quantify sentiment shift.” These numbers, taken as a whole, give marketers a fuller picture of how creative is truly performing to then take action, such as scaling what’s working and pausing what isn’t. ### 4. Use Budget-Friendly Creative Strategies Spending a year’s worth of budget on the holiday season isn’t possible for most businesses, especially smaller or new ones. Luckily, there are lots of options in modern digital marketing and advertising to make an impact in a cost-effective way. Repurposing strong creative, using stock images in a unique way, and user-generated or DIY-style ads can all help teams stay under budget. Use analytics to identify top-performing creative assets and see how you might reuse those. Be sure to consider the assets or possibilities you already have in-house. “These days, a selfie-video ad from a CEO or founder can be your strongest marketing campaign,” said Blatner. “Grab your smartphone and hit record. People listen to people more than companies.” Taboola’s playbook also recommends creating looping, short-form, human-driven HD motion videos to capture attention and bring products to life. And remember, seemingly simple tools available to today’s marketers are really very sophisticated. “Canva is the most affordable and full-featured creative tool you could need at the SMB level,” says Blatner. ### 5. Incorporate Personalization as an ROI Driver Personalization strategies for holiday season marketing are only limited by marketers’ imaginations. Personalized messaging can boost relevance and ad performance, with attentive users browsing in a specific time period. Ideally, marketers will have segmented their audiences already and can then explore personalization strategies for the holidays specifically. Consider including varying emotional triggers and messages based on these segments, including factors like demographics, purchase history, and known interests. Holiday personalization tactics include targeted emails with product recommendations, exclusive offers or discounts for returning customers, and, if possible, landing pages tailored to users. For e-commerce companies, quizzes or other interactive calls to action (CTAs) can lead to personalized gift guides or other assets. ### 6. Conduct A/B Testing for Optimal ROI Optimizing ROI means starting holiday efforts early and conducting useful, thorough A/B testing for creative elements for a more successful end of year. “Start earlier than you think,” advises Blatner. “Creative and copy should be live by late October, so learning phases finish before Thanksgiving.” The results of A/B testing will offer valuable information about which creative users are responding to, helping marketers make ROI-positive decisions. Depending on available resources, advertisers can test images, videos, copy, emails, social media posts, CTAs, and more. As with any good A/B tests, gauge one element at a time. ### 7. Post-Holiday Creative Learnings The holiday season starts well before December for most brands, and it continues into end-of-year and beginning-of-year timed themes. While there may be some respite once the holiday period is over, advertisers should take time to evaluate performance and have plans in place to continue the momentum. “A strong Q1 starts in Q4 and over the holidays,” says Blatner. “Maintaining strong brand awareness over the holidays is key to being considered in those Q1 opportunities, because prospects will start by exploring the brands they already know, like, and trust.” Take the time to build a library of successful holiday creative insights, and analyze performance to inform future creative strategies and campaigns, whether for next year’s holiday season or to see what can be incorporated into ongoing campaigns. ## Key Takeaways Bland, stock-looking, or otherwise tepid creative won’t get marketers and advertisers many returns in the busy holiday season. Storytelling, A/B testing, and personalization are all important components of advertising success, but even more essential for holiday and end-of-year campaigns. Marketers have to pay careful attention to every creative element to see strong ROI and other positive results once the season has ended. --- ### Building Better Financial Services Ads Creative URL: https://www.taboola.com/marketing-hub/finance-advertising-creative-playbook/ Last Modified: 2025-07-27 07:22:16 Financial services companies, from home-loan providers to digital banks, have to stand out in a crowded marketplace. That includes legacy and digital-first providers, all of whom are vying for reach and engagement across search and social media channels. The competition is intense: This year, digital ad spending in financial services is expected to reach close to $37 billion, an 11% increase year-over-year. Sentiment around digital banking has also shifted in a short amount of time: As of 2022, 78% of U.S. adults preferred to bank using a website or mobile app. Still, while much has changed, marketers and advertisers in financial services still have to focus on building and keeping user trust, meeting regulations, and educating audiences on complex topics. At Taboola, we’re constantly gathering data points and identifying trends in digital finance advertising. Our Creative Shop in-house agency builds and advises on ads informed by new and emerging trends across industries, so you can apply these learnings to your own performance marketing work. This guide to financial services advertising trends offers the latest tips on creating ads to engage and convert high-intent users. ## 8 Ways to Make Your Financial Services Ads Shine ### 1. Break Down Complexity Many financial services products are complex or may be new to users. Educating audiences and engaging with finance experts for ads can lend credibility to a brand. Consider numbered lists for explaining abstract financial concepts. ### 2. Emphasize Stability Financial services companies have to earn and keep user trust to succeed. Photos of administrative locations, like banks, often appear on publisher sites and are viewed as trustworthy. ### 3. Get the Imagery Right In addition to location photos, use first-person angles of hands holding financial objects, like credit cards, to drive interest and bring concepts to life. ### 4. Show Off Insider Knowledge In your copy, offer useful tips and tricks to maximize personal income and otherwise benefit users. This type of advice can drive higher engagement and also has the bonus effect of building brand trust by showing authority. ### 5. Build Confidence Discussing money mistakes in finance ads can show users a cautionary tale, and inform them about the potential downside of their own habits. See if you can include any data points or common issues that arise in contrast to good habits. ### 6. Consider Every Identity or Persona Financial services audiences span a huge range of identities across demographics. With personalization important to prospects and users, addressing specific audiences in headlines helps attract the most relevant clicks. ### 7. Try Movement There are lots of opportunities to try interactive ads in financial services, such as subtle motion or rich media designs. Invite users to engage with a click, hover, or swipe. ### 8. Use Ad Descriptions Data analysis found a 5% increase in the conversion rate (CVR) when using ad descriptions in financial services content, with 69% of ads in this vertical using ad descriptions. Paying attention to these types of details can boost conversions. Find more data-backed, creative-minded suggestions in the playbook. ## Key Takeaways Financial services advertising is a busy market that spans multiple specialized industries. They all share the need to engender user trust and engage high-intent users with advertising that includes useful copy and compelling images, showing financial knowledge and expertise. ## Frequently Asked Questions (FAQs) ### What are the latest trends in financial services advertising? Financial services businesses these days focus on data-driven personalization, artificial intelligence (AI) and automation, mobile-first strategies, and sticking closely to regulations. Advances in tech and consumer interest in fast, easy processes and transactions mean that finance companies are exploring new ways to serve users. ### What are some best practices in financial advertising? Financial advertising shares best practices with marketing best practices broadly, such as understanding your audience, and showing them how your products and services can solve their problems. More specific financial advertising best practices include building trust with social proof and industry certifications, using data-driven personalization, breaking down complex topics and concepts, and keeping up to date on regulations as they evolve. ### What are the 4 Ps of marketing in financial services? The 4 Ps of marketing are product, price, place, and promotion. Within financial services, product refers to the services and products a business offers to its audiences, which may include wide ranges of users. Pricing in finance marketing has to be both competitive and make a profit for the business. Place covers the multiple channels and locations where users can access your brand. Promotion is any of the many ways — social media, search, native ads, display, video, and more — that a financial services marketing team brings their products and services to prospects and customers. --- ### How Travel and Tourism Marketers Are Sailing Toward New Performance Goals URL: https://www.taboola.com/marketing-hub/travel-advertising-creative-playbook/ Last Modified: 2026-03-02 08:54:20 Travel and tourism marketing includes a wide range of audiences, many types of travel and destinations, and ever-changing market trends. Advertisers and marketers creating campaigns and ads for travel and tourism brands have to think fast and use all the available data to accurately target audiences and convert high-intent users. Marketers in travel and tourism are strategizing to reach prospects across many of the current trends: ecotourism, blended business and leisure trips, wellness tourism, and many more. In a quickly changing market, it’s essential to know your audience and get all the details right across visuals, copy, landing pages, and display ads. This means that small shifts in a headline or type of image can make a big difference. At Taboola, we analyze tons of data points to help identify the many changing trends in travel and tourism advertising. Our Creative Shop in-house agency focuses on building ads — and helping others build ads — informed by new and emerging trends, so you can apply these learnings to your own performance marketing work. Check out the creative playbook for travel to see the latest tips on creating ads that will engage and convert high-intent users, and put them to work for your brand. ## 5 Tips to Make Your Travel and Tourism Ads Shine ### 1. Encourage Daydreaming Prospective users in the travel and tourism market might start out daydreaming about a type of trip, like a 40th birthday or family reunion, before booking. Consider the ways you can engage those users, such as with photos of aspirational escapes, like a cruise ship docked or at sea. Or, show off a VIP experience, such as the top-tier cruise ship cabin, premium seating, or a private jet interior. On landing pages, use full-bleed imagery, soft gradients, and aspirational design to help users imagine their trip. ### 2. Tell a Story Travel and tourism is well suited to narrative storytelling in creative advertising across multiple channels and formats. For example, create a story around bucket list adventures for longer-term planning, or show aspirational imagery for iconic destinations. Editorial framing to build trust is also a useful tactic, such as creating top 10 lists, top must-see places, or publishing personal travel stories. ### 3. Invite Readers to Engage Travel and tourism marketing lends itself well to engaging, interactive design and copy. Using clean designs with a single message and high-resolution images will appeal to prospects, as well as always staying on brand through colors, fonts, and imagery. Encourage interaction with subtle motion or rich media designs that invite users to click, hover, or swipe. Other tactics include using quizzes or other interactive calls to action (CTAs) to engage a prospect and move them toward conversion. ### 4. Spark Curiosity Help prospective travelers envision their next trip with techniques to spark their curiosity and imagination. Include travel hacks in ads, such as hidden deals or insider tips that promise secret knowledge. Using social proof, like traveler stories or user-generated social posts or blog posts, can help make the trip easier to imagine. Referencing specific user groups that align with the target audience can also add relatability and credibility. ### 5. Get the Visuals Right Making small tweaks to visuals can lead to huge gains for travel and tourism marketers. Taboola’s creative team data found a 16% conversion rate (CVR) increase for motion ads, a 66% click-through rate (CTR) increase when avoiding text in visuals, and a 74% CVR increase when using close-up imagery. These numbers are a great reminder that paying careful attention to every detail, and conducting A/B testing, can make a huge difference for marketing performance. In addition, try visual price anchoring, putting any discounts and urgency-focused copy above the fold. ## Key Takeaways Travel and tourism marketers constantly navigate changing trends to excite and engage prospective buyers. Every detail counts in travel ad campaigns across visuals and copy, so marketers should do frequent A/B testing to see what performs best. Conduct audience research to understand traveler nuances and what kind of vacation they’re seeking. ## Frequently Asked Questions (FAQs) ### What are some popular trends in travel advertising? Ad trends include visual storytelling, incorporating user-generated content, and showing personalized experiences. In addition, emotional appeal, bucket list trips, budget travel, and work/life experiences are all popular. ### What are some examples of successful travel campaigns? Successful travel campaigns are memorable, showing the impact that unique storytelling and creativity can have on prospective travelers. They also tap into trends and user needs and desires, and help to spark curiosity or inspire wanderlust. For example, Airbnb’s “Live There” campaign shows how travelers can have deeper cultural experiences on trips. Iceland’s Icelandair stopover offer propelled the country toward greater tourism success, showing off stunning visuals and a straightforward way to get more travelers to visit. Tourism Australia’s “There’s nothing like it” campaign proved its longevity, as the group brought the slogan back after wildfires with some beautiful imagery. ### What are the key elements of effective travel ads? Ads for tourism and travel campaigns work best when they tap into emotional resonance and compelling narratives, helping prospects envision their own vacation that connects with their values and what they imagine. Effective travel ads also include visual storytelling with high-quality images, personalization, and clear CTAs. Personalization should include tailored messages for different segments of the audience to target what type of trips users might be interested in. --- ### 6 AI Marketing Trends Performance Advertisers Need to Know in 2026 URL: https://www.taboola.com/marketing-hub/ai-marketing-trends/ Last Modified: 2026-03-08 13:50:38 For digital marketers, incorporating AI into your workflow isn’t a trend, it’s a necessity — from strategy to execution to optimization and tracking, the importance of AI to your business is undeniable. Here’s what you should know for 2026. ### What’s changed in our 2026 update: - All stats and figures updated to reflect current information. - Updated information around legal transparency requirements. - New, current graphs and graphics. ## Trend 1: The Pervasive Integration of AI Across Marketing Functions Digital marketers are increasingly embracing machine intelligence, embedding it in a wide range of business functions. That includes content creation, customer analytics, and campaign optimization. To remain competitive, marketers are restructuring their workflows to prioritize AI, and the number of organizations deploying the tech in one or more business functions rose to 88% in 2025, up from 78% in 2024. As a result, teams are evolving to accommodate the new technology. Manual, time-consuming processes like creating social media posts and analyzing data can be handled by machines, freeing up humans to focus on strategic insight and creative oversight. Here are a few statistics showing how AI is infiltrating almost every area of marketing: - More than 80% of marketing teams actively use generative AI. - 93% of chief marketing officers report clear return on investment (ROI) from using GenAI. - Companies using AI in marketing report 20-30% higher ROI than traditional marketing methods. - Among marketers already using AI, 93% use it to generate content faster and 81% use it to boost brand awareness and sales. Source: McKinsey With so many tools now available, marketers can streamline everything from targeting to creative execution. Predictive analytics tools like Adobe Sensei and Google AI for ads can optimize bidding in real time, while generative solutions like Jasper, Copy.ai, and Midjourney can tackle ideation and design. Tools like HubSpot and ActiveCampaign bring AI to lead nurturing and segmentation. AI-powered performance platforms like Realize, meanwhile, can tackle everything from smart audience targeting to landing page creation to real-time ad optimization. To foster a culture of AI adoption, marketing teams should include training as they introduce new tools. It’s also important to ensure employees understand that AI will enhance, not replace, their value as part of the team. ## Trend 2: Scaling Performance with Precision (The Impact of GenAI Ad Maker) The collaboration between Columbia University and Realize highlights how generative AI is no longer just a tool for efficiency, but a primary driver of performance marketing at scale. By utilizing the GenAI Ad Maker, advertisers are able to move beyond the limitations of manual creative production, achieving results that rival or exceed traditional methods. - Human-Level Performance at Scale: The study, analyzing over 500 million impressions, proved that AI-generated ads deliver click-through rates (CTR) on par with human-made ads, with raw data even showing a slightly higher average CTR of ~0.76% for AI versus ~0.65% for humans. - Built-in Best Practices: The GenAI Ad Maker is uniquely powerful because it integrates long-standing creative best practices. For example, AI-generated ads were actually more likely to include large, clear human faces—the single most influential factor in driving engagement—than human-made ads in the same campaigns. - No Conversion Trade-off: A critical takeaway for performance marketers is that AI improves top-of-funnel engagement (CTR) without sacrificing downstream conversion quality. This disproves the common fear that AI ads only drive "curiosity clicks" or low-quality traffic. - The "Secret Sauce" of Perception: The power of the GenAI Ad Maker lies in its ability to create ads that don't "look like AI". Ads perceived as human-made achieved the highest performance of all groups, demonstrating that when AI is used to enhance human-centered cues, it sets a new ceiling for engagement. ## ## Trend 3: The Rise of Hyper-Personalization Powered by AI Personalization is the name of the marketing game in 2026. Customers expect content to be tailored to their interests, and businesses are rising to the challenge. AI-powered tools are making it easier than ever to personalize content, since businesses can gather behavioral and contextual data, which AI algorithms then use to tailor messages based on real-time signals like: - Browsing behavior. - Device type. - Geolocation. - Time of day. Predictive analytics allows marketers to deliver content before customers even know it interests them. Businesses like Netflix and Amazon are also using AI-driven recommendation engines that suggest content or products a customer is likely to enjoy. These statistics highlight the growing shift toward hyper-personalization: - 71% of customers now expect personalized interactions from brands, and 76% feel frustrated when brands fail to deliver them. - 79% of marketers use AI to personalize content and campaigns. - Personalized calls-to-action (CTAs) can convert ~202% higher than generic CTAs. - 82% of consumers are willing to share personal data in exchange for more customized experiences. - While 85% of companies say they provide personalization, only 60% of consumers agree. With increased use of data comes privacy considerations. Some consumers find ultra-targeted messaging intrusive, particularly if they feel that ads are manipulative. They may also wonder how all that collected data is being used. Brands need to prioritize transparency and user control if they want customers to see them as trustworthy. This can include: - Clear opt-in mechanisms. - Accessible privacy settings. - Easy-to-understand AI practices. Ethical personalization means not only understanding consumer privacy concerns but addressing them in your terms of service. It’s important to respect boundaries and avoid harmful biases when using the data you collect through your website visitors and ad leads. The goal is to ensure you’re offering a top-notch user experience without compromising privacy. ## Trend 4: AI-Driven Automation for Enhanced Efficiency and Productivity One of the biggest benefits of automation for marketers is that it tackles mundane, time-consuming tasks, freeing up marketers to focus on strategy and creativity. But, that’s not the only benefit of AI-driven automation. Here are a few statistics showing how marketers are using and benefiting from it: - 79% of marketers automate their customer journey, from partial to full automation. - 50% of marketers use automation daily, optimizing routine tasks and workflows. - Of marketers who use automation, 80% report increased lead generation. - 71% of marketers use automation for email marketing, making it the most automated channel. AI’s benefits go beyond tackling repetitive tasks, too. In 2026, brands are automating customer service: Using chatbots and virtual assistants, businesses can easily handle initial inquiries, answering questions that previously would have taken hours a day of a team member’s time. That allows employees to deal with more complex questions and problems. AI-powered automation can also improve ROI for your marketing efforts. Algorithms can now automatically adjust your bids, recalibrate ad spend, and shift placements in real time to optimize your efforts for each customer. Marketers also now rely heavily on AI for email marketing. AI-powered software can auto-generate subject lines, tailor delivery times, and trigger follow-ups based on customer behaviors. Once campaigns are launched, marketers now turn to AI-supported reporting that can easily pinpoint anomalies and surface opportunities almost as soon as a campaign has launched. ## Trend 5: The Evolution of AI in Content Creation and Curation When it comes to content creation, blog posts and ad copy tend to be front and center in discussions. But, AI can also create images, videos, podcasts, and more. Statistics support AI adoption across a variety of content creation functions: - Approximately 66% of people worldwide use AI regularly. - 60% of U.S. adults use AI to search for information, increasing to 74% for those under 30. - 86% of global creatives incorporate generative AI into creative workflows, and 81% say it enables them to produce work they couldn’t otherwise make. - About 20.5% of people worldwide use voice search. - AI-generated metadata boosted video view performance by up to 7.1%. Source: Associated Press Where AI really excels is in kickstarting the content creation process. AI can draft blog outlines and ad copy variations, or even provide weeks’ worth of social media posts in just a few minutes. For content curators, AI can dig through decades of videos and articles, then suggest items to repost to engage customers. AI’s role in voice search is expanding as well. As consumers increasingly search the internet using voice assistants, marketers are gathering information on common voice queries and tailoring their content to match typical phrasing. AI can even recommend keywords and automatically make metadata adjustments. Unfortunately, though, AI isn’t flawless. As many marketers have learned, the technology struggles to remain consistent over time, which can be a problem if you’re trying to stick to a particular brand voice. The technology can also output false information, making it essential to have a human review every piece of content before it goes live. ## Trend 6: Navigating the Ethical Considerations and Future of AI in Marketing In the final months of 2026, some trends are emerging that can directly impact marketers’ campaigns. Those include an increase in hyper-personalization, AI-driven visual search, and AI’s assistance in tracking down relevant influencers. As marketers fully embrace AI, though, various ethical concerns have emerged. Here are a few of the top considerations for today’s marketers: - Bias in algorithms: AI was trained on years of data, including books, websites, and photos. As a result, the technology regularly exhibits social, racial, and gender biases in the information it outputs. - Data privacy and consent: When gathering user data, it’s important to respect privacy laws, including General Data Protection Regulation in the European Union (EU) and the California Consumer Privacy Act. - Misinformation: Generative AI can be prone to something called hallucinations, which has it confidently outputting information that either doesn’t make sense or is factually inaccurate. - Intellectual property concerns: Copyrighted material was used to train AI. This brings ethical concerns about its use as marketers navigate the complexities of copyright law and fair use. - Deepfakes and synthetic media: AI-generated photos, videos, and audio have been used for fraud, causing trust issues for consumers. - Environmental impact: Generative AI consumes water and energy, leading to ethical concerns about its environmental impact. On top of ethical concerns, marketers need to navigate regulations relating to the use of generative AI. Currently, marketers in the EU must pay close attention to the EU AI Act, which has entered phased enforcement and expands disclosure, transparency, and risk-classification requirements for AI systems. In the U.S., AI regulation remains fragmented, with a growing number of state-level transparency and consumer protection laws influencing how AI can be used in marketing. Organizations now report they’re actively mitigating generative AI-related risks, with inaccuracy, cybersecurity, and intellectual property infringement being the top issues. - 75% of organizations now have some AI usage policies in place, but only 36% have a formalized governance framework. - 3 in 5 businesses are now monitoring AI systems for fairness, bias, and transparency. - Of the U.S.’s public companies, 36% now disclose AI as a separate risk factor in SEC filings. Source: Harvard Law School Forum on Corporate Governance For marketers, these changes mean a likely future shift in job duties. As AI automates time-consuming processes like pulling reports and segmentation, professionals will gradually shift to strategic oversight and ethical governance. Due to this shift, marketers are being called upon to develop new skills like prompt engineering and data literacy. To ensure responsible AI use, brands should prioritize transparency, accountability, and consent. Establishing internal processes for checking AI content for inaccuracies and bias can help retain customer trust in your brand. ## Key Takeaways AI has made its way into almost every facet of marketing, from ideation to execution and analysis. As these tools become more prevalent, ethical concerns are emerging, including consumer privacy. While AI excels at generating content, human intervention is needed to maintain quality and reduce the risk of costly errors. In 2026, AI is no longer optional — it’s a baseline expectation: Brands need to implement the technology responsibly to remain competitive. ## Frequently Asked Questions (FAQs) ### What are the most impactful AI applications for digital advertisers in 2026? In 2026, digital advertisers are seeing AI’s impact in four key areas: - Predictive audience targeting. - Real-time bid optimization. - Dynamic content personalization. - Automated performance reporting. Advertisers who use predictive AI to anticipate user intent are outperforming traditional advertisers who still rely on targeting users by demographics. Meanwhile, marketers are using AI to test dozens of ad variations in seconds, allowing them to identify the ads most likely to get ROI while adapting creative and spend in near real time. ### How can small businesses leverage AI in their marketing efforts? Small businesses can benefit from AI as much as enterprise-level brands. Chatbots and virtual assistants can be great entry points, helping automate efforts in small ways before expanding to more intensive applications. Small biz marketers can also use AI to repurpose content across multiple platforms, personalize email campaigns, and target ads based on intent signals rather than location or interests. Look for affordable tools that provide drag-and-drop interfaces and guided workflows to minimize the learning curve and reduce dependency on in-house technical expertise. ### What are some common misconceptions about AI in marketing? One of the biggest misconceptions about AI is that it will put marketers out of work. Over the years, automation has proven to serve as more of an assistant, taking care of time-consuming, mundane tasks and shifting professionals to higher-level responsibilities. Another misconception is that AI is infallible and impartial. In truth, AI was trained on human-created data, which means it requires human oversight to ensure its output is both accurate and fair. Lastly, some assume they can simply set up an automation and leave it, letting AI do all the work. Unfortunately, the technology needs ongoing oversight and evaluation to ensure it’s functioning properly as data, algorithms, and regulatory expectations evolve. ### How can marketers upskill to work effectively with AI tools? To excel in this new, AI-driven environment, marketers should focus on building foundational knowledge in data literacy, analytics, and prompt engineering. As AI begins to play a larger role in content creation, marketers can upskill by learning to evaluate its output for tone, quality, and bias — all skills that will become increasingly important. To further solidify their value, marketers are seeking certifications in popular AI platforms and attending workshops. ### What are the key metrics for measuring the success of AI-powered marketing initiatives? As with any marketing effort, it’s vital to measure your efforts. For AI-powered initiatives, here are a few metrics to track: - ROI improvement: Are your efforts bringing an increase in revenue without upping your spend? - Acquisition cost reduction: Has AI helped you bring in new customers with reduced effort and cost? - Click-through-rate uplift: Are more customers clicking on your ads and calls to action? - Conversions: How many clicks turn into actual sales? - Time savings: Has AI saved time for you and/or your team? - Compliance and transparency rates: When conducting compliance audits, is your AI output fair, unbiased, and compliant with any emerging regulations? --- ### Three Reasons You Should Refer An Advertiser to Realize URL: https://www.taboola.com/marketing-hub/three-reasons-you-should-refer-an-advertiser-to-taboola/ Last Modified: 2026-07-14 07:01:21 Referral programs are a type of word-of-mouth marketing tactic that help cut through the noise of today’s digital markets. B2B referral programs in particular can help a company generate new business, bringing in highly qualified leads at a low cost. The Realize referral program does double duty for efficiency: It’s easy for anyone to cut through the sales process and refer a fellow advertiser, and advertisers can then use Taboola's Realize platform for smarter performance marketing. Anyone can refer an advertiser to Realize, even if they’re not a customer or a digital marketer. It’s an accessible, inclusive program, aimed to help any business growing their acquisition efforts. Whether it’s a friend starting a side hustle or a company you’ve worked with expanding their product line, you likely know some candidates for the program. Here’s what else to know about referring advertisers to Realize. ## 3 Reasons You Should Refer An Advertiser to Realize ### 1. It Actually Cuts Through the Noise User fatigue, creative fatigue, and the high cost of advertising have all converged in the past few years, leaving many businesses struggling to find and convert prospects. Realize employs several strategies to help its advertisers succeed: It provides advanced targeting options to users, for example, like demographics, interests, contextual relevance, and user behavior, so ads reach the right audience and avoid wasting time. Realize also looks for intent-driven audiences, finding the high-intent users to nurture and eventually convert. It saves wasted time marketing to the wrong audiences. Its embedded artificial intelligence (AI) and machine learning (ML) capabilities optimize campaign performance over time, too, including the creative elements of ads, which have to be engaging and carefully crafted to cut through ad fatigue. ### 2. It’s Useful for a Variety of Industries Realize platform helps brands across industries, with some companies in particular seeing a lot of success — e-commerce and direct-to-consumer (D2C), health and wellness, finance and fintech, home improvement and gardening, travel and tourism, and more. Businesses in lead generation, affiliation, marketing, and tracking and measurement also use Realize for true performance marketing, as do media experts. No matter the industry, marketers are struggling to break through the noise. When thinking about who to refer, consider who could benefit from a performance marketing tool, and you can probably think of advertisers from a variety of businesses, whether they’re clients, colleagues, or any other industry contact. Small and medium-sized businesses especially are looking for cost-effective ways to find the right users. Consider founders, marketers launching new products or opening new markets, or diversifying their acquisition channels. ### 3. It’s Easy and Rewarding (for You and Your Referral) The Realize referral system is designed to be straightforward and inclusive for lots of referrers, with a reward structure that makes it worth taking the time to participate. Setting up Realize is easy — a new user just needs to place the pixel or Realize server-to-server (S2S) on its landing pages. For you, the referrer, the reward for referring that user (after they’ve spent $5,000 within 45 days) is $1,000. The lead referral form is straightforward and quick to fill out, so send along as many potential advertisers as you like. From there, it’s just waiting for lead qualification and providing some basic details so we can send your bonus. ## Key Takeaways If you’re working in advertising or marketing, you know how much chatter and competition exists today. Referring prospects into Realize platform helps connect advertisers and performance marketers with tools that can help them succeed under budget. Those referring get a cash reward, and those using Realize can cut through the noise and target users the right way. ## Frequently Asked Questions (FAQs) ### How does Realize work? Performance-marketing platform Realize uses AI and proprietary data to optimize ad campaigns and target high-intent users. Realize can analyze content to identify promising audiences and match ads across channels with the right users. Advertisers and marketers using Realize can also take advantage of optimized creative across display ads, motion ads, video ads, and other formats to mitigate the diminishing returns of an expensive, crowded social and search market. Users can also track performance metrics, customize reports, and optimize campaigns within the platform. ### What are the requirements for a qualified advertiser referral? The Realize referral program requires that a referred advertiser is first qualified as a lead, then that the advertiser signs up and spends $5,000 on the platform within 45 days. The person referring the new customer provides basic information to get their payment via bank transfer. ### What is a performance advertising platform? A performance advertising platform is a digital marketing tool that enables advertisers to pay only when a specific action is taken by a user, such as a click, lead, or sale. These platforms focus on delivering measurable results and maximizing return on investment (ROI) by tracking campaign performance and optimizing ad spend based on reliable data. --- ### Real Estate Marketing Trends 2026: What Advertisers Need to Know to Stay Competitive URL: https://www.taboola.com/marketing-hub/real-estate-marketing-trends/ Last Modified: 2026-03-16 13:19:39 From immersive visual content to analytics-driven tools, real estate professionals need to fully understand the importance of monitoring trends in 2026. Here, I’ll run down five of the most powerful trends shaping the future of real estate marketing. What’s changed in our 2026 update: - All entries include updated and current information and advice. - All stats and figures updated with new and current information. - Two new graphs/graphics added. ## Trend 1: The Power of Visual Content and Immersive Experiences in Real Estate Visuals have always been an important part of selling a home. Buyers want to see photos of every aspect of a property, from the front yard to the tiny guest bathroom on the main floor. In 2026, brokers are leaning into the power of video, AI, and human content to sell homes, including virtual staging, a tool that continues to spark debate. Here are a few essential statistics that showcase the power of visual content in 2026: - Video dominates in 2026, with video listings boosting inquiries by 403%. - Homes/listings with professional photos sell 32% faster, spending an average of 89 days on the market as compared to 123 days. - Buyers spend about 60% of their time looking at photos, versus only 20% reading property descriptions. - 60% of buyers’ agents say staging has an effect on some buyers, but not always. Source: RubyHome Luxury Real Estate Tools like virtual tours and 360-degree interactive experiences allow buyers to view listings from wherever they are. This is especially beneficial for out-of-state buyers who can’t travel to look at every property that goes on the market. Augmented reality is still relatively new to the industry, but the future possibilities are endless: Imagine buyers being able to stand in a living room and virtually incorporate their existing furniture, or change paint colors. Visual content goes beyond listings, though. In 2026, competitive real estate professionals repurpose assets across platforms, including on social media. By taking advantage of the popularity of video on social media, agents can reach buyers before they’re browsing multiple listing service (MLS) listings. ## Trend 2: Hyperlocal Targeting and Personalized Communication in Real Estate Authenticity and relevance are more than marketing buzzwords. Across all industries, marketers have realized the value of personalizing their messaging, and real estate is no different. In 2026, the industry has found that consumers respond to agents who understand their neighborhood: From local school ratings to the best coffee shops, agents who position themselves as community experts build stronger, more lasting client relationships. In 2026, hyperlocal marketing has become a game-changer. Search trends back this up: Search queries using keyphrases like “real estate agent near me” and “houses for sale near me” are more popular than ever, driving a need for industry professionals to claim their Google Business Profile (GBP) and focus on local marketing. Personalization continues to be key for real estate agents wanting to stand out. Digital advertisers can help real estate marketers create ads that target: - First-time homebuyers. - Property investors. - Sellers/buyers in a specific zip code. - Buyers who are actively looking. - Buyers within a certain demographic. - Buyers with certain interests. Here are some top trends to help inform your marketing strategies in 2026: - In 2025, 66% of sellers worked with an agent referred to them, or one they’d used before. - Reputation remains a key factor in choosing an agent. 35% of sellers say they considered an agent’s reputation when making their choice. - REALTORS® choose text messaging as their top communication channel, with 94% using it, followed closely by phone and email. - Searches for “real estate agent near me” have increased 41-fold since 2015, with approximately 14,000 searches per month. It should be noted, though, that privacy concerns have impacted the way real estate professionals target. The National Association of REALTORS® has strict standards for data collection and use, and that information can be valuable for anyone working in the industry. ## Trend 3: Leveraging Social Media and Building Online Presence for Real Estate Like other marketers, real estate agents know that social media is the key to reaching today’s homebuyers. Here are some key statistics related to building an online presence in 2026: - 82% of real estate businesses market through social media. - Facebook remains the dominant platform for real estate agents, used by about 90% of agents, followed by Instagram (52%) and LinkedIn (48%). - Around 63% of real estate agents use video content in their social media marketing strategy. - Video content on social media is shared about 12x more than text and image-only posts. - According to tech usage data, about 75% of agents use social media to build their brand and connect with clients online. Today’s real estate professionals go beyond promoting listings by posting engaging content like local community highlights, the stories behind homes they’ve listed, and behind-the-scenes glimpses of their daily lives. While Facebook and Instagram remain important, agents are also optimizing content for YouTube Shorts and TikTok search, treating these platforms as mini search engines. Video is the real standout this year. The industry now has the opportunity to engage both buyers and sellers without investing heavily in advertising: Short-form video content is more popular than ever, giving agents the opportunity to share quick tours, highlight a home’s key features, or walk through the closing process. Live video is one of the most popular social media trends impacting the industry. Videos don’t even have to be polished: One New York City real estate agent shares walk-throughs shot using his phone to impressive success. Agents can also share live videos showcasing their latest listings or responding to typical questions to engage potential customers. ## Trend 4: The Importance of SEO and Local Search for Real Estate Discovery In 2026, home shopping starts with a search, but that’s nothing new. One thing that has changed, though, is the way people search. Instead of typing in keywords like “new homes for sale,” today’s savvy online shopper incorporates details like towns or neighborhoods. That’s where local search comes in. - 46% of all Google searches have local intent. - Google’s Local Pack drives 44% of local search clicks. - Organic search dominates clicks, capturing roughly 94% of total search (versus paid). - The top organic search result typically earns about 27.6% of all clicks, and click share drops sharply further down the page. - Over 50% of search queries include four or more words, indicating long-tail keywords continue to be important. Source: Digital Silk So what can a real estate agent do to reach today’s home shoppers? Localization is key. Long-tail keywords incorporating community information can go far in reaching buyers, as can maintaining consistent NAP (name, address, and phone number) across the web. But, one of the most important things a real estate professional can do is to claim their GBP, since those listings get priority in search results. Reviews are another important SEO aspect. Google weighs reviews heavily when prioritizing businesses in local search, so encouraging reviews can make a big difference. If you can generate a steady stream of positive reviews, you’ll keep the algorithms happy. ## Trend 5: Data Analytics and CRM for Effective Real Estate Lead Management It’s no longer enough to generate leads: To remain competitive, you’ll need to effectively manage those leads. Customer relationship management systems (CRMs) make it easy to track the origination of each lead, how those leads engage with your content, and what actions they’ve taken. Agents can then automate follow-ups and segment contacts to stay in front of potential clients with minimal effort. Once leads have made their way into your ecosystem, you can use data analytics tools to measure the effectiveness of every dollar you spend to bring them there. Urchin Tracking Modules (UTMs) and CRM integrations can help you follow a lead from the initial click to closing. Artificial intelligence is also transforming the way real estate professionals market. AI-powered tools can analyze past buyer behavior and use that information to predict future actions. This allows professionals to prioritize the leads most likely to convert. The industry is also learning that personalizing the user experience can get better results. - Real estate agents who use a CRM system see a 41% increase in lead conversions. - 54% of agents say that AI tools help with identifying high-quality leads. - REALTORS® report social media as the top source of quality leads, at 46%, followed by CRM at 23%. - 90% of marketing professionals use AI to streamline customer interactions. Source: National Association of REALTORS® As you cultivate and track leads, here are some important data points to get you started: - Search behavior and keywords used: What search terms led a high-converting lead to your site? This data point is useful for gathering information on your target customers, and that information can be used for future advertising and SEO campaigns. - Time spent on listing pages: When a customer hangs out on your listing pages for a while, it can be a sign of an interest in a particular location, home layout, or price range. If visitors seem to engage heavily with certain types of content, such as your virtual tours or videos, you can use that data to inform future efforts. - Lead source attribution: One of the best things about analytics is its ability to track the origin of incoming leads. Over time, this gives you a picture of the marketing efforts that pay off most. - Click-through and bounce rates: If people are coming to your listings and quickly leaving, it could be an indication that something’s off. In many cases, it simply means your ads are hitting the wrong targets, but it can also signal that your visual assets need improvement. - Mobile vs. desktop usage patterns: More homebuyers than ever are conducting their search on mobile. Still, it’s important to look at the habits of your own customer base. ## Key Takeaways Today’s real estate professionals are leaning in heavily to assets like immersive visuals and video tours to boost engagement and listing velocity. But, it’s also important to localize your marketing efforts to adapt to the way consumers search for homes in 2026. Social media is another way to reach potential clients, with short-form, social-native video serving as the most popular format. Real estate professionals should also consider CRMs for improved lead generation and AI-assisted analytics to measure return on investment (ROI). ## Frequently Asked Questions (FAQs) ### What are the most effective digital advertising channels for real estate in 2026? In 2026, real estate professionals are finding success by combining strong visual appeal with precision targeting. While social media platforms like Facebook and Instagram remain powerful, thanks to their robust user base and location-based targeting, the continued dominance of short-form, social-native video has made platforms like TikTok and YouTube Shorts increasingly popular, as well. For intent-driven targeting, Google Ads and Google Local Inventory are still the top contenders for marketing spend. Today’s serious homebuyers and sellers search using terms like “homes for sale in ” or “real estate agents near me.” Agents can capture clients with retargeting ads, video-based ads, and local-first placements to stay in front of clients throughout their house-hunting journey. ### How much should a real estate agent spend on digital marketing? Marketing budgets can vary dramatically from one real estate agent to another, but generally speaking, the average agent only allocates 10-20% of their budget on marketing. Digital is a large part of that spend, but many agents also rely heavily on free options like social media marketing. A typical monthly marketing budget might look something like this: - Paid social ads: $300-$1,000. - Google Ads: $200-$800. - Video content and editing: $150-$500. - Marketing tools: $50-$300. Marketing spend is only one part of the picture, though, and it’s important for agents to closely track their efforts to maximize ROI. Track cost per lead, time to close, and lead quality, in addition to conversion rates, to ensure you’re putting your time and money toward efforts that get results. ### What are some common digital marketing mistakes real estate professionals should avoid? Real estate agents have access to more tools than ever to power their digital marketing efforts. But, those tools need savvy marketers to come up with sound strategies and execute them. Whether you’re starting a new campaign or identifying issues with your current marketing efforts, it’s important to watch for some things that can go wrong. - Using inadequate or outdated visuals: Listings with professional-quality photos and videos tend to get better results. Today’s home shoppers expect immersive visuals that give them a clear picture of the property. - Failing to claim or optimize a Google Business Profile: In just a few minutes, at no cost whatsoever, you can claim your Google Business Profile, which will help you appear in local search results and on Google Maps. - Spreading efforts too thin across platforms: It can be tempting to have a presence on every social media platform available, but it’s best to channel all your efforts into a select few platforms that best represent your target audience. - Not tracking performance: Without tracking, it’s nearly impossible to know what’s working. Today’s tools make it easy to measure everything from clicks to conversions. - Ignoring online reviews: When searching for an agent, consumers tend to go with recommendations from others. Online reviews can drive buyers and sellers to your business, so it’s important to find ways to encourage them and respond to them. ### How can real estate professionals build trust online? Trust starts with user experience. Make sure your websites and listings are accessible whether customers are on a mobile device or desktop. You can also build trust by maintaining an active, authentic social media presence and a Google Business Profile with updated contact information. Responsiveness is another key way to show your trustworthiness. Reply to comments and answer any inquiries promptly to not only build trust, but also show clients you’re reliable. Video content is a great tool for trust building. Even something as small as showing your face while walking through a neighborhood or discussing new listings can humanize your brand. The old adage that people do business with people they like definitely applies to real estate. By being accessible, you’ll compel clients to bring you along on their homebuying journey. ### What are the key metrics for measuring success in real estate digital marketing? The best metrics depend on your goals. If you’re hoping to measure success in your digital marketing efforts, here are some good metrics to follow: - Engagement. - Lead generation. - Click-through rates. - Video view duration and completion rate. As customers move through the funnel, here are some metrics that can be useful: - Time spent on site. - Bounce rates. - Listing saves and shares. When measuring how your new leads are doing, take a look at the following: - Inquiries per listing. - Cost per lead. - How leads move through your CRM pipeline. Lastly, it’s important to track conversions. Are those leads turning into clients? The following are the ultimate indicators of success in your marketing efforts: - Appointment bookings. - Showings. - Closed deals. --- ### 5 Benefits of Display Advertising, and How It Can Deliver for Your Brand URL: https://www.taboola.com/marketing-hub/display-advertising-benefits/ Last Modified: 2025-10-26 17:34:41 Display advertising isn’t just about putting pixels on a page, it’s about getting your message in front of the right people, in the right place, at the right time. No longer simply reliant on static banners, display advertising involves making smart placements, delivering responsive design, and achieving measurable results. Whether you're a performance marketer driving conversions or a brand strategist seeking awareness, display advertising remains a crucial component of the marketing puzzle. Below, I’ll explore five impactful benefits display ads bring to the table. ## Benefits of Display Advertising for Marketers ### 1. Increased Brand Visibility Display advertising helps boost brand visibility on the open web, often without disrupting the user experience. Unlike traditional pop-ups or banner ads that users tend to ignore or block, dedicated performance advertising platforms let you embed ads that not only reach relevant audiences where they are but capture their attention with a diversity of placements and styles, including eye-catching motion ads. #### How to Measure Impact: - Impressions: The number of times your ad appears on a web page. - Reach: The total number of unique users who have seen your ad. - Viewability rate: The percentage of ads actually seen by users (not just served). #### Example: Kavak, a leading online platform for buying and selling used cars in Latin America, improved its brand visibility and boosted conversions beyond what it was achieving on Facebook, Google, and TikTok by using a combination of image and motion ads, reaching new audiences across premium publisher sites. Kavak found that motion ads — featuring short, looping animations — delivered stronger performance than static formats, leading them to scale this format using Realize’s in-house Motion Ads Studio. To deepen engagement, they installed the Taboola Pixel to track user behavior and build high-intent retargeting audiences. This strategy led to a 30% increase in programmatic conversions from Q3 to Q4 and more than 500 million impressions, significantly amplifying their brand presence and campaign efficiency. ### 2. Improved Audience Targeting The best modern display platforms are data engines. With audience segmentation, behavioral signals, contextual targeting, and retargeting, they let you pinpoint high-intent users with remarkable accuracy, using real-time intent signals and interest-based audience clusters to deliver content to those most likely to engage. Display doesn’t just show your message to a crowd — it finds the right person at the right moment. #### How It’s Measured: - Click-through rate (CTR): The ratio of clicks to impressions. - Engagement time: How long users stay on your content after clicking. - Conversion rate: The percentage of users who complete a desired action (such as a purchase, signup, or download). #### Example: Ziwo, a cloud contact center provider in the GCC region, turned to Realize to generate high-quality B2B leads across the UAE, Saudi Arabia, and Morocco. Using Realize’s Native Image Ads, premium publisher inventory (including Yahoo!), and first- and third-party audience segments, the campaign reached 33.5 million people across the open web, with 40% more leads versus other platforms, and a 30% lower CPA compared to other channels. Ziwo’s success demonstrates how advertising, when coupled with AI-driven segmentation and intent-based targeting, can benefit companies attempting to improve audience targeting efforts. ### 3. Cost-Effectiveness Display advertising can offer a strong return on investment (ROI) when paired with precise targeting and smart bidding. Flexible bidding models and granular budget control make it easier for marketers to optimize performance at every funnel stage. For instance, performance platforms like Realize can leverage AI-powered algorithms to automatically adjust bids based on real-time signals (e.g., user behavior, page context, and likelihood of engagement). #### How It’s Measured: - Cost-per-click (CPC): What you pay for each click. - Cost-per-acquisition (CPA): Total ad spend divided by the number of conversions. - Return on Ad Spend (ROAS): Revenue generated per dollar spent on advertising. #### Example: Peugeot, in collaboration with Publicis Media, aimed to enhance their campaign's efficiency by reducing acquisition costs. Utilizing Taboola's retargeting capabilities and the Taboola Pixel, they tracked user interactions and optimized their campaigns using the Maximize Conversions bidding strategy. This approach led to a 14% lower cost-per-acquisition (CPA) compared to their other targeting campaigns. Furthermore, the campaign achieved a 2.45x lower cost-per-lead (CPL) than the brand’s benchmark expectations, demonstrating the cost-effectiveness of their efforts. ### 4. High Engagement As mentioned earlier, display advertising has grown far beyond static images. Today’s formats include responsive carousel ads, interactive video, and immersive content previews. These rich creatives not only capture attention but also drive meaningful engagement. Realize’s responsive ad formats, for example, adapt to both screen size and context, helping ensure higher interaction rates — especially on mobile. #### How It’s Measured: - Engagement rate: Interactions — like swipes, clicks, or hovers — divided by impressions. - Video completion rate (VCR): Percentage of users who finish watching your video. - Dwell time: Time spent actively engaging with the creative. #### Example: HPE Automotores launched a programmatic campaign using Realize-powered video ads across premium publisher sites. As a result, their CTR tripled and their engagement rate nearly doubled (up 46%) compared to other channels. This shows that modern display tools can lead to mindful campaigns that create sustained engagement. ### 5. Scalable Optimization Through AI Automation Display advertising today isn’t just about placement: It’s about scaling and refining content in real time. AI-driven tools now enable advertisers to auto-generate ad assets, conduct live A/B testing, and continuously optimize campaigns based on real-time engagement and performance signals. #### How It’s Measured: - Click-through rate: Tracks how well AI-optimized creatives perform. - Conversion rate: Measures the effectiveness of real-time testing and targeting. - Creative refresh rate: Frequency of updates to assets to combat ad fatigue. #### Example Using AI-powered Predictive Audiences within Realize, brands like NerdWallet, The Motley Fool, and QuinStreet are unlocking up to 270% increases in conversions while increasing spending by just 40% more year over year. Further, they did this without major increases to their cost-per-acquisition. This approach combines first-party conversion data with Realize’s behavioral insights to automatically identify and target high-converting users, leading to powerful performance gains. ## Key Takeaways Campaigns that use behavioral signals and intent-based targeting consistently outperform broader approaches. Rich, interactive formats deepen engagement, and ads that match their context build trust, reduce bounce rates, and improve on-site performance. ## Frequently Asked Questions (FAQs) ### What are the advantages and disadvantages of display advertising? Display advertising often offers marketers the following benefits: - Reaches large and diverse audiences. - Enables precise targeting. - Offers real-time performance data. - Supports various ad formats (image, video, carousel, etc.). - Cost-effective, especially for advertisers with a small budget. Despite the benefits of display advertising, there are drawbacks. As you consider this type of marketing, keep the following in mind: - Banner blindness or ad fatigue is common. - Ad blockers may reduce visibility. - Requires regular testing and optimization to stay effective. ### What are the best practices of display advertising? A successful display advertising campaign implements the following best practices: - Avoid a one-size-fits-all approach, opting for a segmented approach based on audience needs and their place in the funnel. - Use strong visual elements that attract and engage readers. - Leverage compelling copy that targets your audience. - Use A/B testing to hone efforts. - Create cohesion across all channels to promote brand awareness and improve the user experience. --- ### Three Reasons You Should Refer An Advertiser to Realize URL: https://www.taboola.com/marketing-hub/three-reasons-you-should-refer-an-advertiser-to-realize/ Last Modified: 2026-07-14 07:03:35 Referral programs are a type of word-of-mouth marketing tactic that help cut through the noise of today’s digital markets. B2B referral programs in particular can help a company generate new business, bringing in highly qualified leads at a low cost. The Realize referral program does double duty for efficiency: It’s easy for anyone to cut through the sales process and refer a fellow advertiser, and advertisers can then use Taboola's Realize platform for smarter performance marketing. Anyone can refer an advertiser to Realize, even if they’re not a customer or a digital marketer. It’s an accessible, inclusive program, aimed to help any business growing their acquisition efforts. Whether it’s a friend starting a side hustle or a company you’ve worked with expanding their product line, you likely know some candidates for the program. Here’s what else to know about referring advertisers to Realize. ## 3 Reasons You Should Refer An Advertiser to Realize ### 1. It Actually Cuts Through the Noise User fatigue, creative fatigue, and the high cost of advertising have all converged in the past few years, leaving many businesses struggling to find and convert prospects. Realize employs several strategies to help its advertisers succeed: It provides advanced targeting options to users, for example, like demographics, interests, contextual relevance, and user behavior, so ads reach the right audience and avoid wasting time. Realize also looks for intent-driven audiences, finding the high-intent users to nurture and eventually convert. It saves wasted time marketing to the wrong audiences. Its embedded artificial intelligence (AI) and machine learning (ML) capabilities optimize campaign performance over time, too, including the creative elements of ads, which have to be engaging and carefully crafted to cut through ad fatigue. ### 2. It’s Useful for a Variety of Industries Realize platform helps brands across industries, with some companies in particular seeing a lot of success — e-commerce and direct-to-consumer (D2C), health and wellness, finance and fintech, home improvement and gardening, travel and tourism, and more. Businesses in lead generation, affiliation, marketing, and tracking and measurement also use Realize for true performance marketing, as do media experts. No matter the industry, marketers are struggling to break through the noise. When thinking about who to refer, consider who could benefit from a performance marketing tool, and you can probably think of advertisers from a variety of businesses, whether they’re clients, colleagues, or any other industry contact. Small and medium-sized businesses especially are looking for cost-effective ways to find the right users. Consider founders, marketers launching new products or opening new markets, or diversifying their acquisition channels. ### 3. It’s Easy and Rewarding (for You and Your Referral) The Realize referral system is designed to be straightforward and inclusive for lots of referrers, with a reward structure that makes it worth taking the time to participate. Setting up Realize is easy — a new user just needs to place the pixel or Realize server-to-server (S2S) on its landing pages. For you, the referrer, the reward for referring that user (after they’ve spent $5,000 within 45 days) is $1,000. The lead referral form is straightforward and quick to fill out, so send along as many potential advertisers as you like. From there, it’s just waiting for lead qualification and providing some basic details so we can send your bonus. ## Key Takeaways If you’re working in advertising or marketing, you know how much chatter and competition exists today. Referring prospects into Realize platform helps connect advertisers and performance marketers with tools that can help them succeed under budget. Those referring get a cash reward, and those using Realize can cut through the noise and target users the right way. ## Frequently Asked Questions (FAQs) ### How does Realize work? Performance-marketing platform Realize uses AI and proprietary data to optimize ad campaigns and target high-intent users. Realize can analyze content to identify promising audiences and match ads across channels with the right users. Advertisers and marketers using Realize can also take advantage of optimized creative across display ads, motion ads, video ads, and other formats to mitigate the diminishing returns of an expensive, crowded social and search market. Users can also track performance metrics, customize reports, and optimize campaigns within the platform. ### What are the requirements for a qualified advertiser referral? The Realize referral program requires that a referred advertiser is first qualified as a lead, then that the advertiser signs up and spends $5,000 on the platform within 45 days. The person referring the new customer provides basic information to get their payment via bank transfer. ### What is a performance advertising platform? A performance advertising platform is a digital marketing tool that enables advertisers to pay only when a specific action is taken by a user, such as a click, lead, or sale. These platforms focus on delivering measurable results and maximizing return on investment (ROI) by tracking campaign performance and optimizing ad spend based on reliable data. --- ### Moving Automotive Marketing Forward: What’s Working in Auto Ads URL: https://www.taboola.com/marketing-hub/auto-advertising-creative-playbook/ Last Modified: 2025-08-11 06:59:11 The automotive industry has gone digital, with 95% of car shoppers these days relying on online resources to do their research. The automotive industry still counts on in-person experiences, though, since most shoppers will go to a dealership to purchase a car. For advertisers and performance marketers, this means that a positive buyer journey is essential. In addition, only one in three potential car buyers know exactly what they want to buy, so marketers have a lot of potential influence on those undecided shoppers. Automotive marketing strategies generally include data-driven personalization, use of social and search, and strong creative to cut through user fatigue. At Taboola, our Creative Shop in-house agency uses huge amounts of data from our Realize platform to identify trends, then build better creative accordingly and advise others. For the automotive industry, the team has uncovered best practices across copy, visuals, landing pages, and display ads. These include details on what type of imagery performs best, which headline topics are popular, and how to improve landing pages, all tailored for automotive marketers. Get all the details yourself in the Automotive Creative Playbook: ## 6 Tips to Speed Up Your Automotive Marketing Performance These are some of the essential pieces of a successful automotive marketing campaign, based on the latest trends. ### 1. Tell a Story Successful car brands go much farther than just showing their products in photos or videos — they tell a compelling story that resonates with prospects and their lifestyle. That might include connecting the brand to travel, off-road adventure, or city life, in addition to highlighting a brand’s legacy to tap into nostalgia, if relevant. ### 2. Experiment With Photos and Copy For automotive marketing in particular, Realize’s Creative Shop found that including a person in visuals leads to a 4% increase in conversion rate (CVR), and that avoiding text in visuals leads to an increase of 11% CVR. Test variations of your imagery to see if you can capture these gains as well. ### 3. Get Every Detail Right Automotive marketers should make sure every image is current for savvy online users. Keep it natural, too, infusing authenticity with candid shots, real-life settings, and natural lighting. Strong colors can also help ads stand out to prospects. ### 4. Make Engagement Easy Try some fun ways to compel users to learn more, such as with interactive, gamified, or quiz-based calls to action like “Find Your Perfect Model” for auto shoppers. See how you might incorporate immersive visuals, like large, high-quality hero images and videos. ### 5. Put the Car Front and Center Car shoppers like to see luxury vehicle interiors, so include close-up shots of premium materials, sleek, modern dashboards, or other advanced features. Make sure to capture unique or new features (particularly when marketing electric vehicles), and incorporate movement wherever possible. ### 6. Encourage Interaction For display ads, consider how you can bring audiences in with subtle motion or rich media designs, so users can engage with a click, hover, or swipe. In general, use elements of surprise and exclusivity in ads to elicit reader curiosity and urgency. ## Key Takeaways Automotive brands require top-notch creative advertising, with visuals and copy that can engage car shoppers throughout their buying journeys. Marketers and advertisers can take advantage of trends data that shows which ad details prospects respond to best, incorporating those tips into their own campaigns. ## Frequently Asked Questions (FAQs) ### What are the best practices for ad creative in automotive campaigns? While ad creative for automotive campaigns follows many of the best practices of marketing generally, there are some key areas of focus. High-quality imagery is essential for automotive ads. Visuals should be appealing, with the correct aspect ratio, and show cars in motion as well as their premium features. In addition, encourage click-through rate and conversions with special offers, particularly local ones, and make sure the mobile experience is seamless. ### What are some effective creative strategies for automotive video ads? Video is a great fit for automotive, since there’s so much movement to capture, and marketers can create successful videos by telling a human story. Automotive purchases are frequently driven by emotion, nostalgia, or a sense of the brand connecting with a driver’s lifestyle, so marketers can create engaging storylines. In addition, marketers can use interactive video features for more engagement and personalization, while AR (augmented reality) and VR (virtual reality) also work well for automotive. ### What makes a successful automotive creative ad campaign? So many factors go into a successful automotive creative ad campaign, and it starts with knowing the intended audience and the campaign’s goals. Use a multi-channel approach across digital, social, and more traditional options where appropriate. Tell a creative, unique story about the automotive brand you’re promoting to connect with prospects, and include an emotional connection where it makes sense. Finally, it’s essential to show off key features and benefits so users are fully informed as they’re researching vehicles. --- ### 4 Benefits of Referral Programs: How They Connect Brands and New Users URL: https://www.taboola.com/marketing-hub/referral-programs-benefits/ Last Modified: 2025-07-15 08:52:48 In an era of constant messaging and ads from multiple brands, referral programs are all about who you know. It’s a refreshing change from trying to squeeze more leads and buyers from an increasingly fatigued base of web users, and increasingly crowded social and search channels. Referral programs are an inexpensive, effective type of marketing, and for B2B companies in particular, they can bring in highly qualified leads more easily and naturally than many other marketing programs. Those referring in new leads may have worked closely with that lead, bringing an understanding of why the product will be helpful. Referral programs work best as a type of acquisition tactic. Well-run, established referral programs can help boost revenue and bring other benefits for advertising, e-commerce, and other types of companies. ## 4 Benefits of Referral Programs ### 1. Cost-Effectiveness By any measure, marketing and sales costs only keep growing as channels and mediums proliferate. One survey found that customer acquisition costs (CAC) rose 222% between 2013 and 2022. In addition, advertising costs keep rising for businesses — especially vexing for small and medium-sized businesses — and tactics like bidding on popular keywords can make some ads unaffordable. The phasing out of third-party cookies also added a big challenge for advertisers, who have had to start from scratch with gathering first-party data to enable better targeting. With all these challenges, a referral program can bring in highly qualified leads who are already familiar with your brand. It’s a more personal, less digital way to connect with clients, former colleagues, and others who can benefit from a brand, and it’s a much shorter sales cycle or funnel journey. When a lead starts out as a referral, it tends to lead toward a lower churn rate — 18% lower than from other marketing channels, according to Wharton School data. That saves money for the business, as does the lowered customer acquisition cost. ### 2. Better Outcomes While brands save money implementing and using referral programs, they also see better outcomes across many typical marketing and sales metrics. Those hard numbers are in addition to improvements in trust, enthusiasm, and brand recognition. Since referral programs depend on the referrer knowing and understanding the business challenges and goals of those they’re referring, there are typically upticks across several metrics categories. Positive outcomes for leads or customers through referral programs include: - Referred customers are more likely to stay with your company, with a 37% higher retention rate. - Those qualified, referred leads bring in more revenue — one study found that referred customers bring in at least 16% more in profits. - When referral marketing is implemented correctly, customer acquisition cost decreases by 13%, since those customers make faster decisions. - The customer lifetime value (CLV) of a referred customer is, on average, 16% higher than that of a non-referred customer. ### 3. A Brand Boost Brand-building is an ongoing project, and for smaller businesses, it can be too pricey to run large-scale, splashy digital ads, or to compete on search and social channels. These companies can benefit from referral programs as another way to build name recognition and brand trust, in addition to the other metrics mentioned above. Word-of-mouth marketing is powerful for lead generation — 85% of small businesses say that word-of-mouth referrals are the number one way that new prospects discover them. Another survey also found that word of mouth is a highly trusted tactic, with 90% of people trusting word-of-mouth marketing over other types of advertising. ### 4. Loyalty Increase Capturing customer loyalty and turning new customers into return users is harder than ever, with competition high and consumer trust in marketing lower than in years past. Referral programs cut out much of the funnel to convert a referral into a user, and from there, into a repeat customer. Referred customers are more loyal, with higher lifetime values, and they tend to purchase more than customers acquired via other channels. Referral programs can increase customer retention rates by up to 37%, so those customers are more valuable over time. For B2B referral programs, those referring in new leads likely understand the challenges and needs of the people they’re referring, and how the product can specifically help them. That connection can lead to yearslong subscriptions or product usage. ## Key Takeaways A well-run, established referral program can benefit both the company running it and the users who are referred in. Referral programs are an inexpensive, high-value marketing tactic that bring lots of benefits, including higher CLV, faster conversion rates, and increased brand trust and loyalty. ## Frequently Asked Questions (FAQs) ### How can I start a referral program? Start with your goals and potential referral sources, then pick the incentives you can offer and decide how you’ll promote the program. Once the program is established and launched, make sure you’re tracking referrals, qualifying leads, and measuring the success of those users who came in as referrals. Make sure to offer training for sales and customer service teams so they can nurture and serve those leads, and then customers. To capture that brand excitement up front, businesses should initiate referrals as soon as they’ve converted a new customer, when they’re excited about the product. ### What are the different types of referral programs? There are a few types of referral programs to consider. There are those based on an incentive structure, such as ones that reward just the referrer, or the customer; those that reward both; and tiered programs that offer increasing rewards based on the number of referrals. Other aspects of referral programs include how users can refer others, whether by email, social media, reviews, or more. Businesses can also explore approaches like contests and sweepstakes, brand ambassador programs, gift card or discount offers, exclusivity offers, and waiting lists, where a user can move up by referring others. ### How do you measure the success of a B2B referral program? Many typical marketing metrics work well for referral programs. Focus on the ones that typically do well with referrals or word-of-mouth tactics: conversion rate, lifetime value, repeat purchases or renewals, and cost per acquisition. For referral programs specifically, track referral rate, or the percentage of new customers acquired through referrals, to see how the program is performing over time. --- ### 6 Financial Services Marketing Trends to Watch in 2026 URL: https://www.taboola.com/marketing-hub/financial-services-marketing-trends/ Last Modified: 2026-03-16 12:10:11 In the financial services sector, there’s always a tricky balancing act, raising brand awareness while maintaining strict adherence to consumer privacy laws. In 2026, as these privacy concerns take center stage, banks, credit unions, fintech solutions firms, and other financial services providers are all rethinking how they engage with consumers, while also trying to build trust. With such a constant evolution of consumer expectations and the relentless march of innovation in technology, it can be tough to keep on top of it all. To help you figure out what’s what, here are six trends currently shaping financial services marketing strategies. What’s changed in our 2026 update: - All entries include updated and current information and advice. - All stats and figures updated with new and current information. - All new graphs and graphics added. - All FAQs updated with current information. ## Trend 1: Building Digital Trust and Security in Financial Services Marketing Artificial intelligence (AI) has become a useful tool for financial services providers. At the same time, though, it’s also creating new challenges. The emergence of deepfake scams and other types of AI-generated fraud means consumers need to remain vigilant. For financial services marketers, that brings a new challenge to overcome: How do you make it clear that your platforms and services are safe? Here are some key statistics eroding consumer trust in 2026: - A.I.-driven fraud now constitutes 42.5% of all detected fraud attempts in the financial sector. This highlights the growing need for robust digital trust measures. - 60% of companies faced AI-enabled attacks in the past year, showing a sharp acceleration of AI-based fraud and scam tactics - Personalization is a must. Of customers surveyed, 71% expect personalized interactions with brands and 76% get frustrated with more generic messaging. - 72% of financial institutions are concerned about data quality. In 2026, it’s more important than ever to be transparent with both prospects and existing customers. Communicate clearly the measures you’re taking to keep consumer data safe, including security measures and data collection protocols. Even if customers don’t read every word of the privacy policies on your site, notifications when you have updates can remind them that you take security seriously. Customer testimonials can also build credibility for financial services brands. In addition to statements about your services, make sure you have at least a couple of testimonials that speak to your commitment to security. A word of warning, though: The financial services sector is tightly regulated to protect consumers. As a marketer, you’ll need to avoid misleading claims of the type laid out by the Federal Trade Commission as Unfair or Deceptive Acts or Practices. You’ll also have to make sure your disclosures are in compliance and keep your email marketing in line with the CAN-SPAM Act. Building consumer trust goes beyond your marketing tactics. When your business prioritizes security and transparency, that will show through to your customers. ## Trend 2: Personalization and Customer Experience in Financial Services Financial services marketers have a number of cutting-edge tools at their disposal, and those tools have made personalization easier than ever. That means for marketers in the space, personalization is becoming crucial for staying competitive in a field crowded with banks, credit unions, and solutions providers. Today’s financial services marketers harness data analytics and AI to deliver tailored experiences to both potential and existing customers. Here are some of the top strategies they’re using: - Data-driven insights: Using customer information, marketers are personalizing advice, offers, and messages across all platforms, whether they’re communicating by email or creating and distributing display ads. - Seamless digital journeys: Consumers don’t always stick to one device when shopping for financial services. Savvy marketers are focusing on providing a seamless journey as customers move between mobile devices and desktop. - Long-term engagement: Winning a new customer is only the beginning: Financial services marketers are now focusing on fostering long-term relationships through personalized interactions that evolve over time. Here’s what else you need to know: - 72% of customers cite personalization as a deciding factor in where they bank. - 84% of customers would switch banks for timely, personalized advice, and 74% would stay loyal when tailored insights and automated money management are offered. - 73% of customers expect more personalization as technology advances, while 64% expect it as their spending increases. - Omnichannel personalization is essential to good customer service. While 79% of customers want consistent communication across departments, 55% percent feel communications are too disjointed. Source: Salesforce Personalization isn’t an option in 2026 — it’s required to remain competitive. The right tech tools make it easy to responsibly target customers and personalize interactions to boost engagement. ## Trend 3: The Rise of Mobile-First and App-Based Financial Services Marketing At one time, mobile was considered a “second screen,” with most consumer activity happening on a laptop or desktop. In 2026, though, apps dominate the space, which means mobile marketing should take center stage in your strategies. Here are some stats showing just how consumers access financial services today: - 54% of U.S. bank customers say they use a mobile app as their primary way to manage an account. - 52% of customers check their favorite financial app at least once per day. - Customer satisfaction with financial apps remains high. 84% of mobile banking users report satisfaction with the mobile experience on their primary financial app. Source: American Bankers Association Banks and credit unions have especially embraced the app experience, putting pressure on other lenders to offer fully functional apps to their customers. But, it’s not enough to offer an app: To fully engage customers, financial service marketers should optimize their app’s functionality and user experience. Customers should find the app easy to use, with intuitive screens and all the features they need. Personalizing each user’s experience can be a great way to engage customers. For banks and credit unions, it’s especially important for apps to be fully featured, giving account holders the same services they’d receive if they visited a branch. Once customers have the app, they should be able to opt in to receive push notifications. Make sure these notifications are related to information a consumer wants, such as when a bank balance drops below a certain threshold or a new statement is ready. Even the best app will flounder if customers can’t find it, so be sure to include easy download options in customer emails and other communications. In terms of making it discoverable to people searching, use the right keywords for your app’s title, subtitle, and description, and always check to ensure you’ve chosen the right category. You’ll also want to use eye-catching images for your icon and screenshots, making sure they accurately describe what your app offers. Competition is fierce in the financial services sector, but with a little extra work, you can stand out. Personalization and optimization are key to winning new customers and keeping them around for years to come. ## Trend 4: Leveraging Content Marketing and Financial Literacy Initiatives Content marketing has emerged as a powerful way for financial services marketers to connect with customers. By providing information that answers questions consumers have, you can stand out from your competitors. A few key stats show the demand for financial education in the United States: - Financial literacy rates in the U.S. remain low, with adults answering only 49% of personal finance questions correctly on average. - Baby Boomers and the Silent Generation showed the highest financial literacy, at 55%, while Gen Z had the lowest, at 38%. - 61% of U.S. adults say they’d be uncomfortable discussing their bank account balances with close friends or family. - Women tend to report lower confidence and often score lower on financial literacy assessments than men. GMF, a leading mutual insurance group, demonstrated how powerful educational content can be. Through display ads pointing to a landing page offering helpful information about GMF’s products, along with the option to submit their information to move further down the funnel, GMF boosted lead volume by 82%, exceeding its goal. Whether you choose blog posts, videos, webinars, interactive calculators, or a combination of all of the above, you’ll need to touch on the topics that interest your customer base. Depending on the type of services you offer, here are some ideas for educational content: - Budgeting: Step-by-step guides on creating a budget, along with budgeting templates, tend to be popular with consumers. - Credit scoring: Whether consumers need to improve their credit score or build one from nothing, a guide to how scoring works can help. - Debt reduction: With the average consumer debt topping $100,000, it’s more important than ever to explore manageable ways to pay off loans and other debts. - Saving: Setting aside money, whether for next summer’s vacation or eventual retirement, can be tough. A guide going through small steps to save can help consumers start down the right path. - Investing: Many consumers are daunted at the prospect of putting money into the stock market. A beginner’s guide covering various investing options, along with their level of risk, could work well for an investment firm or bank. As you’re creating educational content, don’t forget search engine optimization. With consumers increasingly relying on voice search, AI-powered search tools, and conversational queries, financial service providers have a prime opportunity to create content that answers some of the most frequently asked questions within their vertical. ## Trend 5: The Strategic Use of Social Media and Influencer Marketing in Finance Social media remains one of the best places to find an audience. In 2026, however, many financial marketers are reporting flatter returns from traditional social campaigns. Audience saturation, ad fatigue, algorithm inefficiencies, and rising ad costs are some of the top reasons marketers believe social is less effective these days. Don’t toss your social media strategy out the window just yet, though, because with the right approach, you can still get a decent return on investment for your social media efforts. The first step is to choose social media platforms that match your marketing goals and typical demographics. If you’re trying to reach younger consumers, for instance, short-form videos on TikTok might be a better approach, while customer testimonials on Facebook or YouTube could help you reach older audiences. Influencers can help financial services marketers get in front of consumers, too, but again, it’s important to remain compliant while doing so. The Federal Trade Commission maintains strict rules for disclosing paid partnerships, and those rules apply whether you’re compensating the influencer with cash or free products or services. Analytics are the last piece of the puzzle when it comes to social media marketing. It’s important to regularly track performance on all your efforts, from social media posts to paid ads to influencer collaborations. By gathering this data and letting it inform future marketing efforts, you can boost efficiency. A few interesting stats on social media and influencer marketing include: - 24% of investors say they receive financial information from social media, and 35% of respondents under 30 rely on social media for financial insights. - In the general consumer population, 47% say social media has helped improve their financial decision-making, but that rises to 62% among Gen Z. - 58% of adult consumers have purchased products because of an influencer endorsement. ## Trend 6: Navigating Regulatory Landscapes and Compliance in Digital Finance Marketing Financial service providers face tighter regulations than many other industries, and those regulations have to be considered when you’re planning and launching marketing campaigns. Here are some statistics relating to regulation and compliance in 2026: - Only 10% of financial institutions use a fully automated compliance management system, while 58% combine automation with spreadsheets and/or email. - 25% of financial institutions expect their compliance budgets to increase in the next 12 to 18 months, while 45% expect budgets to remain flat. - Compliance teams using manual processes report 7x more examiner questions and concerns and 4x lower satisfaction with staffing and strategic influence than those using automated tools. - In 2024, the Financial Industry Regulatory Authority (FINRA) reviewed 75,125 advertisements and sales communications. The agency logged 730 disciplinary actions. Source: Ncontracts In 2026, marketers need to pay close attention to: - U.S. Securities and Exchange Commission (SEC): Financial services marketers must comply with the SEC’s Rule 206(4)-1, otherwise known as the marketing rule. This rule sets strict requirements on statements that can be made in marketing and advertising. - FINRA: This nonprofit regulates brokers and exchange markets. Under FINRA regulations, marketers must ensure all statements are accurate, fair, and not misleading. - Truth in Savings Act of 1991: Also known as Regulation DD, this act requires disclosures on all terms and conditions that come with a new account or lower interest rate. - General Data Protection Regulation (GDPR): This applies specifically to European customers. If you market outside the U.S., it’s important to be aware of GDPR’s restrictions on how EU consumer information can be collected and used. Following these regulations will not only help you avoid legal repercussions and steep fines, but you’ll also be able to build trust with your customers. Just remember, you’ll need to be proactive to ensure your marketing campaigns are 100% compliant. Some key compliance strategies include: - Keeping an eye out for any changes in federal and local regulations that might apply to your marketing. - Conducting frequent audits of your marketing materials to ensure you’re in compliance with all current regulations. - Educating all relevant staff on compliance regulations and putting best practices in writing. - Leveraging tools that help automate the process of checking marketing materials for any compliance issues. ## Key Takeaways In a privacy-conscious sector, financial services brands can build trust by prioritizing transparency and security in all marketing efforts. In 2026, personalization is essential, and AI can help with delivering tailored experiences that satisfy individual consumer needs. When gathering data and using it for targeting, though, it’s crucial that financial services brands pay close attention to privacy regulations in addition to all compliance requirements. By creating engaging educational content and optimizing your mobile apps, you can build trust while making your marketing work for you. ## Frequently Asked Questions (FAQs) ### What are the most effective digital advertising channels for financial services in 2026? Social and search remain relevant in 2026, but soaring ad costs and customer fatigue have led financial service marketers to diversify their efforts. Pay-per-click, native, and display advertising, alongside email marketing, are solid performers for finance messaging, with greater emphasis on privacy-safe targeting, contextual relevance and creative optimization. Using technology to improve targeting and personalize your ads can keep you competitive. ### How can financial institutions generate leads and acquire and retain customers cost-effectively online? For financial marketers, 2026 is all about winning and retaining customers while also keeping costs low. Banks and credit unions are focusing on personalizing their marketing efforts to ensure each message has maximum impact. Many financial institutions now rely heavily on AI-powered tools that can deliver targeted content and offers, boosting the chances that recipients will take action. In addition, first-party data strategies, improved onboarding flows, and frictionless mobile experiences play a growing role in lowering acquisition costs. Investing in user-friendly apps and offering useful educational content can improve customer loyalty, effectively reducing acquisition costs. ### What are some successful examples of financial services digital marketing campaigns? Successful digital marketing campaigns in the finance sector often combine personalization, educational content, and the strategic use of technology. JPMorgan Chase is just one example of some of the biggest brands prioritizing financial literacy. The multinational corporation dedicates a section to its website called Research, and in that section, you’ll find thought leadership content and original research that helps establish it as a trusted resource. If you decide to go this route, don’t stop at publishing the content: It’s important to promote it in different formats on different channels to make sure it’s viewed by the widest possible audience. ### How is AI impacting the financial services marketing landscape? AI stands poised to impact every area of financial services marketing, from targeting and hyperpersonalizing content, to image and copy creation, to chatbot-powered customer service and analytics. Financial services marketers can analyze large swaths of data to get never-before-available insights, then use that information to identify trends and customer needs. In 2026, AI is increasingly used for predictive modeling, creative testing at scale, and real-time campaign optimization, and it can also be used to audit marketing processes to ensure regulatory compliance. ### What are the key metrics for measuring success in financial services digital marketing? Before you can use the latest tools to measure your marketing success, you need to understand the most important metrics. In 2026, key performance indicators for banks, credit unions, investment brokerages, and other financial services providers include customer acquisition cost, customer lifetime value, cost per lead, and conversion rates. Financial services marketers should also pay attention to engagement metrics like click-through and open rates, as well as retention, lifetime engagement, and incremental lift across channels. Overall, paying attention to your return on investment can help you ensure you’re getting the most out of your marketing efforts and expense. --- ### Building High-Converting eCommerce Funnels with Native Content: Xevio Blog Series (Part 2) URL: https://www.taboola.com/marketing-hub/building-high-converting-ecomm-funnels-with-native-content/ Last Modified: 2025-07-19 15:23:58 In the first part of our e-commerce interview series, Xevio co-founder and CEO Nadim Kuttab gave his thoughts on the ins and outs of driving direct response sales on the open web. In this installment, we shifted our focus to discussing how to effectively use native advertising in e-commerce funnels. Read on to discover an all-too frequent landing page error, advice for accurate attribution, and Kuttab’s two biggest conversion killers. ### How do you design an e-commerce funnel specifically optimized for native advertising traffic? Native funnels are linear, so what you want to avoid is having a lot of different product alternatives or options. For example, sending people from your advertorial to your homepage, where they have 400 products to choose from, is a terrible idea. Too much choice will lead to people not knowing what to buy and, therefore, not buying directly. Use linear funnels where there's maybe a couple of additional bundle options, but that's it: You don't want to give people the option of 300 different types of socks or pants or shirts, just keep it very straightforward. If anything, upsell in the checkout process, but the key is not giving them too much choice, because you'll lose them. You want to push them through that funnel, not get them out on the sides. In terms of designing that funnel, there are a lot of good examples. I would look at what you see on the internet, see how they're structuring it — I would never copy content, but look at the format of how they're running: What is the messaging in the ad, what is the structure of the content pages, how are their checkout pages built, how are their bundles structured? Those are things you can look at, learn from, and apply to your own setting. I should stress, if you copy their actual content, you're never going to be able to compete with the original content owners, you'll always be below them on the feed and therefore have substantially worse performance. I always say ripping off content is like, RIP — you're just going to lose in the game. ### What are the essential components of a high-converting landing page for e-commerce products on the open web? I'm going to define a landing page in this context as an advertorial — the content piece between an ad and the product page. We have a ton of information on this in our native hub, which I highly recommend people look at! In terms of the essential components of a landing page, I always say, give them context. Why are they here? Why is it relevant for them? Introduce the problem and explain why it’s relevant to the user, and what the user can do about that problem. Then, introduce a product that solves that problem before pushing them to action with testimonials and FOMO. That's the standard funnel of every single landing page out there, whether it's 300 words or whether it's 30,000 — there's a huge span there, but in essence, they're all structured like that. Capture the attention, capture the interest, take that interest, bring it towards a solution, and then sell a solution to their problem. ### What's the optimal balance between educational content and direct product promotion within an e-commerce native funnel? You need to start with education and then move towards promotion. If you have an ad that says “buy these shoes,” you’re directly selling, you’re not educating. You can be relatively salesy in your content, but you're going to pay a lot for that click on the ad level, and probably not going to be able to scale it as much, because your CTR is piss poor. Most of the really good Taboola native campaigns don't sell in the initial stages. They might present a problem, or a theory — there's a lot of people that want to save money on something, or who want to feel better about a problem, so you use that in your ads to capture the relevant audiences. Think of your ad as being like a net. If you have a super broad ad — imagine an ad that says, “Everybody who clicks here gets $100.” That isn't setting a filter, everybody wants $100! You're just going to get a bunch of people on your ad who want a hundred bucks for clicking, which obviously you're not going to give them. Because it’s misleading, a lot of people are going to jump off. On the other hand, if you have an ad that says, “Are you 38 years old and born in Atlanta? Click here,” it's too specific — you're never going to get anywhere because it resonates with too few people. You have to find the middle ground. If I have a product for back pain, for example, I would have my first paragraphs be about how many people suffer from back pain, why it's so annoying, and how it can really inhibit living a happy and healthy life. Then I’d introduce a solution that's easier and cheaper than most people think and therefore you should check it out here, etc. That would be the structure, where most of the pages are educating instead of selling. ### How can advertisers leverage Realize's flexible creative formats to enhance their e-commerce funnels? Realize is relatively fresh, so I don't think we’ve fully cracked everything that can be done with Realize’s new ad formats, yet! But, what we have noticed is that creating campaigns in four specific formats and targeting these different pockets of traffic individually can be incredibly effective. Then, we're just playing around with different ideas for each type of format until we find something that works. I recommend everyone try it! If you have a general campaign that performs, break it out into, say, email, in-app, display — try these different pockets of traffic that will perform differently and try to crack an approach, especially on the ad level. ### What are common conversion killers in e-commerce native funnels, and how can they be avoided? First off, treating it like Facebook. It's not Facebook! You do not have an algorithm that will find you pockets of people to run, just remember that — you have a bidding mechanism that will limit your costs and try to be efficient in finding you clicks, but you don’t have the ability to essentially microtarget people with an algorithm. Secondly, as I said, giving people too many options. I've seen so many brands fail, especially in e-commerce, because they're sending people to the homepage. It's depressing because it's so avoidable. Choose a product, choose a niche, choose a bundle, then focus on that bundle. Yeah, you can cross-sell them later via email, that's fine, but just get them to buy something first so you can scale and spend your money profitably on Taboola. Those are the two biggest pitfalls. There are so many small things that you can do wrong, but they normally don't have as big of an impact. To go back to the Facebook point, I’ve seen people come from Meta and just expect to see success because they were successful on Meta, but the truth is that I've seen brands fail because they didn’t look at Taboola as its own traffic source, instead seeing it simply as an extension of what they were doing on other other social channels. That's a great way to lose a bunch of money on learning nothing, because you're not testing, and you're not using the product correctly. ### How do you ensure seamless tracking and attribution across the entire e-commerce funnel on the open web? You need to have a tracker, period. Choose a model, stick with that model and make sure that you're using it across your entire business. Otherwise, you'll be double counting. If you only use the conversions in the individual traffic sources — say, if you look at Meta and Google and Taboola independently — you're going to have a lot more conversions than you actually have sales. You need to have a system in place that allocates revenue to spend. I'm not here to tell you what the best system for that is, because it really depends on the use case, but we've seen a lot of success with Triple Whale, and are relatively happy there. --- ### Google Core Web Vitals: The 101 Metrics Google Uses When Evaluating Web User Experience URL: https://www.taboola.com/marketing-hub/google-core-web-vitals/ Last Modified: 2026-06-22 08:34:16 When you have a doctor’s appointment, what are the first things the physician checks out? Is it the health of your nail beds and the hair on your neck? Or, is it your lungs, your heart, your eyes, and so forth? Google is much like an examining doctor when it comes to evaluating a web user’s experience with any given website: It evaluates the most important parts of a web page’s form and function, and this analysis helps determine how well (or how poorly) a website will be featured in search results. When a site has “healthy” pages, it will rank better. When a site is on the sickly side, to extend my metaphor, it might be lost far down in those search results, hardly ever to be found at all. Which can be a shame, because a site with well-written copy, good deals, funny memes, and so on might be essentially lost just because a few elements weren’t properly in place. ## What Are Core Web Vitals? There are three main core web vitals (CWV) Google looks to, and I’ll break those down in detail in a moment. To give you a top-level understanding, though, Google’s core web vitals are essentially an analysis of how well a web page loads. That is to say, how quickly the page loads, how well formatted the page appears once loaded, and how responsive it is to user interaction. If Google were the doctor in my example above, the vitals can be thought of as those things that just have to be in place to call a page healthy. ## What Are the Main Core Web Vitals? The three main core web vitals that Google considers when evaluating a web page — and determining how well it will rank in search engine page results (SERPs) — are as follows: ### LCP: Largest Contentful Paint This is the measure of how long it takes the largest “content element” of a page to load fully. Think of a large carousel of images atop a page, or a video waiting to become visible. Google can measure a page’s overall loading speed (or sluggishness). ### CLS: Cumulative Layout Shift You’ve probably noticed that some web pages are shaky, unstable, and erratic as they load, with content appearing in different places, loading layered over other parts of the page, and generally what might be called wonky, to use a nontechnical term. That’s high layout shift, which is not a good thing in the eyes of Google’s algorithms. Pages that maintain visual stability as they load have lower shift, which Google likes, because it makes a user’s experience better. ### INP: Interaction to Next Paint Formerly called first input delay (FID), INP measures the responsivity of a web page to a user’s action, be that clicking on a link or video on a computer, or tapping something on a smartphone or tablet. ## Why Are Core Web Vitals Important? Core web vitals are important because they measure potential user experience, and in so doing, greatly influence search engine rankings. They measure the aforementioned aspects of a web page's performance (loading speed of major visual elements, visual stability, and user interactivity efficiency) which significantly impact how users perceive and interact with a website. By focusing on making sure these metrics are in good order, site owners/managers improve user satisfaction and potentially increase conversions. In this way, paying attention to core web vitals is a part of search engine optimization (SEO) that can’t be overlooked. This is also very much the case for marketers placing paid ads online. Core web vitals can have a large impact on the performance and cost-effectiveness of paid advertising campaigns, for better or for worse. Optimizing core web vitals for landing pages can lead to better ad rankings, lower cost-per-click rates (CPC), and improved conversion rates. ## How Did the Last Google CWV Update Affect Websites? The last major CWV update at the time of this writing was in March of 2024, though Google has made several additional changes since then, such as that INP replacement of FID noted above. The updates have mostly re-emphasized the importance Google puts on user experience, further attempting to improve the quality of user experience by prioritizing high-quality content and penalizing low-quality pages and sites. ### Quality Content Prioritization The update intends to show users more content that is high-quality, genuinely useful, and fact-based (when applicable), and less content that feels as though it has been created primarily for search engines to track it down and serve it up. ### Reranking The March 2024 core web value update initiated the reranking of many sites, with those found to be populated largely with AI-created images and content being reduced in rank, while those with genuinely original copy and graphics were weighted more heavily. ### Site Deindexing Google has become much more active in its deindexing of websites it finds to have extra low-quality content, or that it finds to be in violation of its policies. The company stated that it aimed for a 45% reduction of poor-quality content. ## Core Web Vitals Effect on Ads If the sites on which you’re running ads have poor core web vitals, you’re probably throwing away a lot of cash. ### CLS Issues CLS stands for cumulative layout shift, a measure of a web page’s visual stability as it loads. CLS can significantly impact ad performance and revenue, as poor CLS (often caused by ads loading asynchronously and by shifting page content) leads to a negative user experience, potentially resulting in lower ad engagement, missed impressions, reduced conversions, and lost ad revenue. ### Heavy Ad Issues Heavy ads are online advertisements that consume excessive system resources, leading to slow website loading times and increased data usage. CWVs significantly impact the performance and effectiveness of heavy ads by influencing user experience and ad revenue. Poor CWV caused by heavy ads can result in missed impressions, higher CPCs, and lower ad revenue. ### Paid Ads Core web vitals significantly impact paid advertising, particularly within Google Ads. Websites with strong CWVs often experience better ad rankings, lower cost-per-click rates, and increased ad revenue. ### Lazy Load Ads Lazy loading is a technique where ads are loaded only when they are about to become visible on a user's screen, rather than all ads being loaded at once when the page itself initially loads. This approach improves a site’s performance by reducing the initial load time. CWVs significantly impact lazy-loaded ads. Properly implemented lazy loading can improve page load times and enhance user experience, which positively affects CWV evaluation. ## Key Takeaways Core web vitals are a pulse check Google uses to analyze how well a web page would perform in the eyes of a user. CWV involves three primary metrics: LCP (largest contentful paint), INP (interaction to next paint), and CLS (cumulative layout shift). These measure load speed of major visual elements, the speed at which a page reacts to user actions, and the visual stability of a page as it loads, respectively. When these metrics are all in good working order, it improves a site’s search engine optimization and can lead to more views and, in many cases, better ad revenue. When a site’s CWV is off, it can mean the site is buried in Google search results and ad spend is wasted. ## Frequently Asked Questions (FAQs) ### Why do people worry about core web vitals? People who manage sites care about CWV because they are a set of metrics that directly impact user experience and, consequently, influence search engine rankings. By optimizing for these vitals, websites can improve user engagement, reduce bounce rates, and potentially achieve higher rankings in search results, making them a crucial aspect of SEO. ### Why is it called “paint” in CWV? The term “paint” in this context is used because much of core web vital analysis is about how quickly pixels load and change. Think of those pixels as digital paint that shifts to show new content, be it video, ads, text, and more. ### What is the most important core web vital? All are important measures of overall web page health and performance, but one could argue that LCP, or largest contentful paint, is the most important, because the loading of major visual elements plays a major role in how users perceive the overall efficacy of a site. --- ### Customer Journeys: What Are They? How Are They Managed? URL: https://www.taboola.com/marketing-hub/customer-journey/ Last Modified: 2025-07-15 09:57:42 To convert a potential customer into a lifelong one, it's essential to understand every step of the customer journey. This journey offers businesses a chance to attract, meet, engage with, and earn customers across multiple stages. A comprehensive understanding of these stages will enable your business to create tailored solutions that strengthen loyalty and improve brand interactions. It's important to recognize the data received at each stage, the challenges encountered (such as misconceptions or service issues), and to explore effective strategies for improving engagement and customer satisfaction. ## What Is the Customer Journey? The customer journey is the complete experience that each customer has with a brand, and is an essential element of business. It can be likened to any other journey, with a starting point, a destination, and various steps in between. The ultimate goal is not just making a purchase, but ensuring a positive experience with your product or service, which can include after-sales support, leading to customer loyalty. Overall, the customer journey offers valuable insights into how customers think and behave, starting from their initial awareness of the brand to their interactions after making a purchase. Incorporating these insights into your marketing strategies will enhance lead generation and facilitate more efficient marketing. ## Customer Journey Mapping: What Are the Customer Journey Stages? Mapping the customer journey typically consists of five key stages. Three of these stages occur before the acquisition of a customer (Awareness, Consideration, and Decision), and two occur after the acquisition (Retention and Advocacy). Each of these stages requires different psychology and activation of the customer, so let's break them down: ### 1. Awareness The awareness stage represents the initial point at which potential customers become aware of a specific need or desire. This realization leads them to investigate various brands or products that could meet that need. During this phase, marketing approaches focus on increasing brand visibility and recognition, making sure that the target audience not only notices but also remembers the available options. The priority here is establishing a strong market presence, utilizing compelling visuals and messaging to attract attention and spark interest. ### 2. Consideration During the consideration stage, customers actively explore their options, comparing products or services that might meet their needs. They gather information, read reviews, and assess the advantages and disadvantages of various options. ### 3. Decision In the pivotal decision phase, customers find themselves poised to make a purchase. Influenced by their past experiences, along with the valuable insights gathered during the consideration stage, they carefully evaluate their options before determining what product to buy and which vendor to choose. ### 4. Retention After a purchase, it is essential to maintain the customer’s interest. Brands should strive to deliver an exceptionally smooth and enjoyable experience, providing comprehensive support and actively nurturing relationships to cultivate lasting loyalty. ### 5. Advocacy If you’ve done your job well, advocacy becomes simple. The customer is warmed up, educated, nurtured, led towards a purchasing decision, supported after making that decision, and now wants to share their experience with their peers about your product or service. At the advocacy stage, the customer will serve as a powerful asset, actively championing you to their friends, family, and colleagues. ## Why Are Customer Journey Stages Important? As a business owner, you want to convey the right message at the right stage, just as you would for each stage of a child's development, or the progression of a relationship. Understanding the stages of the customer journey is vital for several reasons: ### Tailored Experiences Each stage of the customer journey requires a unique approach. Understanding the needs of customers at each phase enables brands to create more personalized interactions, making them feel valued and understood. This strategic approach is key to improving customer engagement. ### Identifying Pain Points Mapping the journey helps to highlight areas where customers may face challenges. This proactive approach allows organizations to intervene and improve the overall experience, demonstrating a strong commitment to customer satisfaction. ### Resource Allocation Recognizing which stages require the most attention and resources can help in developing a more efficient marketing and sales strategy. ## Content and Messaging by Stage Each of the five stages of the buyer journey requires different content and visuals to connect with the customer. ### Awareness Stage At this point, informative and engaging content is essential. Brands can utilize blog posts, social media ads, and educational videos to highlight problems and solutions that are relevant to their customers. ### Consideration Stage Here, businesses should focus on providing detailed product information, reviews, case studies, and comparisons. Content that answers frequently asked questions can be particularly beneficial. ### Decision Stage At the decision stage, offer clear calls to action, promotions, and reassurance (such as guarantees and testimonials) to encourage customers to complete their purchase. ### Retention Stage Sending follow-up emails, offering loyalty programs and providing helpful resources, can ensure that customers feel valued and appreciated. ### Advocacy Stage Encourage satisfied customers to share their experiences through testimonials and social media engagement, thereby fostering a sense of community and attracting new customers. ## Advantages of a Customer Journey Map ### Enhanced Customer Understanding A well-structured map provides businesses with valuable insights into customer behavior, preferences, and pain points, enabling them to cater directly to their customers' needs. ### Improved Communication By identifying the appropriate messaging and channels for each stage, brands can enhance their communication efforts to better resonate with customers. ### Increased Customer Loyalty By optimizing each stage of the journey, businesses can create a more satisfying experience, resulting in higher retention rates and increased brand loyalty. ## Biggest Challenges of Mapping a Customer Journey There are also some challenges to this mapping, which are good to be aware of in order to head them off at the pass. These include: ### Complexity of Customer Paths Customer journeys are rarely linear, and sometimes come with friction points or even gaps. Customers may switch channels or regress to previous stages, making it challenging to create a straightforward map of the customer journey. It’s vital to understand and address these points in order to avoid preventable loss of leads mid-funnel. ### Data Overload With numerous touchpoints and data sources, distilling relevant insights can be a daunting task. Businesses need to focus on the most important metrics and customer feedback. This includes the post-purchase portion of the journey. ### Organizational Alignment Ensuring that all teams understand and are aligned on the customer journey can be a challenge. Collaboration across departments is vital for delivering a consistent experience. ## How to Build a Customer Journey Map Creating a customer journey map to outline potential paths should be the first step in your marketing plan. You can use data to analyze customer interactions, improving each step to remove barriers that potential and existing customers face. ### Research and Gather Data Collect both qualitative and quantitative data through customer interviews, surveys, and analytics to better understand customer behavior and preferences. ### Define Stages Clearly outline the key stages of the customer journey relevant to your business — a customer journey for an enterprise buyer would look very different to that of a freemium SaaS user, e.g. — focusing on the customer's interactions and emotions at each stage. ### Identify Touchpoints List all touchpoints along the journey, noting how customers interact with your brand through various channels. ### Analyze and Optimize Use insights gathered during mapping to identify areas for improvement in the customer experience, continuously revising the map as necessary. ## Ways to Use Automation in Customer Journeys You can use automation in several ways to personalize your customer's journey and make them feel attracted to your brand, such as: ### Personalized Messaging Automation tools can send personalized messages to customers based on their behavior, ensuring relevant communication at the right time. ### Streamlining Follow-Ups Automated emails can handle post-purchase follow-ups, offering customers assistance or incentives for future purchases. ### Gathering Feedback Utilize automated surveys to gather customer feedback following specific interactions, enabling the identification of areas for improvement. ## Customer Journey Metrics ### Conversion Rates Tracking how many customers progress from one stage to the next can indicate the effectiveness of the journey. ### Customer Satisfaction Scores Measuring customer satisfaction at various stages can provide valuable insights into how effectively customer needs are being met. ### Repeat Purchase Rates Monitoring the rate at which customers return can help evaluate the effectiveness of retention efforts and the success of loyalty programs. ## Customer Journey Optimization Ongoing testing and continuous improvements will significantly enhance your business’ marketing efforts. The following approaches not only ensure quality but also foster innovation and adaptability, paving the way for greater achievements: ### A/B Testing Conduct A/B testing on messaging and content to determine what resonates most with customers at each stage of the journey. ### Continuous Feedback Loop Implement a system for ongoing feedback to continually adapt and optimize the customer journey. ### Performance Metrics Assessment Regularly review key performance indicators and metrics to adjust strategies in response to customer behavior and preferences. ## Multi-Channel and Omnichannel Journeys Creating omnichannel journeys for your customers ensures a smooth experience across in-store, online, and app interactions. When everything feels connected, it boosts engagement and builds loyalty. ### Create a Consistent Journey Across All Channels It’s important to maintain a coherent brand voice and experience, regardless of the channel customers choose to engage with. ### Map Cross-Channel Journeys Understand how customers transition between channels, ensuring a seamless experience whether they’re shopping online or in-store. ### Unify Customer Experiences Across Platforms Utilize technology to share customer data across silos, creating a unified view of the customer journey to foster better interactions. ## Customer Journey Tools & Platforms ### Mapping Software Utilize specialized mapping tools to visualize the customer journey, facilitating the identification of gaps and the optimization of experiences. ### CRM Systems Customer Relationship Management platforms can help track customer interactions, segment audiences, and automate communications. ## Key Takeaways Understanding the customer journey is important for any business seeking to enhance customer experience and engagement. By mapping the journey, recognizing key stages, and continuously optimizing the experience, businesses can foster loyalty and drive growth. ## Frequently Asked Questions (FAQs) ### What are the five main points of a customer journey? Awareness, consideration, decision, retention, and advocacy ### What’s the difference between the customer journey and the buyer’s journey? The buyer’s journey is a shorter, three-step process encompassing awareness, consideration, and decision. However, it doesn’t address how to retain customers after a purchase. ### How often should I change my customer journey mapping? The customer journey is constantly evolving. Once you've outlined your customer journey, it's essential to periodically reassess it to ensure it stays relevant and updated. Utilize website analytics, customer feedback, and social media metrics to aid this evaluation. --- ### 18 Digital Advertising Trends Performance Marketers Should Know Of URL: https://www.taboola.com/marketing-hub/digital-advertising-trends/ Last Modified: 2026-06-23 08:58:29 The digital marketing landscape is always evolving, but in 2025, the pace of change appears to be accelerating faster than ever, thanks to rapid technological advances and shifting consumer behaviors. From the rise of artificial intelligence (AI)-powered tools to the dominance of short-form video and the increasing reliance on first-party data, advertisers must continually adapt if they want to stay ahead. In this article, I’ll examine the key digital marketing trends shaping the industry today and provide insights on how to focus your efforts to maximize results. ## 18 Insightful Digital Marketing Trends Advertisers Should Know of ### 1. Social Media Short-Form Video Content Social media platforms remain critical to digital advertisers, but it’s important to note that the way users engage with social media is shifting. According to Hubspot, marketers say that short-form video delivers the highest return on investment (ROI), which means that advertisers need to adapt their strategies. Major trends include a shift toward short-form video on platforms like TikTok, Instagram Reels, and YouTube Shorts — even Threads is emerging as a promising platform for short-form content. - According to Sprout Social’s Benchmark report, almost 80% of people prefer to learn about new products through short-form video content. - 93% of marketers say they will invest more time in social media marketing in 2025, per the same report. ### 2. Combat Rising CPM Costs with Channel Diversification As advertisers continue to shift towards social media advertising, increasing CPMs and diminishing returns on popular platforms like TikTok, Meta, and YouTube are requiring them to increase their social ad spend. To lower costs, consider diversifying via efforts on the open web, where a mix of native and display ads, alongside content recommendations, can drive strong performance. Channel diversification also prevents advertisers from becoming dependent on closed platforms that can change dramatically with little notice, e.g., Twitter/X and Meta making changes to their content moderation policies. - TikTok’s Q1 2025 U.S. ad spend rose by 15.6%, according to EMarketer. - CPMs (cost per mille, i.e., the price an advertiser pays for each 1,000 impressions of an ad) continue to rise across several popular platforms, per the same report. ### 3. AI and Emerging Technologies In 2025, brands are using AI more than ever to create, test, deliver, and optimize their ad campaigns. Take content creation, for example: AI tools now allow you to generate multiple headlines and ad copy variations tailored for different campaign goals, audience segments, and platforms. This speeds up the creative process for advertisers, allowing them to experiment more quickly than in the past — and free up their time for more strategic work. Major ad networks, such as Google, Meta, and Realize, are also now utilizing AI-powered bidding strategies that analyze user behavior, enabling them to adjust bids in real time, helping to optimize spend. Some brands are also experimenting with features like AI-generated voices, synthetic media (sometimes referred to as “deepfake” tech when used for more nefarious purposes) for virtual influencers, and lifelike AI avatars, making digital content far more interactive. Ultimately, all of this will continue to change the way marketers work. - According to Hubspot, 92% of marketers say AI has already impacted their role. - Despite this, the same report found that only 47% of marketers clearly understand how to use AI in their marketing strategy. ### 4. Search In the past, marketers relied on finding high-volume, low-competition keywords that matched what people were typing into search engines. In Google Ads or other pay-per-click (PPC) platforms, this meant bidding on specific keywords. Theoretically, the better the keyword targeting, the more likely a campaign would convert. However, this method is less effective in 2025, as Google’s AI Overviews and Microsoft Copilot often provide direct answers instead of webpage links. This means that marketers need to consider AEO (answer engine optimization) as well as traditional SEO, and ensure their content clearly answers questions and covers topics in depth. They also need to optimize for new formats, such as voice and visual search. ### 5. Content Perhaps the most significant trend in content for 2025 is that AI-powered content has become the new norm. AI tools such as ChatGPT, Jasper, Synthesia, and Descript are transforming how written and video content is created and edited, and how quickly. Additionally, the combination of first-party data and AI is allowing advertisers to create highly personalized content based on real-time behavior, instead of relying simply on audience demographics. ### 6. Video Video remains a critical medium for advertisers in 2025, but other, highly interactive mediums, such as quizzes and polls, augmented and virtual reality (AR/VR), and audio (podcasts and voice content) are also performing very well. Regarding audio, consider this: There are more than 584 million podcast listeners worldwide, and the number is growing. (Source: Hubspot Blog) But, that doesn’t mean marketers should shift all their focus away from video, since, as mentioned, short-form video in particular continues to perform very well. ### 7. Voice Search/Visual Search AI-powered tools, such as Google Lens, Apple’s Siri, and Amazon’s Alexa, are changing how people interact with brands, as consumers rely on these voice assistants and visual discovery tools more than ever to search and shop. Google reports that commercial intent is detected in one out of every five Google Lens searches, and 80% of Gen Z rely on Google for all aspects of their shopping. ### 8. AR/VR Augmented reality (AR) and Virtual Reality (VR) technologies are starting to change the shopping experience for consumers. AR works by adding digital elements, such as images or text, to your real-world view through your mobile device or tablet. VR takes it a step further by creating a fully immersive digital environment that you can experience through a headset, such as the Quest 3, from Meta. In a recent blog post, Artlabs, an AI-powered 3D company, highlighted the following real-world examples of how AR is being used in retail shopping: - Nike Fit now utilizes AI-powered virtual try-on technology to provide shoppers with accurate shoe size recommendations. All you need to do is scan your feet with your smartphone camera, and the AR does the rest. You can use the technology for in-store shopping, but it’s especially useful for online shopping when you can’t try on the shoes before you make a purchase. - Eyewear brands are using AR technology to analyze customers' facial features. They can then use your image to show you precisely how different pairs of glasses will look on your face. - Swedish furniture giant IKEA has an AR app called IKEA Place. You scan your living space with your smartphone, and the app will show you how furniture will look in your living room or bedroom. This makes it easy to choose the right furniture colors or dimensions. This AR-powered virtual try-on technology offers many benefits for advertisers, including personalizing the shopping experience, reducing return rates, and increasing conversions. ### 9. Nicke Influencers for Higher Engagement Influencer marketing has been around for a while, but it’s maturing and even fragmenting somewhat, with marketers increasingly leaning toward micro- or nano-influencers (those with fewer than 100,000 followers): 68% of marketers reported working with a niche influencer during the past year, while only 32% worked with larger influencers. Advertisers can benefit from working with smaller influencers because, alongside their increased relevance, they’ve established a high degree of loyalty with their audience, and as such, can generate higher engagement and conversions. - 93% of consumers feel brands should stay culturally relevant online. - 90% of consumers use social media to keep on top of trends and cultural moments. ### 10. Personalization Gone are the days when advertisers could only target large groups of users by demographics or location. Using AI-powered ad networks, you can now adjust your ad creative in real time based on user behavior for a highly personalized experience. In a recent article on its website, Upspring.ai noted this example: “Starbucks uses predictive analytics in its app to tailor offers based on behavior. A morning commuter might get a coffee coupon, while someone else sees a cold drink promo during a heatwave.” What’s even more impressive is that this personalization is happening at scale. Ad platforms can make millions of tweaks to ad creative, including text, images, calls to action (CTAs), colors, times, and more. Multiple viewers streaming the same video, e.g., might all see different ads from the same brand, based on their individual preferences. - 81% of digital marketers say AI-driven personalization has boosted sales and brand awareness. - 96% of marketers believe personalized experiences have increased sales. ### 11. Targeting toward First-Party Data One of the most significant trends in targeting is the shift away from third-party cookies and toward first-party data. Increasing consumer privacy expectations and stricter regulations, like the GDPR and CCPA, are raising the bar for how personal data can be collected and used. At the same time, major browsers including Google Chrome, Safari, and Firefox, are phasing out third-party cookies or exploring alternatives to protect user privacy. This makes it more challenging for advertisers to use traditional cross-site tracking and targeting methods. The good news is that first-party data, which can be collected by platforms with log-in and purchase data, email sign-ups, and surveys on your own website, tends to be of higher quality anyway. ### 12. Creatives Ad creative is no longer just about good design — you need to build your ads based on previous campaign performance and predictive AI models. You also need to constantly A/B test, measure performance, and make real-time adjustments to boost engagement and conversions. In 2025, marketers are creating more content tailored to each platform, in an effort to make each format feel native and attention-grabbing. While short-form video remains popular, motion in general is playing a bigger role — think subtle animations and dynamic elements in display ads or social posts. These help capture viewers’ attention and improve click-through rates. - 70% of marketers reported higher CTRs with animated ads than static ads. - 60.8% of marketers use animation in social media posts. - 37% of marketers are using animation in explainer videos. ### 13. Analytics and Measurement In the past, multi-touch attribution (MTA) assigned value to each touchpoint in the customer journey. But, with third-party cookie loss and the increased prominence of walled gardens like Meta, Google, and Amazon, complete funnel visibility has become more difficult. In 2025, MTA continues to be replaced by media mix modeling (MMM), which relies on historical data to predict how different marketing channels, such as paid search or email marketing, contribute to outcomes (e.g., conversion, sales). With MMM, you don’t need to track individuals across devices or platforms, so it's unaffected by the loss of third-party cookies. ### 14. Data-driven Approaches Data-driven marketing has been around for a long time, but the expectations on marketers to use the data to make smarter decisions, optimize in real-time, and increase personalization are higher in 2025. You can achieve this by focusing on first-party data, using the correct attribution in your customer touchpoints, and improving how you leverage data tools, such as customer relationship management (CRM) platforms and Google Analytics. You should also take advantage of advertising platforms that offer real-time AI-powered A/B testing. ### 15. Omnichannel Experiences In 2025, consumers want a seamless experience when interacting with brands, whether they’re clicking on a display ad, browsing a website, or walking into a store. To deliver, advertisers need to focus on omnichannel marketing, which includes creative messaging, consistent branding, and connected journeys. What does that look like in practice, though? Let’s say someone sees a display ad for a new pair of headphones while reading an article online. They’re interested, but they don’t buy right away. Later, they see a sponsored content recommendation for a blog post reviewing the same headphones. They still don’t buy. The next day, they get a personalized email with a special offer. They click on the link and finally purchase the product. This is an example of a connected customer journey and the omnichannel experience at work. ### 16. First-Party Data I’ve mentioned first-party data several times throughout this article, and for good reason. With fewer cookies and more privacy rules, first-party data has become incredibly valuable to advertisers. The beauty of first-party data is that it is owned, permission-based, and highly accurate. Here are some ways advertisers can collect first-party data in 2025: - Website or app interactions: User account creation, sign-ups, purchase history, product wishlists, on-site search behavior, browsing history, shopping cart activity. - Owned channels: Email newsletter sign-ups, SMS opt-ins, email campaign engagement. - Interactive content: Surveys and polls, quizzes, product reviews, feedback forms. ### 17. Lead Generation Gone are the days of static form lead generation. Today’s best lead magnets are interactive and helpful, blending seamlessly into the user's journey. For example, tools like quizzes and assessments provide users with immediate value through insights or recommendations, and give marketers much more specific data than just a name and email. Online calculators help users make decisions, and chatbots gather valuable information while providing users with friendly conversations. These all work because they’re highly interactive, personalized, and, if implemented correctly, seamless for the user. ### 18. Mobile In 2025, the majority of consumers use mobile devices to consume content and shop online. As a result, ads need to look great on smaller screens. There are several things advertisers need to ensure for this to happen: - Mobile-optimized creative: Ads must be mobile-responsive and feature clear, bold visuals with concise ad copy. - Fast-loading pages: People’s attention spans are shorter than ever. If your landing pages are slow to load, it will crush your conversions. - Vertical video formats: Vertical videos look better on mobile devices — think ads on TikTok, Instagram Reels, YouTube Shorts, and even mobile websites. - App-based experiences: Apps are optimized for mobile. They are easier to navigate, promote deeper engagement, and offer a richer mobile experience. Brands that can nail these attributes on mobile in 2025 will boost user attention, loyalty, and conversions. - 63% of consumers prefer to use mobile devices to find brand and product information. - The average American household has 21 connected devices. ## Key Takeaways As you can see, the most effective digital advertising strategies for 2025 are highly customized, data-driven, and built to withstand increased privacy regulations. Marketers are seeing the highest ROI from short-form video, personalized ad creative, and omnichannel experiences that blend seamlessly with the customer journey. As third-party cookies continue to be phased out, expect to see a continued reliance on first-party data and highly interactive lead generation tools. ## Frequently Asked Questions (FAQs) ### What's the role of AI in digital advertising today? AI can now drive a significant portion of content creation, as well as optimize ad campaigns. This enables advertisers to create content at scale, gather valuable data, and use that data to make real-time adjustments and boost ad performance. ### How will data privacy regulations affect the future of digital advertising? Laws like GDPR and the CCPA are making it more difficult for marketers to obtain and use personal information. Coupled with this, more users are opting out of cookie consent banners, and ad blockers and private browsing are on the rise. These trends mean that marketers are being forced to shift from third-party targeting to using first-party data. ### What are the best practices for using first-party data in advertising? When you use first-party data as an advertiser, always get clear, informed consent from users. Tell them what data you’re collecting, why you’re collecting it, and how you will use it. Remember to use first-party data to refine your targeting when running ads on platforms like Meta, Google Ads, LinkedIn, TikTok, and Realize. --- ### Walled Gardens: Pros, Cons, Alternatives URL: https://www.taboola.com/marketing-hub/walled-gardens/ Last Modified: 2025-07-15 08:22:35 The marketing ecosystem contains many avenues to reach potential consumers. Each platform has its own benefits, as well as its specific limitations and restrictions. For each product or service, a marketer or advertiser would want to choose the platform which can help them best reach their target consumers. Whether that’s within the confines of a walled garden, or on the open web, will ultimately be down to what the advertiser hopes to achieve. ## What Are Walled Gardens? A walled garden is a platform where the technology provider has full control over what gets displayed. The owner of the platform can prohibit specific advertisers or messaging from appearing in the space. They allow advertisers to reach that platform’s users in a way that might make their ad seem trusted, because the user might already be poking around there for entertainment and education. ## Who Are the Main Examples of Walled Gardens? Apple, Amazon, Facebook, Google, LinkedIn, and TikTok are all examples of walled gardens. ## Walled Gardens: Pros/Cons Pros Cons Allows for highly personalized and precise campaigns. Targeting is limited to the data within the platform. Platforms provide detailed performance measurement data within their walls. Measurement attribution is siloed. Walled gardens restrict external access to user data, protecting their privacy. Don’t provide a complete picture of performance. Improved user experience and engagement because ads feel more native and personalized. Potential channel bias and limited consideration of user experience. ## How Do Walled Gardens Impact Marketing Strategies? ### Limited Targeting Walled gardens like Meta, Google, and Amazon offer extremely rich first-party data, enabling advertisers to build highly personalized and precise campaigns, explains Colby Flood, founder of digital marketing agency Brighter Click. “Their algorithms can effectively match ads to user behavior, interests, and purchase intent, which is great for ad targeting,” he says. That said, unless marketers capture user information (e.g., emails or phone numbers), retargeting across other platforms becomes fragmented and inefficient since, as Flood points out, targeting is limited to the data within the platform. This is where advertising on the open web opens up new opportunities. ### Siloed Attribution These platforms provide detailed performance measurement data within their walls, allowing for fast optimizations — but without the proper third party attribution tools, you may struggle to get a clear image of how you’re performing across platforms. “There has been a decline in third-party cookies and information over the years, and we don’t have the granularity of reporting today that we did in 2019,” says Flood. “Each platform claims credit for conversions, leading to inflated results — for example, Meta claiming 60 sales and Google claiming 50, when only 100 sales actually happened. This skews true blended performance and hinders accurate budget allocation.” ### User Privacy and Data Sharing Walled gardens restrict external access to user data. “This helps protect privacy by avoiding data leaks and misuse,” says Flood. “It benefits enterprise companies looking to keep sensitive customer data from being seen by employees and external vendors that do not need access.” The downside is that, again, it makes it harder for advertisers to see the wider picture. “This closed nature also prevents advertisers from understanding performance across demographics, devices, or channels, limiting insight into holistic performance,” says Flood. ### User Experience User experience and engagement are improved within walled gardens, because ads feel more native and personalized. “From the media buyer's perspective, once someone learns the platform, they can quickly navigate it,” says Flood. That said, there is potential for channel bias and not seeing the full customer experience. “Marketers often gravitate toward the platforms they’re most comfortable with, which can lead to channel bias and a suboptimal customer experience when the full buyer journey isn’t considered,” says Flood. “There is also an issue with marketers not being familiar with the UI of the platform, limiting their ability to access the insights and data needed to perform their jobs.” ## Why Are Digital Marketers Reevaluating Their Use of Walled Gardens? For many marketers, walled gardens — think search and social, for example — are showing diminishing returns. “Marketers must structure campaigns around each platform’s unique setup, which fragments customer journeys and can cause businesses to decide between platform best practices and business needs,” says Flood. “For example, Meta targeting best practice is to consolidate audiences to help reduce CPMs (cost-per-milles), but a business with brick-and-mortar locations may need to run city or zip code-specific campaigns.” Overconcentration of the budget on a single platform, such as Meta, is the primary cause of brands reevaluating budgets, Flood continues. For example: 65-70% of many direct-to-consumer (DTC) budgets go to Meta, as shown in the Northbeam screenshot below: “These recurring platform issues with Meta cause uncertainty in forecasting, which is not ideal,” Flood adds. ## What Strategies Can Brands Use to Succeed in Walled Gardens? “Success requires a strong creative strategy,” insists Flood, including a marketing mix with a range of channels, which is essential to keep reaching consumers on different platforms and in different ways. “As media buying becomes automated, creative becomes the new targeting, signaling relevance to algorithms and customers alike.” Brands should focus on continuous testing of ad creative, using platform-specific formats and messaging, and capturing first-party data (email/SMS) for retargeting, Flood adds. While walled gardens can offer certain benefits, as with any strategy, it’s important to add more to your marketing mix, and including performance advertising on the open web as part of that mix should practically be a given. ## Measuring Campaign Performance Inside Walled Gardens Flood says that the most accurate way to measure performance is to look outside the platform. He recommends these tactics: - Blended metrics like Marketing Efficiency Ratio (MER), which divides total revenue by marketing spend. - Measuring cost-per-acquisition (CPA) and return on ad spend (ROAS) by the breakdowns showing new customers vs. existing customers, to ensure you are not overpaying for repeat purchases, while believing you are acquiring net new customers at a low cost. - Third-party attribution tools like Triple Whale or Northbeam. - Post-purchase surveys to map the actual customer journey. “All of these provide a fuller picture of return on investment (ROI) and prevent over-crediting from in-platform dashboards,” says Flood. ## What Alternatives Exist Outside of Walled Gardens? ### Open Web The open internet, or open web, refers to the rest of the internet landscape, such as websites and apps that are not subject to the limitations of walled gardens. Performance advertising platforms such as Realize, for example, offer access to a large network of premium publisher sites across the open web. ### Programmatic Advertising “Through demand-side platforms (DSPs) like The Trade Desk and StackAdapt, brands can place ads across thousands of websites and apps,” says Flood. ### Retail Media Networks With retail media networks, your customer is already looking around with the idea of buying, so it’s up to you to deliver the right message at the right time in the right way. “Walmart Connect, Target Roundel, and others allow advertising to high-intent shoppers using first-party retailer data,” says Flood. ### Connected TV (CTV) “Platforms like Hulu, Roku, and Tubi offer premium video inventory, great for brand awareness,” says Flood. The three main ways marketers can buy ads on CTV are via direct deals with a streaming service provider, through programmatic buying on automated platforms, or by partnering with companies that make CTV devices. ### Owned Media Sometimes the best marketing is still the content you create yourself. “Email, SMS, blogs, and branded apps give full control and long-term value,” says Flood. ## Key Takeaways Walled gardens can be beneficial to advertisers in a variety of ways, such as providing rich first-party data and detailed performance information. However, these details only exist as they relate to the activity within the specific platform. Walled gardens have both pros and cons, and ultimately, they should be part of a larger marketing mix. ## Frequently asked questions (FAQs) ### What are the latest trends in walled garden advertising? The biggest trend is the shift toward creative-led performance. “Platforms like Meta are openly saying, ‘Creative is the new targeting,’” says Flood. “As automation handles more of the media buying, creative becomes the primary lever for differentiation. ### How do walled gardens relate to the end of third-party cookies? Walled gardens offer first-party data, which can help make up for the demise of third-party cookies, which were used to track user activity across websites, but raised privacy concerns. ### What are the key differences between walled gardens and the open internet? With the open internet, you have freedom as to what you advertise and how you do it, but you might not have access to the specific audience you’d like to reach. The walled garden is a specific arena where you have to follow certain rules about what you present and how you present it. The platform owner controls user data, unlike an advertiser’s interactions on the open web. The advertiser only communicates with consumers inside the walled garden, as opposed to a potentially wider audience on the internet. --- ### 5 Small Business Marketing Trends To Be Aware of in 2026 URL: https://www.taboola.com/marketing-hub/small-business-marketing-trends/ Last Modified: 2026-03-08 13:48:44 Marketing as a small business in 2026 looks very different to how it used to. Rising advertising costs, increased competition for attention, and the rapid shift toward AI-driven discovery mean small businesses must be more intentional about where and how they show up online. Whether you’re running marketing solo or as part of a small team, staying competitive in 2026 isn’t about being everywhere at once: It’s about understanding how customers discover, research, and decide which brands to purchase from, and building marketing strategies that support that journey. ### What’s changed in our 2026 update: - All entries include updated and current information, figures, and stats. ## Trend 1: The Power of Local SEO and Hyperlocal Marketing for Small Businesses Local SEO remains foundational in 2026, but it’s evolved into something broader: local discovery across search engines, maps, social platforms, and AI-generated answers. Consumers are increasingly searching for “near me” businesses while browsing on their mobile devices, and search engines are increasingly favoring businesses with strong, local relevance. 32% of Americans look up information about local businesses online at least once a day, so it’s vital that your business is discoverable by those users. An important change here, though, is that many of these searches now result in no-click actions from users, with search engine AI overviews often meaning that browsers don’t have to visit a website at all to find the information they’re looking for. This makes the basics even more important. Start with accurate NAP (Name, Address, Phone Number) listings online, especially on your Google Business Profile. Optimizing your website for localized keywords that include your city or town, or creating location-specific landing pages, can be helpful in driving organic traffic to your business online. Using structured data markup on your site can also help your business to appear in local search results pages and maps. This information doesn’t only impact rankings in search results, but also shapes how AI tools summarize and recommend local businesses. Local website optimization still plays a vital role in this, too, as location-specific landing pages and localized keywords help reinforce geographic relevance. For businesses using digital advertising, hyperlocal targeting can go a step further with ads being served to specific neighborhoods via geofencing. Combining this approach with intent-based targeting can help you serve ads to customers who are ready to convert in your target locations. This is an especially strong strategy to use if you’re running a brick and mortar business. Don’t forget about reviews, too: Recent reports suggest that over 83% of customers use Google to search for local business reviews, per Bright Local’s 2025 Local Consumer Survey. Responding to any reviews that come in, regardless of where, is an important way to build trust with both search engines and customers. Reviews are also an important way for AI tools to better understand your business and produce recommendations based on what others are saying about your business online. Other key stats from the same survey: - 89% of U.S. consumers were more likely to use a business that responded to reviews. - 83% of consumers use Google Reviews to research businesses. - 48% of U.S. adults use local news outlets as sources for local business reviews. ## Trend 2: Affordable and Effective Social Media Strategies for Small Businesses Outside of managing your own website with an effective SEO strategy, social media remains one of the best small business marketing avenues in 2026, but it’s important to remember that not every platform will offer the same kind of return. For most small to medium businesses, Instagram, TikTok, and Facebook continue to provide the best return on investment (ROI), particularly when your content is tailored to each audience. The most cost-effective way to produce content for these platforms as a small business is repurposing where possible. For instance, turning product ads into reels or a carousel ad on Facebook or Instagram means that you can maximize the shelf-life of your creative. Short-form content still performs well in 2026, with most social media platforms prioritizing quick, engaging clips that can entertain or inform your audience. Livestreams also offer you the opportunity to connect with and build your community, with real-time interaction that increases trust with new and potential customers. Alongside this, however, you should also provide educational and explanatory content, as we begin to see a return to longform video and written content through platforms like YouTube and Substack. User-generated content (UGC) is also a powerful tool for small businesses. Encourage customers to tag your business in their content or share their genuine reviews of your products or services. Resharing this content can also naturally integrate into your existing content calendar, helping you stay visible and relevant across social platforms. UGC is also performing well when incorporated into paid campaigns, reinforcing credibility during the customer consideration phase. Not only that, but this approach fosters community for long-term success in a way that chasing virality doesn’t. Directly engaging with your customers through responding to comments and sharing customer posts increases engagement and sustained growth. Even for those working on a limited budget, social media can be very effective, both paid and organic. Users in the middle of the sales funnel, with awareness of your brand but not quite at a purchase decision, are often the best group to target with paid social ads. Many small businesses can see an average of ROI of around $5 for every dollar spent on social advertising. Social media marketing stats you should know in 2026: - 78% of marketers said that social media has become consumers' preferred customer service channel. - 52% of marketers use funny content as part of their strategy, followed by relatable content, and then behind-the-scenes content. - Facebook remains the most popular social media network worldwide, followed by YouTube and Instagram. - For younger audiences, TikTok is now being used as a search engine, with 77% of Gen Z TikTok users using the platform for product discoverability. ## Trend 3: Email Marketing and Building Direct Customer Relationships for Small Businesses Email marketing has stood the test of time as one of the most powerful tools in small business marketing. With direct access to your audience, there are no algorithms to worry about, but you have to get it right to be effective. In 2026, that means personalization and segmentation with a broader owned-audience strategy. With over half of consumers saying that marketing emails impact their purchase decisions, tailoring your email content to customer behavior and preferences can make a significant impact. You can speak to your customers directly by breaking them into relevant groups like past purchasers, first-time site visitors, or abandoned cart users to see better open rates, improved click-through rates, and more conversions. Using smart automation tools can support this even further, without needing a large marketing team manually programming it all. Messaging is still the most important part of any successful email marketing campaign, whether you’re using automation or not. Educational content, seasonal offers, and new product announcements should all be shared with a personal feeling and provide real value to your customers. Including first names, location information, or purchase history insights can make these emails feel highly tailored to the individual recipient. No matter what type of content you create, it should always be value-driven to help maintain engagement and build long-term customer relationships. Think about how you can organically grow your email marketing list. Discounts for first-time signups are a great strategy for product-based businesses, along with giveaways. Embedding sign-up forms throughout your site is also useful for capturing this data and starting a lead nurturing process that targets your users directly in their inbox. 2026 email marketing statistics you should know: - Email marketing sees an average ROI of $42 for every $1. - 81% of businesses say that email drives customer acquisition. - 64% of small businesses rely on email as their primary customer acquisition tool. ## Trend 4: Content Marketing on a Budget for Small Businesses Content in any form continues to be a vital part of all effective digital marketing strategies in 2026, for businesses big and small, when it comes to attracting, educating, and converting customers. Success, though, depends now on quality, structure, and repurposing rather than simply volume of content alone. This is especially important when it comes to both AI-overviews in search results and AI-generated answers within tools such as ChatGPT. How-to guides, behind-the-scenes views, and customer stories are still some of the most successful pieces of content that resonate with both new and existing customers. Creating without a large budget, though, is still a challenge for many small businesses. Planning and repurposing existing content can help combat some of these issues: For instance, a blog post could become a video script, while an ad could be turned into an organic social post or an email callout. AI-driven copy and image generators are now some of the most powerful tools for helping businesses repurpose this content and streamline the whole process. In 2025, 67% of small businesses were using AI tools for their content marketing, and that number will only continue to grow this year. When it comes to sharing your newly-created content, both owned and paid channels can be effective. Social channels like Instagram and TikTok are ideal for short-form visual content, while written content is best on owned platforms like your website in the form of blogs and newsletters. Not only does this provide a home for your written content, but you can also benefit from SEO growth over time. Discoverability is crucial for a successful SEO strategy, both in traditional search results and AI tools, and content is a critical part of supporting this. Even low-budget content should be optimized for both users and search engines, with focused keyword research, optimized titles and descriptions, image alt text, and internal linking. Alongside any local SEO you may be including, these basic SEO steps are essential for giving your content the best chance of being found and ranked in search engine results pages (SERPs). But, remember the most important factor — at the heart of any good business content, you should be telling the story of your why, what, and how. People want to connect with people, not a faceless business, so every piece of content should be tied back to your brand’s mission and vision, along with providing real-life stories of your customers and how your products or services help them solve a specific problem. Insights from the Content Marketing Institute’s 2025 research shows that: - 1 in 3 marketers have a scalable approach to content marketing. - 54% cited lack of resources as the biggest challenge to content creation. - Only 19% of B2B marketers are regularly using AI as part of their content workflow. ## Trend 5: Leveraging Digital Tools and Automation for Small Business Marketing Efficiency Time is often one of the biggest barriers to entry when it comes to small business digital marketing, but using the right tools to support your goals means you’ll get back some of that time, while also exploring new avenues for growth online. From social media schedulers to AI writing assistants and drag-and-drop landing page builders, there are endless tools out there to help you and your team move quickly and efficiently. Automating repetitive tasks like sending customer emails, advertising optimization and budget management, reporting, or even running A/B tests on your ad creative can all free up time for your team to focus more on the big picture strategy. Automation in 2026 is no longer optional, but an essential tool for saving time, improving efficiency, and making smarter marketing decisions. AI tools are revolutionizing small business marketing and can give you a competitive edge that was previously only available to larger enterprises. CRM systems to consolidate customer data, track behavior, and manage outreach can be leveraged alongside AI tools to better target your customers and personalize every interaction they have with your business. Analyzing performance has also become easier. Real-time dashboards that integrate data from various sources like your ads, website data, email, and social media channels mean that you can make fast, data-informed decisions that impact your bottom line. With the right performance marketing metrics, small businesses can finally move beyond last-click attribution and focus more on assisted conversion and engagement quality, reflecting how your customers actually move through the sales funnel. Key stats around using the right tools that you should know: - Only 23% of marketers are confident that they’re tracking the right metrics. - 80% of marketers are looking to reduce time spent on repetitive tasks by using AI tools. - 68% of marketers report positive ROI from AI investments. ## Key Takeaways Small business marketing in 2026 is all about working smarter, not harder. Local SEO, combined with focused content marketing and social media marketing strategies, can capture the attention of local, high-intent customers while building long-term relationships. Email marketing remains one of the most cost effective methods of reaching middle-of-the-funnel prospects as a small business. Incorporating AI tools into the mix can offer you a strategic advantage that many of your competitors are not yet experiencing. ## Frequently Asked Questions (FAQs) ### What are the most effective digital marketing channels for small businesses in 2026? Local SEO and email, along with social platforms like TikTok and Instagram, are the most effective digital marketing channels for small businesses in 2026. Additionally, longform channels like YouTube are also starting to see a resurgence. ### How much should a small business spend on their digital marketing? A typical small business will spend, on average, 7-10% of their revenue on marketing every year. ### What are some common digital marketing mistakes small businesses should avoid? Businesses who try to capture traffic from every platform often end up spreading themselves too thin. Instead, focused attention should be paid to a handful of channels. Neglecting basic SEO on your website and not paying attention to mobile optimization are also big issues that many small businesses overlook. ### How can small businesses compete with larger companies online? Authentic storytelling and niche targeting are the best ways to compete against larger businesses online. Remember, people buy from people — one of the biggest advantages of being a small business is the personal attention you can pay your customers in a way that a large enterprise struggles to replicate. ### What are the key metrics for measuring success in small business digital marketing? Conversion rate, customer acquisition cost (CAC), return on ad spend (ROAS), email open and clicks, and engagement rates are some of the most critical metrics to track. --- ### Personally Identifiable Information (PII): Leveraging Personalization with Care URL: https://www.taboola.com/marketing-hub/personally-identifiable-information/ Last Modified: 2025-07-10 12:54:13 In digital advertising, personally identifiable information, or PII, plays a complex role: It enables personalization and audience targeting while simultaneously raising serious privacy concerns. Marketers often rely on both direct and indirect identifiers to serve relevant content. However, even anonymized data can become identifiable when combined with behavioral patterns or third-party datasets. As a result, advertisers must walk a fine line between performance and privacy, ensuring their data practices are both compliant and consumer-friendly. ## Defining Personally Identifiable Information Personally identifiable information refers to any data that can be used to identify a specific individual, either on its own or when combined with other information. PII can include obvious identifiers, such as names, phone numbers, email addresses, or fingerprints, as well as indirect identifiers, like IP addresses or device IDs. ### What Are Some Common Examples of Direct PII? Direct PII refers to unique information that can identify an individual without any additional data. Common examples of indirect PII include an individual's: - Full name. - Email address. - Phone number. - Social security number. - Passport. - Drivers license. Direct PII is collected regularly through various channels, including website forms, newsletter sign-ups, customer service inquiries, and account registration. Companies that collect this type of data must prioritize security, as even minor data breaches can result in significant privacy violations. ### What Are Some Examples of Indirect PII and How Can It Become Identifiable? Indirect PII, also known as quasi-identifiers, can’t identify an individual on their own, but can do so when combined with other pieces of data. Common examples of indirect PII include: - IP addresses. - Device IDs associated with smartphones, laptops, and tablets. - Browser fingerprints, which can include a user’s operating systems, browser plug-ins, time zone, and language. - Demographic information, such as race, age, and gender. - Geolocation. There are numerous ways indirect PII can become identifiable when cross-referenced with other pieces of information. For instance, an IP address coupled with an individual's browsing history, device ID, and/or ad interactions can reveal a user’s identity. The ability to associate multiple pieces of indirect PII to establish an individual’s identity is becoming increasingly easy, thanks to advanced algorithms and data brokers that use this data to create user profiles. As such, what is often considered anonymous data can become personally identifiable when aggregated. ### What Are the Potential Risks Associated with Mishandling PII? Mishandling PII can have serious consequences for both individuals and the companies and organizations that collect and use the data. When collecting entities improperly store, share, or access PII, the risk of the following can increase, often significantly: - Data breaches that expose private user information to malicious third parties. - Legal and regulatory penalties for violating laws like the General Data Protection Regulation (GDPR) or the California Consumer Privacy Act (CCPA). - Loss of consumer trust, which can significantly and potentially irreparably damage a brand’s reputation and customer loyalty. - Operational disruptions caused by security investigations, breach notifications, legal engagements, and remediation efforts. - Financial losses from fines, lawsuits, damage control campaigns, and operational disruptions. For companies in the digital ecosystem, where data is central to performance, personalization, and brand growth, the stakes are significantly high. According to recent data, 56% of U.S. adults say that they’re unlikely to share their personal information with a company that has experienced a data breach. 45-54-year-olds represent the most hesitant group, with 76% saying it's not likely they’d share information after a breach. ## Categories and Sensitivity of PII The sensitivity of PII can vary significantly, and companies should utilize variations to inform their collection, management, and storage efforts. Understanding the types of PII can help organizations apply appropriate safeguards and prioritize risk mitigation strategies accordingly. ### What Are the Different Categories of PII? Below are common categories of PII, each representing a unique risk level and subsequent protection strategy: - Contact information: E.g., names, mailing addresses, phone numbers, and email addresses. These are commonly collected in user profiles, newsletter sign-ups, and customer service or account portals. While this information is relatively basic, it can be exploited in phishing efforts or identity theft schemes if not adequately protected. - Financial data: E.g., credit card numbers, bank account details, peer-to-peer payment data, and transaction histories. Financial data is among the most targeted data in cyberattacks and requires encryption, secure payment gateways, and strict access controls. - Health information: E.g., medical records, prescriptions, and insurance policy numbers. When handled by advertisers, particularly those in the wellness and pharmaceutical industries, these details must be managed in compliance with several regulations, including HIPAA. - Biometric data: E.g., fingerprints, facial recognition scans, iris patterns, and voice recognition. Biometric PII is uniquely tied to an individual and cannot be changed if compromised. As such, it’s critical for organizations to store and manage this data with extreme care, relying on advanced encryption and strict access protocols. - Online identifiers: E.g., IP addresses, cookie IDs, device IDs, and browser fingerprints. While these are generally considered indirect identifiers, if improperly accessed, they can be cross-referenced with other data points to identify an individual, especially in targeted advertising contexts. Each category above carries different levels of risk, but if compromised, there can be serious consequences for all parties. Businesses and organizations, in particular, can face regulatory fines, reputational harm, and a loss of consumer trust. Therefore, it’s vital that companies assess the sensitivity of the data they collect and implement tailored data protection protocols that take into consideration both legal requirements and ethical standards. ### How Do Regulations Like GDPR and CCPA Define and Categorize PII? Regulations such as the General Data Protection Regulation (GDPR) in the European Union and the California Consumer Privacy Act (CCPA) in the United States have significantly shaped how organizations collect, store, and use personally identifiable information. Both definitions of personal data are broad, reflecting modern data practices and carrying legal consequences for misuse or non-compliance. Here’s how these regulations categorize and govern PII: #### Direct Identifiers Includes names, email addresses, mailing addresses, Social Security numbers, and passport details. Both the GDPR and CCPA treat these as clearly identifiable data points that require explicit consent and secure handling. #### Online Identifiers Data such as IP addresses, device IDs, cookies, and geolocation data are explicitly considered personal data under the GDPR and are similarly recognized under the CCPA. These identifiers are especially relevant in digital advertising and user tracking, where consent and transparency are essential. #### Behavioral and Profiling Data This includes browsing history, search queries, purchase behavior, and inferred consumer profiles. Both regulations emphasize that even derived or inferred data falls under the umbrella of personal information, particularly when used for targeted advertising or personalization. #### Sensitive or Special Category Data GDPR outlines special categories — such as health data, political opinions, religious beliefs, and biometric identifiers — that are subject to stricter protections. CCPA doesn’t use the same classification, but it still imposes heightened responsibilities for sensitive data. #### Household and Inferred Data (Under CCPA) Unique to CCPA is the inclusion of data linked to households, not just individuals, and inferred characteristics, such as predictions about preferences or behavior, if tied to a consumer profile. Both laws also grant individuals expanded rights over their data, including the ability to access, delete, correct, or opt out of the sale or sharing of their information. For advertisers, compliance requires more than checkbox policies: It demands active data governance, clear privacy notices, consent management, and a readiness to respond to consumer requests. Failure to comply can result in significant financial penalties, legal exposure, and erosion of customer trust, making proactive PII management both a legal requirement and a business imperative. ## PII in Digital Advertising ### How Is PII Used for Targeting, Personalization, and Measurement in Advertising? In modern digital advertising, PII is a fundamental part of strategy and growth, helping brands grow awareness, connect with new and loyal customers, and make data-driven decisions about the products and services they provide. The exact way that advertisers use PII can vary by industry and goals. Here are a few common ways advertisers use PII: - Audience targeting and segmentation: Advertisers use direct and indirect PII, such as email addresses, device IDs, and geolocations, to create targeted campaigns that increase engagement and conversion rates. - Cross-device and cross-platform tracking:With the help of cookies, mobile ad IDs, and fingerprinting techniques, advertisers can better understand user behavior patterns, following a user’s journey across devices and platforms, even when that data doesn’t truly identify the user. - Personalized ad delivery: PII allows for a more dynamic approach to content, enabling brands to cultivate a personalized experience based on a user's preferences, demographics, or past interactions. - Data sharing and third-party partnerships:Ad networks and data brokers can and often do exchange user data to expand reach or enhance targeting efforts. However, this can notably lead to enhanced risk if all parties aren’t actively protecting data. With regulations like GDPR and CCPA, advertisers must treat any data that can reasonably identify an individual as PII. Clear disclosures, opt-in consent, and easy opt-out tools are required to avoid fines and build long-term trust. As privacy concerns grow alongside marketing technologies, the responsible management of PII is more than a legal requirement: It's a key component in any brand reputation strategy. ## Legal and Regulatory Frameworks for PII Understanding the legal landscape is crucial for compliant PII handling in advertising. ### What Are the Key Data Privacy Regulations That Advertisers Need to Be Aware Of? In digital advertising, PII is used in a variety of ways to enhance performance and precision, but each use case carries privacy implications that companies must carefully manage: - General Data Protection Regulation (GDPR): Applies to entities that process the personal data of EU residents, emphasizing the importance of user consent and data protection. - California Consumer Privacy Act (CCPA): Grants California residents rights over their personal information, including access and deletion. - Children’s Online Privacy Protection Act (COPPA): Protects the personal information of children under 13 in the U.S. ### What Are the Fundamental Principles of These Regulations Regarding PII Processing? Both GDPR and CCPA outline key principles that organizations must follow when handling PII. These principles serve as the foundation for compliant and ethical data practices: - Transparency: Businesses must clearly inform individuals about what data is being collected, how it’s used, and with whom it’s shared. - Purpose limitation: Data should only be collected for specific, legitimate purposes and not used in ways that are incompatible with those purposes. - Data minimization: Organizations should collect only the data that’s necessary to fulfill their stated purpose. - Accuracy: Personal data must be kept accurate and up-to-date, with reasonable steps taken to correct any inaccuracies. - Storage limitation: Data shouldn’t be stored longer than necessary and must be securely deleted or anonymized when no longer needed. - Security: Businesses must implement appropriate technical and organizational measures to protect PII from unauthorized access, loss, or damage. - Accountability: Companies are responsible for demonstrating compliance with these principles and must document how they meet each obligation. Combined, these principles reinforce the idea that personal data belongs to the individual, not the organization collecting it. Adhering to them isn’t just about avoiding penalties; it’s about fostering trust and showing respect for consumer privacy. ### What Are the Potential Penalties for Non-compliance with PII Regulations? For advertisers, non-compliance with data privacy laws like the GDPR in the European Union and the CCPA in the United States isn’t just a legal issue — it’s a business risk. Missteps in handling personal data can lead to steep financial penalties, loss of consumer trust, and long-term brand damage. #### Under the GDPR - Administrative fines: Advertisers found using personal data without valid consent — such as targeting users without proper opt-in mechanisms — may face fines up to €20 million ($22.7 million) or 4% of their global annual revenue, whichever is higher. - Lesser violations: Even operational oversights, like delayed breach notifications, can result in fines up to €10 million or 2% of global turnover. - Additional sanctions: Regulatory bodies can also issue orders to pause or permanently halt certain advertising activities that involve unlawful data processing. This could include suspending programmatic campaigns or disabling third-party tracking systems until full compliance is achieved. #### Under the CCPA - Civil penalties: Businesses may incur fines up to $2,500 per unintentional violation and up to $7,500 per intentional violation of the CCPA's provisions. - Private right of action: In the event of certain data breaches, consumers have the right to sue for statutory damages ranging from $100 to $750 per incident, or actual damages, whichever is greater. - Enforcement actions: The California Attorney General and the California Privacy Protection Agency are empowered to enforce the CCPA, with the authority to investigate violations and impose penalties ## Best Practices for Handling PII in Advertising When it comes to digital advertising efforts, it’s vital that organizations create strong protocols that allow for compliance, data security, and brand credibility. As regulations like GDPR and CCPA continue to evolve, advertisers must take an active role in developing privacy-first strategies to safeguard their organization and those they serve. The following best practices can help advertisers collect and manage PII effectively and ethically. ### What Are the Best Practices for Securely Collecting and Storing PII? - Collect PII over encrypted connections (HTTPS) to ensure data is protected during transmission. Encryption helps block man-in-the-middle attacks and secures sensitive user inputs, such as email addresses or form submissions. - Use secure, user-friendly forms that request only the minimum details needed to fulfill the ad campaign’s objective. - Implement bot protection tools, such as CAPTCHA, to reduce spam, prevent automated abuse, and ensure the integrity of incoming data. - Apply role-based access controls (RBAC) to ensure only authorized personnel can view, edit, or export PII. Limit data access according to job function and enforce the principle of least privilege. - Audit access logs regularly to monitor who accessed sensitive data, when, and for what purpose. Unexpected patterns or unauthorized attempts should trigger immediate review and, if necessary, corrective action. - Revoke access promptly when roles change or employees leave. - Require multi-factor authentication (MFA) for all systems that handle PII to add an extra layer of security beyond just passwords. ### What Are the Guidelines for Obtaining User Consent for PII Collection and Use? Advertisers should ensure users know exactly what data is being collected and how it’s being used. Earning trust starts with being upfront and clear. - Use clear, plain-language privacy notices that outline what data is collected, for what purpose, and who it may be shared with. Avoid legal jargon or vague phrases that could confuse or mislead users. - Make consent easy to give (and to withdraw). Avoid default opt-ins or pre-checked boxes. Users should have ongoing control over their data, not just at the point of entry. - Present cookie banners and permission prompts prominently on landing pages and ad experiences. Make sure users can access their settings again later if they change their minds. - Log and store consent records to demonstrate compliance in case of audits or user inquiries. These records should be kept securely and updated whenever consent preferences change. - For sensitive data like geolocation or behavioral history, provide granular controls so users can choose what they’re comfortable sharing. Offering layered privacy options can increase user confidence and reduce opt-outs. ### How Can Anonymization and Pseudonymization Techniques Protect PII? Techniques such as anonymization and pseudonymization help advertisers strike a balance between utilizing data effectively and protecting individual privacy. By minimizing the risk of identification, these methods support responsible data-driven marketing. - Anonymization: Involves permanently removing or altering personal identifiers so the data can no longer be linked back to a specific individual, even with additional information. This is useful for aggregate reporting, campaign analysis, and trend forecasting, while maintaining user privacy. - Pseudonymization: Replaces direct identifiers (like names or email addresses) with artificial identifiers or tokens. While it still allows internal teams to perform segmentation or performance tracking, re-identification requires access to a separate key or reference table, adding a critical layer of security. When implemented correctly, these techniques significantly reduce privacy risks while still enabling valuable insights, especially in audience measurement and attribution. ### What Are the Procedures for Handling Data Breaches Involving PII? - Act fast to isolate the breach: If ad tech systems, CRM platforms, or tracking tools are compromised, disconnect affected components immediately. Work with your IT and privacy teams to contain the issue and stop further access. - Notify users and authorities promptly: If PII collected through ad forms, lead-gen tools, or analytics is involved, disclosure may be legally required within a strict time frame (e.g., 72 hours under GDPR). Coordinate with legal and compliance teams to craft appropriate notifications. - Investigate the root cause: Was the breach tied to a third-party partner, a compromised API, or a misconfigured tracking pixel? Understanding how the data was exposed helps determine whether your ad tech stack or internal workflows need revision. - Remediate and reinforce: After the breach, secure affected systems, update access controls, and review campaign workflows that touch PII. This might include reassessing vendor agreements, tightening form integrations, or retraining your team on secure data handling. A well-documented breach response protocol tailored to your media workflows and ad stack helps reduce downtime and shows regulators and users that you take data protection seriously. ## Key Takeaways PII includes both direct and indirect identifiers that can reveal an individual's identity, so proper handling of PII is critical for legal compliance and maintaining user trust. Regulations like GDPR and CCPA set stringent standards for PII management, and in the event of a data breach, prompt action and transparent communication are vital. Advertisers should adopt best practices, including data minimization, user consent, and robust security measures. ## Frequently Asked Questions (FAQs) ### Is an IP address considered PII? Yes, an IP address is often considered personally identifiable information (PII) because it can be used to identify or track a user across devices or sessions. Under GDPR, IP addresses are explicitly classified as personal data, especially when paired with other identifiers like cookies or device IDs. Advertisers using IP-based targeting or analytics must ensure they have proper consent and safeguards in place. Even if the data seems anonymized, it may still fall under privacy regulations if it’s linkable to a person. ### How long can advertisers legally retain PII? Retention rules vary, but advertisers should only retain PII for as long as it's needed for the specific purpose for which it was collected. For instance, if data supports a campaign, it should be deleted or anonymized after the campaign concludes. Define clear retention timelines for each data type and document them as part of your compliance plan. Keeping data longer than necessary can increase legal exposure and undermine trust. ### What are the rights of individuals regarding their PII? Individuals have the right to know what personal data is collected, how it’s used, and who it’s shared with. They can request access, corrections, deletion, and opt out of profiling or data sales. Advertisers must respond to these requests promptly, particularly under laws such as GDPR and CCPA. A straightforward, coordinated process across internal teams and vendors is essential. ### How does PII differ from non-personally identifiable information? PII includes any data that can directly or indirectly identify a person, such as names, emails, or IP addresses. Non-PII refers to information, such as device type or aggregated trends, that can't identify someone on its own. However, combining non-PII with other data can sometimes make it identifiable. Advertisers should handle any data that could reasonably point to an individual with heightened care. --- ### Social Media Advertising: How It Works URL: https://www.taboola.com/marketing-hub/social-media-advertising/ Last Modified: 2025-10-27 10:21:48 I’m old enough to remember an internet without social media. Back then, there were message boards and chatrooms, but everything changed in the early 2000s, and I’ve watched the evolution of it take shape from its inception. As an advertising copywriter who's spent lots of time trying to figure out how to get people's attention, social media completely rewrote the rulebook. It's not just about screaming your message the loudest anymore; it's about conversations, communities, and finding your people and audience where they already are online. ## Defining Social Media So, what exactly counts as "social media" now, anyway? Social media is about connecting people — and that doesn’t need to happen only on the biggest social websites. Nowadays, there’s endless online platforms and tools that let people and groups share content, interact, and build relationships in their own communities. It’s a massive, constantly evolving digital forum, where conversations happen, photos and videos are shared, and ideas spread like wildfire (for better or worse). It’s been quite a journey to watch. Social media transformed from simple online forums into complex ecosystems that often seem to have a hive mind of their own. There’s a place for all interests and communities and media now — platforms that specialize in everything from short-form video to professional networking and super-niche interests. It’s already altered how we communicate, how we consume information, and basically how society functions. Everyone has a voice, and the landscape of news, entertainment, and, of course, marketing is changed forever. ### What Are The Different Types and Categories of Social Media? Again, the term “social media” has grown to incorporate a whole lot, and it's not a one-size-fits-all thing anymore. There are more than a few different types, each serving a unique purpose. Here’s just a few: Social networks: These are what most people think of immediately — platforms like Facebook or LinkedIn. Their primary function is to connect individuals with each other, letting them build profiles, share updates, and interact inside and outside their circles. Content communities: These are all about sharing specific types of content. YouTube, obviously, is for video. Pinterest is for images. Instagram also started primarily as an image-sharing platform before expanding. Blogs and microblogs: Blogs are personal or professional websites where people regularly post articles or thoughts. Microblogging, like what you’d find on X (formerly Twitter), is about sharing short, frequent updates instead. Discussion forums: Places like Reddit have huge networks of communities where people discuss specific topics. Review sites: Yelp and TripAdvisor fall into this category, letting users share their experiences and opinions about businesses or places. ### What Are the Key Characteristics and Features of Social Media Platforms? Despite their different user experiences, most social media platforms share overall common traits. They're built around user-generated content, meaning the users themselves are creating and sharing the bulk of the information. They’re the ones that drive interaction through comments, likes, shares, and direct messages, and the site usually offers profiles for users to customize and express themselves. Most of all there’s the “social” part, where the platforms encourage the building of networks (or friends, followers, etc). What's interesting, at least to us advertising nerds, is how they often use algorithms to personalize what you see in your feed, trying to show you content they think you'll engage with. This algorithmic aspect is incredibly important for advertisers to grasp, as it directly impacts how you can make sure your content and ads are discovered. ## Popular Social Media Platforms for Advertising Different platforms mean different audiences and different ways to engage, so choosing the right one is important for keeping your goals on track. ### What Are Some of the Most Popular Social Media Platforms Used for Advertising? The biggest names always come to mind: Facebook (and its family, including Instagram), TikTok, YouTube, LinkedIn, Pinterest, and sometimes X. These platforms have wide user bases and robust tools that allow for incredible precision in targeting users. Plus, each platform has its own personality, its own crowd, and its own strengths for advertisers. ### What Are the Unique Features and Demographics of Platforms Like Facebook, Instagram, X, LinkedIn, TikTok, etc.? Taking some time to understand the nuances of the different platforms is crucial for deciding the best place to put your ad dollars. #### Facebook It might seem outdated, but it’s still a beast. Facebook has a broad demographic reach, with strength in detailed audience targeting based on interests, behaviors, and demographics. It's good for almost any advertising goal, from brand awareness to direct sales, thanks to its versatile ad formats. #### Instagram Visually driven, Instagram is a powerhouse for brands that can tell a story with stunning images or short videos. It skews younger than Facebook, with a strong emphasis on lifestyle, but Insta is still great for brand building, product showcasing, and driving engagement, especially for e-commerce. #### X The former Twitter is all about real-time and trending topics. Ads here are often about driving conversations, website traffic, or app installs, and capitalizing on trending hashtags can be effective. That said, its demographic has changed a lot in the past few years, to say the least. #### LinkedIn The professional networking network. If you're a B2B business, or you're looking to reach professionals, LinkedIn is often your go-to. The ability to target by job title, industry, company, and skills makes it ideal for lead generation and brand awareness in the corporate universe. #### TikTok The king of short-form, highly engaging video. Its user base is predominantly younger, and content is often raw, authentic, and trends quickly. #### Pinterest Pinterest is more of a visual search engine for inspiration. Its users are often planning projects, and that makes it excellent for brands in creative, lifestyle, or retail categories, particularly for driving traffic to product pages. ## The Role of Social Media in Digital Marketing Social media shouldn’t be an isolated afterthought in your marketing plan. Make it a central hub that connects to pretty much everything else you're doing. ### How Does Social Media Integrate With Broader Digital Marketing Strategies? Social media ties into your whole digital marketing ecosystem. It can drive traffic to your website (where your search engine optimization (SEO) efforts should already be at work), serve as a customer service channel, fuel your content marketing by distributing blog posts or videos, and even help segment audiences for your email marketing. It's a crucial point in the customer journey, from initial awareness all the way through to loyalty. When I’m working on a campaign, I often tell clients to think of it as the common thread connecting all their online efforts. ### How Can Businesses Use Social Media for Brand Building and Awareness? Social media is really an irreplaceable tool for brand building. Social gives you a platform to express your brand's personality, share your values, and tell your story in a consistent voice. But, you’ve got to keep up with it. By regularly posting engaging content, running awareness campaigns, and interacting authentically, you can build recognition, trust, and create a strong brand identity that resonates. ### What Role Does Social Media Play in Customer Engagement And Community Building? This is where social media truly shines. Social media isn’t just a broadcast channel — it's a two-way street. Businesses can engage directly with customers through comments, DMs, and live sessions. By having constant interaction, you can build a sense of community around your brand with customers who are genuinely interested. Plus, when people feel heard and connected, they become more loyal, often turning into advocates for your product. ### How Can Social Media Be Used for Lead Generation and Sales? Through targeted ads, lead forms built into the platforms, and direct shopping features, you can easily move users down the sales funnel. High-intent users can be captured through compelling calls to action (CTAs), or you can nurture leads by providing content that guides them toward a purchase. ### What Is Social Listening and Why Is It Important for Advertisers? Social listening is the act of monitoring social media conversations to understand what people are saying about your brand, your competitors, and your industry. For advertisers, this is pure gold, and it’s just sitting there waiting for you. Social listening helps you identify trends, discover customer pain points, find unmet needs, and even spot new opportunities for ad campaigns or product development. It's like having a giant focus group running 24/7, giving you real-time insights all the time. Be sure to utilize it. ## Social Media Advertising The free side of social media offers plenty of benefits, but advertising is where you need to start putting some budget behind your efforts to really amplify your message and reach specific audiences. ### What Are the Different Types of Advertising Available on Social Media Platforms? Unlike the old days when you were pretty much limited to print, outdoor, and radio/TV spots, the ad formats on social media are much more diverse, and designed to fit various objectives. You've now got everything from standard image and video ads to carousel ads, collection ads, story ads, and even playable ads for games or apps. Each platform usually has its own unique spins on these, so it's worth exploring what's available where your audience is. ### How Does Social Media Advertising Targeting Work? This is arguably the most powerful aspect of social media advertising. Unlike traditional advertising, social platforms have vast amounts of data on their users, and this allows for precise targeting based on things like: - Demographics: Age, gender, location, language. - Interests: What pages they follow, what content they engage with. - Behaviors: Purchasing habits, device usage, travel patterns. - Custom audiences: Uploading your customer lists to target existing customers or create lookalike audiences. - Connections: Targeting people connected to your page, or their friends. This level of accuracy means you can show your ads to people most likely interested in your product or service, which makes your ad spending way more efficient. ### What Are Some Best Practices for Creating Effective Social Media Ads? Effective social media ads take some creative thinking and lots of tweaking and experimenting. A few things I've learned along the way: - Know your audience: Really dig deep into who you're talking to. - Compelling visuals: Social media is visual-first, so your image (or video) needs to be compelling enough to stop their scrolling. - Clear, creative, concise copy: People scroll fast, so keep it short and meaningful. - Strong call to action (CTA): Tell people exactly what you want them to do. - Test, test, test: A/B test different headlines, visuals, and CTAs. - Mobile-first: Most people are on their phones nowadays, so ensure your ads look great and function perfectly on mobile devices. ### How Do You Measure the Success of Social Media Advertising Campaigns? Track those metrics! Keep a close eye on key performance indicators (KPIs) like impressions, reach, clicks, click-through rate (CTR), engagement rate, and, most importantly, conversions — and align those metrics with your initial campaign goals. Most platforms have built-in analytics dashboards that provide all this data and are a huge help. ## Challenges and Opportunities in Social Media Marketing The social media landscape is constantly shifting, providing both challenges and exciting opportunities all the time. ### What Are Some of the Challenges Businesses Face in Social Media Marketing? One of the biggest hurdles is simply staying current as social media evolves. Platforms update their algorithms, features, and ad policies constantly and it’s really hard to keep up. You’ll need a commitment to continuous learning and adaptation. My advice for staying current is to dedicate time to it: Follow industry blogs, subscribe to newsletters from the platforms themselves, participate in online communities, and regularly experiment with new features. It's an ongoing education that never stops. Ad fatigue is another challenge; users see so many ads that yours need to consistently stand out. That can be a tough challenge to your creative side. Measuring true return on investment (ROI) can also be tricky, especially for brand awareness campaigns where the direct sales impact isn't immediately obvious. Then there's the ever-present issue of handling negative comments or feedback. Try to address them promptly, politely, and professionally — don't get defensive. Acknowledge their concern, apologize if appropriate, and offer to take the conversation offline if it's a specific issue that requires more detail. Turning a negative experience into a positive one through excellent customer service can actually build brand loyalty. Silence, however, just makes things worse. Right now there’s a big push towards creator economy partnerships and influencer marketing. Live commerce is gaining traction, too. Augmented Reality (AR) filters and experiences are becoming more common in ads, and as privacy concerns grow, platforms are investing in first-party data solutions, which means understanding your own customer data will become even more valuable for targeting. Determining how often to post on social media can also be a challenge. There's really no magic number here — it depends heavily on your industry, your audience, and the platform itself. A good starting point is to post frequently enough to stay relevant in your audience's feed, but not so often that you become annoying. Consistency is generally more important than sheer volume. ## Key Takeaways Social media advertising is a powerful, dynamic tool that lets you connect with your audience where they spend a significant amount of their time. It's about understanding the nuances of each platform, leveraging precise targeting capabilities, crafting engaging content, and constantly measuring and optimizing your efforts. While challenges definitely exist in such a rapidly changing environment, the opportunities for brand building, customer engagement, and driving real business results are endless. It's a journey, not a destination, so embrace the learning process. ## Frequently Asked Questions (FAQs) ### What is the difference between organic and paid social media? This one’s important to know, and it’s easy to spot the difference. Organic social media refers to all the content you publish on your social channels that doesn't involve making direct payments to the platform. That includes your regular posts, stories, reels, and interactions with your followers. Organic is more about building up a community naturally, sharing valuable content, and earning reach and engagement over time through the quality of what you put out there. While it’s true that the visibility of organic content is influenced by algorithms, which decide what content users see based on their past interactions and perceived interests and can be slow to get going, it generally builds long-term brand loyalty and community. Paid social media, meanwhile, is exactly what it sounds like: You pay the social media platform to promote your content, whether that's an existing post or a specially created ad. This has its advantages too, allowing you to bypass the organic algorithms to a degree, ensuring your message reaches a specific, targeted audience that might not otherwise see it. Unlike organic, paid campaigns offer immediate visibility and precise targeting options based on demographics, interests, and behaviors, as well as detailed analytics to measure clicks, leads, or sales. I wouldn’t go all-in on one or the other: A smart strategy usually involves a blend of both, using paid efforts to boost key messages and reach new audiences, while organic keeps your existing community engaged. ### What is influencer marketing and how does it relate to social media? Influencer marketing is a powerful branch of social media marketing. It involves collaborating with people who already have a dedicated following and credibility within a specific niche or industry — a.k.a "influencers." These influencers, through their authentic voice and established trust with their audience, can promote your products or services in a way that’s more natural and feels less "sales-y" than traditional advertising. They might create sponsored posts, videos, reviews, or even participate in brand campaigns. If they use and like the product, that honestly comes through. Social media platforms are where influencers build their communities and exert their influence. Without those platforms, influencer marketing as we know it simply wouldn't exist. Brands utilize influencers as an inroad into their highly engaged audiences, and benefit from the influencer's built-in trust and reach. It's proven to be an effective strategy for brand awareness, driving consideration, and even direct sales, especially when targeting younger, more digitally native users who might be more skeptical of traditional ads. The key is finding influencers whose audience truly aligns with your brand and values, making sure the partnership feels genuine to their followers, as audiences can sense inauthenticity. ### What are some key metrics to track for social media performance? Tracking the right metrics is essential for finding out if your strategies are actually working, and that goes for both organic or paid. For awareness, you'll want to look at metrics like reach (the number of unique users who saw your content) and impressions (the total number of times your content was displayed). These alone are going to tell you how widely your message is being seen. When it comes to actual engagement, examine metrics such as likes, comments, shares, saves, and video views. A high engagement rate indicates that your content is resonating, sparking interest and interaction and building a stronger community around your brand. For website traffic and conversions, track measures like click-through rates, website visits driven by social media, and ultimately conversions (purchases, lead form submissions, downloads, etc.). These metrics tie your social media efforts directly to your business goals. Cost metrics like cost-per-click (CPC), cost-per-mille (CPM): cost per thousand impressions, and cost-per-acquisition (CPA) are also vital for paid campaigns to ensure you're getting an efficient return on your ad spend. It’s best to combine and analyze a mix of these metrics for an overall view of your social media performance, allowing you to continually refine and grow your strategy. --- ### Youtube Ads: All You Need to Know URL: https://www.taboola.com/marketing-hub/youtube-ads/ Last Modified: 2025-11-13 12:29:25 YouTube is one of the most powerful digital advertising platforms available today. With billions of users watching videos each month, it offers businesses a unique opportunity to reach targeted audiences through engaging visual content. Understanding YouTube Advertising, how it works, and how to set up effective campaigns can help drive better marketing results and support your overall business goals. ## What Is YouTube? YouTube is a global video-sharing platform owned by Google. It allows users to upload, view, share, and comment on videos across a range of topics, from entertainment to education. For advertisers, YouTube provides tools to reach potential customers with tailored video ads placed before, during, or after videos on the platform. ## YouTube Campaign Setup Setting up a YouTube ad campaign is a relatively straightforward process that occurs within Google Ads. Once you've set up your Google Ads accounts, the platform guides you step by step through creating your ads, choosing a target audience, and setting a budget. By following these steps, you can ensure your ads reach the right viewers effectively. ### Add Your Business Information Start by adding your business name and website to your Google Ads account. Linking existing Google accounts can help you receive tailored recommendations for your campaigns. ### Choose a Campaign Goal Choose from several campaign goals, such as purchases, lead form, and brand awareness. Select an objective that best matches what you want to achieve with your campaign. Choosing the right goal helps YouTube deliver your ads to the viewers most likely to act. ### Create Your Ad Upload your video and write headlines and a description to support it. Google Ads utilizes AI to test combinations of visuals and text to determine what works best for your audience. ### Choose an Audience and Budget During this step, you can set who will see your ads. This is done through various targeting settings, including demographic details, interests, keywords, and placements. Then, decide on your daily advertising budget. Google manages budget pacing to distribute your spend efficiently over time. ### Launch Your YouTube Ad Campaign Complete the process by reviewing your campaign setup and confirming payment details. Set up conversion tracking before launching the campaign to accurately measure the results. If everything looks good, publish your ad! ## What Are the Different Types of YouTube Ads? YouTube offers a range of ad formats to meet different business goals and viewer preferences. Choosing the right ad type for your campaign helps ensure your message reaches viewers in ways that align with how they consume content on the platform. Each format has unique features, costs, and placement options, so understanding them can help you maximize your budget. ### Skippable In-Stream Ads Skippable in-stream ads play before, during, and after videos on YouTube. Viewers have the option to skip these ads after five seconds. Advertisers are only charged when someone watches at least 30 seconds of the ad or interacts with it, whichever comes first. They're a flexible option for delivering longer messages while keeping costs tied to actual engagement. ### Non-Skippable In-Stream Ads Non-skippable ads are up to 15 seconds long and must be watched in full before the viewer's chosen video begins. Because viewers cannot skip these ads, they guarantee the complete delivery of your message. They are ideal for concise, impactful promotions where brand awareness is the primary goal and broad outreach is needed within a short timeframe. ### Bumper Ads Bumper ads are short, non-skippable ads that last up to six seconds. They are designed for quick, memorable messages that reinforce brand recognition without interrupting the viewer's experience for too long. Bumper ads are effective when paired with longer ad formats in a campaign, as they help strengthen brand recall through repetition. ### In-Feed Video Ads In-feed video ads appear as thumbnail images with text in YouTube search results, alongside related videos, or on the YouTube homepage feed. When viewers click the thumbnail, they are taken to the video watch page to view the full content. This format encourages active engagement by allowing viewers to choose whether to watch your video based on their interest. ### Masthead Ads Masthead ads display at the top of the YouTube homepage feed and autoplay without sound for up to 30 seconds. Because of their prominent placement, masthead ads are reserved for campaigns that require wide visibility in a short time, such as major product launches or national campaigns. ### YouTube Shorts Ads Shorts ads appear within YouTube Shorts, the platform's short-form video feed optimized for mobile viewing. These ads are best suited for quick, impactful messages that capture attention in fast-scrolling environments. Shorts ads should include strong visuals and concise messaging to stand out within seconds. ## Tips for Maximizing Your YouTube Campaign Strategy Using YouTube effectively involves more than simply launching ads: Implementing thoughtful strategies can enhance engagement, reduce cost, and amplify the impact of your campaigns. Focusing on creative variety, targeting, and calls to action can help your budget go further while reaching your goals. ### Vary Your Ad Formats Use a combination of ad formats to increase your reach. Each format engages viewers in different ways, so using multiple formats can help reinforce brand messaging across touchpoints. ### Refresh Your Creatives Regularly Update your ad creatives frequently to prevent ad fatigue among viewers. Refresh ads with new visuals, messaging, or calls to action to keep campaigns engaging and maintain performance over time. ### Test Targeting Options Try different targeting settings, such as demographics, interests, and keywords. Experimenting with these settings can help reveal which audience segments engage most with your ads. Review performance data and adjust targeting to ensure your campaigns remain effective. ### Include Strong Calls to Action Every ad should tell viewers what you want them to do next. Add direct and specific calls to action to increase the odds of viewers taking your desired action after watching your ad. ## YouTube Targeting and Audience Best Practices Targeting on YouTube enables you to reach viewers who are most likely to be interested in your products or services. Using the right targeting options ensures your ads are shown to people who fit your ideal customer profile, which can improve engagement rates and the return on your advertising investment. ### Define Core Demographics Narrow your audience using demographic filters such as age, gender, location, and language. Focus on viewers who match your target market to avoid spending your budget on people less likely to engage with your ads. Selecting the right demographics builds a strong foundation for the rest of your targeting. ### Use Advanced Targeting Reach viewers based on their interests, habits, and recent online behavior. Select affinity audiences to connect with individuals who share specific lifestyles or interests. Use custom intent audiences to target viewers who are actively researching topics or products related to your business. Applying these advanced options helps you engage people closer to making a purchase decision. ### Expand Reach With Similar Audiences Grow your audience by targeting similar audiences. YouTube identifies new viewers who share characteristics with your existing customers or website visitors. This feature helps you find more potential customers while maintaining relevance and targeting efficiency. ### Use Target-Specific Placements Choose specific YouTube channels, videos, or partner sites where you want your ads to appear. This ensures your ads show alongside content that matches your message and audience interests. Targeting placements strategically increases the chances that viewers will engage with your ads. ## YouTube Ad Creative Best Practices Creating strong ad creatives is essential for YouTube campaign success. Your ads should grab attention quickly, deliver your message clearly, and inspire viewers to act. Follow these best practices to improve your ad performance. ### Capture Attention Quickly Grab attention within the first few seconds. Use bold visuals, direct openings, and strong audio to make viewers want to keep watching. Start with a clear hook that connects directly to your product or message to prevent viewers from skipping your ad. ### Keep Video Length Appropriate Make sure your video length aligns with what viewers expect for each ad format. Short ads are ideal for quick reminders or brand awareness, while longer formats provide space to tell a more comprehensive story. Each YouTube ad type has time limits. Keep your videos within those time limits to ensure they run properly and keep viewers engaged without feeling rushed or dragged out. ### Adapt Creatives for Each Format Tailor your ad creatives for their specific placements. Use vertical videos for Shorts ads to fill the mobile screen and horizontal videos for in-stream ads to match standard viewing formats. Adjust visuals, text, and calls to action to ensure clarity and effectiveness on both desktop and mobile devices. ### Write Clear Copy Craft headlines and descriptions that are direct, benefit-focused, and easy to read. Avoid jargon or complex phrasing that may confuse viewers. Use simple language to highlight why your product or service matters to them. Always include a strong call to action to guide viewers to their next step. ### Use AI to Optimize Creatives Leverage AI tools within Google Ads to improve your ad creative performance. AI can analyze which headlines, visuals, and calls to action drive the best results with your target audience. Apply these insights to test new variations and refresh your ads regularly without relying solely on manual guesswork. ## YouTube Ads Measurement Measuring your campaign's performance helps evaluate its effectiveness and identify areas for improvement. As digital video ad spending in the U.S. continues to rise, reaching an estimated $72.4 billion in 2025, it's more important than ever to track key YouTube metrics to ensure your ads perform competitively. ### View Rate View rate measures the percentage of viewers who watch your ad after it appears. The average YouTube view rate is 31.9%. A higher view rate indicates that your ad is capturing attention effectively. If your view rate isn't where you'd like it to be, test new hooks or visuals to make your ads more engaging. ### Cost Per View Cost per view (CPV) indicates the amount you pay when someone watches at least 30 seconds of your ad or interacts with it. The average CPV for YouTube ads is approximately $0.026. ### Click-Through Rate Click-through rate (CTR) measures how often viewers click your ad after watching it. The average CTR for YouTube ads is about 0.65%. A strong CTR suggests your message and call to action are motivating viewers to learn more or take the next step with your business. ### Cost Per Thousand Impressions Cost per thousand impressions (CPM) shows how much it costs to display your ad 1,000 times. Recent data shows YouTube CPMs cost around $3.53 on average. Monitoring your CPM helps you determine if your ads are reaching viewers cost effectively compared to broader industry averages. ### Video Completion Rate Video completion rate (VCR) shows the percentage of viewers who watch the entire ad. High completion rates indicate that your creative is maintaining attention from start to finish, strengthening brand recall and message delivery. If completion rates are low, adjust your pacing, hook, or messaging to improve engagement. ## How to Optimize YouTube Ads Performance Optimizing your YouTube ads can improve results and maximize your advertising budget. By continuously testing and adjusting your ads, your campaigns are more likely to stay effective as viewer behavior, competition, and costs change over time. ### Review Data to Improve Creatives Review your campaign performance data regularly to identify which ads drive the highest view rates, click-through rates, and conversions. Use these insights to update underperforming ads with new visuals, messages, or calls to action. Refreshing your creative based on data keeps your ads engaging and prevents ad fatigue. ### Adjust Targeting for Better Results Analyze which audience segments generate the best engagement and conversion rates. Refine your targeting to focus on these high-performing groups while excluding segments with low performance. This makes your budget more efficient by concentrating spending where it delivers the most impact. ### Review Bidding Strategies Regularly Check your bidding approach to ensure it aligns with your campaign goals and current market conditions. Switching between bidding types — such as cost per view and target cost per action — can help you improve cost efficiency and achieve better results, depending on your objectives. ## What Are Common Pitfalls to Avoid in YouTube Advertising? Even well-planned YouTube campaigns can underperform if you overlook common mistakes. Understanding these pitfalls will help you to create ads that engage viewers effectively and use your budget wisely. ### Ignoring Mobile Optimization Most YouTube viewers watch videos on their phones. Failing to design ads for mobile can lead to poor performance. Ensure your visuals are clear, your text is large enough to read on small screens, and your calls to action are easy to tap on mobile devices. ### Keeping Ads Running Without Refreshing Them Running the same ad creative for too long can cause ad fatigue, where viewers become accustomed to your content and stop engaging with it. Refresh your creatives regularly with new visuals, messages, or calls to action to maintain viewer interest and performance. ### Using Weak Calls to Action Ads that don't clearly indicate what viewers should do next will fail to drive results. Include a direct and specific call to action with each ad to guide viewers to the next step. ### Overloading Ads With Information Including too much detail in a single ad can confuse viewers and weaken your message. Focus on one clear idea or benefit in each ad to keep it simple, direct, and easy for viewers to remember. ## YouTube Campaign Budget Your YouTube ads budget plays a key role in campaign planning and success. A clear budget helps control costs while ensuring your ads achieve their goals. When you understand how YouTube ad pricing works and the factors that influence it, you can create campaigns that reach the right viewers without overspending. ### Understand What Affects Your Ad Costs YouTube ad costs depend on several factors such as your bidding strategy, campaign goals, ad type, targeting options, audience size, and seasonal demand. For example, running ads during peak seasons like the holidays often costs more because of higher competition. Knowing these factors helps you plan your budget better and avoid surprises. ### Allocate Budget to Match Campaign Goals Consider your goals when deciding how to split your budget. If building brand awareness is the priority, consider investing more in formats like skippable in-stream ads to reach a wider audience. For driving quick conversions, dedicate a portion of your budget to Shorts ads or in-feed video ads to capture attention in shorter formats. ### Set Daily Limits Daily budgets keep your spending under control. Google Ads uses budget pacing to spread your spend strategically across the campaign period, ensuring your ads appear when viewers are most active. Regularly checking your budget settings ensures your campaigns remain aligned with your broader advertising goals. ### Apply Frequency Caps to Improve Efficiency Using frequency caps limits how often the same viewer sees your ads. This helps prevent overspending on repeated impressions that may not lead to additional results. Frequency caps also make your budget go further by reaching more viewers, rather than showing ads repeatedly to the same people. ## YouTube Bidding Strategies The bidding strategy determines how much you pay for ads and how effectively they reach your audience. Each approach supports different goals, so understanding these options helps you choose the best fit for your campaigns. ### Use Cost Per View to Build Awareness CPV bidding charges you only when someone watches at least 30 seconds of your ad or interacts with it. This approach is effective for building brand awareness because you only pay when viewers show real interest. It's a good option if your main goal is to efficiently reach as many people as possible. ### Choose Cost Per Thousand Impressions for Broad Reach CPM bidding focuses on maximizing your ad's visibility. You pay based on the number of times your ad is shown, regardless of whether viewers engage with it. Use CPM when your priority is to increase brand exposure and reach a large audience quickly. ### Optimize Conversions With Target CPA Target CPA bidding helps you drive specific actions, such as website sign-ups or purchases. You set the average amount you're willing to pay for a conversion, and Google Ads automatically adjusts bids to get the best results within that goal. This strategy is most effective when you have a clear conversion objective and want to control acquisition costs. ## YouTube Ad Compliance and Approval Before launching a YouTube ad campaign, ensure your ads comply with YouTube's policies and approval requirements. Following these guidelines helps prevent delays, disapprovals, or account issues that can impact your campaign results. ### Copyright Violations Use only music, images, video, and other creative assets you own or have licensed properly. Uploading ads with copyrighted materials you don't have the rights to can lead to rejections or even account penalties. Double-check usage rights for every asset included in your ads before submitting them for review. ### Content Policies Review YouTube's advertising policies to ensure your ads do not include misleading claims, prohibited products, or restricted topics. For example, ads cannot promote certain healthcare products, adult content, or financial schemes that violate platform standards. Stay up to date with these guidelines to avoid rejections and keep your campaigns running smoothly. ### Approval Reviews After submitting your ads, YouTube reviews them to confirm compliance with all guidelines. Most ads are approved within one business day, but more complex content may take longer. Plan ahead for the approval window to ensure your campaigns launch on time, without unexpected delays. ## YouTube vs. Other Digital Marketing Platforms YouTube stands out from other digital marketing channels because it combines the power of video storytelling with precise targeting options. Unlike static image or text ads, YouTube ads engage viewers through sound, motion, and narrative, making it easier to build emotional connections and explain complex ideas clearly. Its integration with Google Ads allows you to target specific demographics, interests, and search behaviors. You can reach viewers while they watch content that entertains or educates them. Reaching people in these moments increases your chances of engaging those already interested in similar topics or products. When used in conjunction with display advertising on the open web, as well as search and social media ads, YouTube adds a strong visual component to your marketing strategy. It helps reinforce your brand message across multiple platforms and moves viewers closer to the action. ## YouTube Trends to Know for 2025 YouTube continues to grow as both a content platform and an advertising channel. Changes in how people watch videos and interact with brands are creating new opportunities for marketers in 2025. Awareness of these trends can help you maintain a competitive edge in a crowded space. - YouTube Shorts growth: Shorts now average over 200 billion views daily, nearly a 185% increase since mid-2023, making them a dominant format for short-form, mobile-first content. - New AI tools: YouTube is rolling out Veo 3 to help creators generate AI-enhanced videos for YouTube Shorts. While the tool is designed for creators, it raises the bar for video quality. Brands and advertisers must consider their creative strategy to stay competitive. - Follow-on views optimization: YouTube's Demand Gen campaigns now include optimization for follow-on views. This targets people likely to watch more videos from your channel after seeing an ad, helping brands grow organic engagement. ## Key Takeaways YouTube advertising offers a powerful way to reach targeted audiences through engaging video content. Your ads perform best when you know what you want to achieve and create messages that resonate with viewers. Review your campaigns regularly to measure results and identify areas for improvement. Stay up-to-date on trends and adjust your strategies as needed to keep your YouTube ads competitive and drive better results for your business. ## Frequently Asked Questions (FAQs) ### How does video SEO relate to YouTube ads? Video SEO improves the visibility of your videos in YouTube search results and suggested videos. Optimizing titles, descriptions, and other elements makes it easier for people to find your content organically. Even if you run ads, good video SEO supports your campaigns by building stronger ad placement within YouTube's algorithm, increasing overall views and reinforcing your brand presence. ### Why was my YouTube ad rejected? YouTube rejects ads for several reasons, including policy violations, the use of copyrighted content without permission, or the inclusion of misleading claims. Common issues include promoting restricted products, failing to meet technical requirements, and using unlicensed music. Before resubmitting, review YouTube's ad policies and make any necessary changes. ### How can I use YouTube Shorts for advertising? YouTube Shorts are short-form videos up to 60 seconds long, designed for quick viewing on mobile devices. To advertise effectively with Shorts, create videos that feature strong hooks within the first few seconds and maintain clear, direct messaging. Use captions or on-screen text to reinforce your call to action since many viewers watch without sound. Shorts ads are best for short, impactful promotions that grab attention quickly. --- ### Driving Direct Response Sales on the Open Web: E-comm Insights from Xevio (Part 1) URL: https://www.taboola.com/marketing-hub/direct-response-sales-open-web/ Last Modified: 2026-04-12 14:48:49 As marketers continue to look for alternatives to the diminishing returns of search and social, the open web is becoming an ever-more appealing option. To help you get the most out of this channel, I talked with Xevio co-founder and CEO, Nadim Kuttab, an advertising veteran who’s worked closely with Taboola to bring affiliate, brand, and arbitrage spend to new heights globally. Having grown a multi-million-dollar native ads agency, built out the premier native advertising community on Native Hub, and spent hundreds of millions of dollars on ads globally — as well as working on a sizable chunk of the largest Taboola ad accounts out there — it’s safe to say that Kuttab has advice worth listening to. ### Why is the open web, and particularly native advertising, becoming a dominant channel for e-commerce direct response? It's always been a good channel, it’s just that most people have never fully understood how to use it! It's becoming more dominant now because there are a lot more direct response e-comm brands, and the traditional channels where people start, whether it's search or social, are becoming harder — they're becoming less predictable, and if you're building a business, one of the things you hate most in life is unpredictability, right? You want to know that you can spend X and get Y in return, and native is exceptionally good at that. We’ve had dozens of cases over the last two years where e-commerce brands that worked with us really saw consistent and sustained revenue and growth from Taboola, but also from native ads and performance display advertising in general. So, I just think that people are becoming more aware of it. As I said, it's always been good, it's just becoming more popular now. ### What are the core differences in strategy when approaching e-commerce direct response on the open web, versus social platforms? On social, you're competing for attention with a lot of other people, so you have to be very creative — you have to have catchy videos, or very eccentric or colorful things that catch the user's attention while they're scrolling. Social media feeds are also tailored towards the user behavior of that person — channels like Meta and TikTok are very good at finding you the right people. The downside of this is that it's often the same people for a lot of different brands, and therefore the prices can be quite high, especially during peak seasons. With the open web, it's very different: You don’t target individuals, you target placements. If you’re an individual on Meta who has a user profile, Meta knows what you like and what you don't like, what you click on and what you don't click on. On Taboola, we're targeting placements, so you might not read the same websites every day, and therefore we might not consistently reach you, but we will definitely reach you, because you’re reading on the open web. That’s why I’d say that generalistic approaches always work better, because it's about reaching as many people as possible, finding the right people in that mass of users, then bringing them into your content-rich environment, where you can present them with ideas and sell them your product. It's a very different form of advertising, because one is very platform-assisted in targeting an individual, and the other is essentially mass media. That's also why products that have a broader appeal tend to work well on Taboola. ### Can you share a compelling success story of an e-commerce brand achieving significant sales growth with native ads? Yes! We had probably the craziest e-commerce case I've ever seen as a marketer, where a brand went from low seven figures in revenue to nine figures in one year. I could barely believe it myself! Taboola was responsible for close to 100 million of that, on a first-click basis, which goes to show the actual power of native if used right. This brand didn't just do native correctly, though — they did everything else correctly, too. They built their strategy on native, but then made sure that the people who were coming in were caught on retargeting campaigns, on search campaigns, on shopping campaigns. They did a phenomenal job with email retargeting and remarketing: They have a fashion product that you don't expect people to continuously buy or re-buy, but we saw a ton of repeat sales, which obviously drove up their LTV. ### How do you identify and target high-intent e-commerce buyers within the vast network of the open web? We always start with the open web ad space and then transition to Meta, etc. Most brands will do it the other way around: They'll have Google and Meta campaigns, but what they should look at is, what approaches are working on a broad network? Then build content around that. ### What role do engaging content experiences play in driving immediate purchases on native platforms? Everything! Taboola is a content discovery platform, in essence: It recommends content, whether it's organic content from publishers or paid content from advertisers. So, say you don’t create content: You’re just sending people from a Taboola ad to the homepage of a shop with 600 products. People will not buy, because you have no story — you're essentially just throwing things out there and hoping somebody picks it up. That's not how the world works, especially nowadays, where people have so many choices and options that allow them to make more informed decisions. If you do it right, though, good content on Taboola can mimic the quality of a clickout from branded Google search, which is wild — you're taking a cold user and warming them up to a point where they want the product. So, yeah, engaging content is everything. We do not have campaigns succeeding on Taboola right now without engaging content. And we spend six figures a day, easily! ### How does the ability of Realize (Taboola’s performance platform) to offer diverse inventory and placements benefit e-commerce advertisers? So, I was the first guy to criticize this update and be like, "Oh, no, they should stick to what they know!" But, I was wrong, and I'll admit it. I checked yesterday and our top ad on the case I just described is no longer native, it's display, and with a 30% lower CPA than our top native ad. Display offers real potential for growth. The ads aren't aggressive, they're just very direct and they work exceptionally well when done right. Really, these display placements can be incredibly powerful by just having direct messaging that appeals to people. Again, don't sell a product, sell a problem. ### And in terms of being able to place those display ads in multiple spots and in different styles — static, motion, etc. — how much does that affect your success? The fact that Realize gives us the ability to play around with so many different ad formats is a huge asset for the good marketers and media buyers out there. The truth is that the big agencies and brands are going to be slower at iterating and experimenting with these new formats, so if you get ahead of that curve, you can benefit from much less competitive inventory that performs just as well. ### What's the most critical factor for an e-commerce brand to succeed with direct response campaigns on the open web? Let's start with the basics: You need to have a product that people can buy immediately. If you're going to sell a $16,000 scooter, that’s probably not a direct response play on Taboola. I hate to burst the bubble of the scooter brand owners, but it's just not going to work! We found that 70-120 bucks is the sweet spot. You can go higher, but you need to expect substantially higher CPAs. It needs to be direct to consumers, not selling some networking equipment to businesses — the audiences are too broad on Taboola for that. If you're running at scale on other channels, you also need to have an attribution tool in between — whether it's Triple Whale or Klar or something else, it doesn't matter, you just need to have a system in place to assess first and last click attribution properly, because every platform will take as much credit as they can. If you're only optimizing on each individual platform independently, you're not going to get anywhere. After that, it's retargeting on other channels. If you're only looking at the last click, you're missing out on probably three or four times as many sales on a first-click basis just by not properly retargeting, doing email marketing, Google search marketing, or Meta. As with everything in the e-commerce space, it's always a mix: If you’re only focusing on one part of something, you’re missing out. To see success on e-comm, you need to do everything right. --- ### Ideas to Enhance Your Fourth of July Marketing Campaigns URL: https://www.taboola.com/marketing-hub/fourth-july-marketing-ideas/ Last Modified: 2025-09-01 08:27:32 Fourth of July in the U.S. is the summer’s biggest holiday, with holiday and vacation time carved out for many workers. It’s a one-day event that often becomes a long weekend, where friends and family gather to eat, drink, and watch fireworks. 87% of respondents in a recent survey said they celebrate the Fourth of July, and it’s the second most popular American holiday after Christmas. For performance marketers, this has also become a shopping event, with a range of industries selling products relevant to the holiday. Shoppers are busy during this period: Americans spent an estimated $15.5 billion for the Fourth last year. Marketers can set off their own fireworks — in the form of great performance metrics — with some fresh ideas and well-timed campaigns. ## 5 Ideas to Enhance Fourth of July Marketing Performance ### 1. Focus on Your Industry’s Strengths A long weekend marking the peak of summer, with fireworks, food, and fun, is especially suited for some industries in particular. Retail and automotive are some of the biggest winners during Independence Day. Consumers are primed to spend — especially with the long weekend, which gives people more time to research and buy. Think of items like grills, outdoor furniture, pools and pool toys, lawn chairs, and all the other accessories that make summer memorable. It’s a time for cookouts, picnics, and parties, so any product that enhances those moments can win big. You’ll also see success in categories like travel and hospitality, as people plan getaways or events around the long weekend. Consider localized promotions, depending on what makes sense for your brand — think fireworks, parades, and event tie-ins. CPG and beverage brands (especially alcohol and soft drinks) have a natural fit around the Fourth. Capital One Shopping found that wine and beer are the top drinks purchased, with nearly 42% of Americans buying alcoholic beverages for their Fourth of July celebrations last year. It’s a golden moment for high-ticket items and major discounts that can drive urgency, so plan for offers, discounts, and creative copy and imagery that target this event. ### 2. Time It Right for Maximum Sparks For short-term events like this, start awareness efforts early. Consumers increasingly start their purchase planning far in advance for these types of events, knowing that there will be deals available for a short window of time. Think of Amazon’s Prime Days, where marketers spend weeks priming their audience with notifications, influencer summits, early deals, and cross-channel buzz. That same playbook can apply to the Fourth of July. Great marketing starts weeks ahead, so the brands that win are creating anticipation and generating urgency by using teaser messaging, creating “countdown to savings” or early-access content, airing influencer previews, and creating pre-event wish lists. By the time the actual day arrives, the best campaigns have already moved their audience from awareness to intent — and now just need to capture the conversion. If you’ve done your up-front work, the purchases should build up nicely on Independence Day itself. Early momentum is key to get ahead of competitors. Of course, marketers shouldn’t forget about post-holiday momentum. Often there’s a second wave of opportunity right after the Fourth, as people return to their routines or realize they missed a deal — and that there’s still plenty of summer left for seasonal purchases. Consider running extended promotions (as you would for Black Friday through Cyber Monday), remarketing to browsers who didn’t convert, or using first-party data collected during the holiday to inform your next campaign push. ### 3. Channel Your Channel Efforts Make sure to avoid single-channel thinking on the Fourth of July. Success doesn’t come from just one format or platform: Use a multi-touch strategy across the open web, social, email, and native placements to ensure you’re meeting your audience where they are in this short timeframe — and nudging them from interest to action. A cross-platform presence can help spread your messages, reaching consumers through email, social, native display, and video for a full-funnel impact. In addition, for a winning Fourth of July strategy, think mobile-first, not just mobile-responsive. Prospects are out of office, away from their desks, and often traveling. Prioritize mobile optimization before desktop and tablet to ensure a frictionless experience for all your shoppers. SMS shouldn’t just mirror email campaigns — it should reinforce urgency. Use text messages to remind shoppers about expiring, time-sensitive deals, last shipping days, and exclusive offers. Many consumers wait until the last minute, and a one-day event goes by quickly, so SMS can cut through the noise and drive swift action. Boost your text marketing impact after a purchase with real-time updates and tracking info. A well-timed SMS can turn hesitation into conversion, ensuring your brand captures every possible sale. ### 4. Bring In the Fun Leaning into patriotism, community, and summer vibes are all effective strategies for marketers. Campaigns that celebrate America, barbecue culture, and togetherness tend to resonate. One standout was Pepsi’s 2023 “#BetterWithPepsi” campaign, which paired the brand with iconic July 4th foods — even poking fun at classic burger chains to drive conversation. In addition to these themes, it’s a great time to test new creative, like “freedom to save” or “independence from high prices” to open up playful, engaging narratives. But, avoid being generic. Don't just slap a flag on your creative; spend some time making it stand out with strong copy and imagery beyond stock photos. Make sure the tone and visuals actually speak to the audience’s mindset and behaviors during the holiday. ### 5. Think Beyond the Consumer The Fourth of July has traditionally been a major B2C moment, but its marketing potential extends beyond just consumer-facing brands. While it’s undeniably a high-impact sales period — especially in industries like retail, automotive, and home goods — it also represents a cultural milestone: a celebration of summer, freedom, and community. For many consumers, it's become shorthand for “big deal season.” Consumers might say, “I’m waiting for Fourth of July sales,” before making a significant purchase, whether that’s a car, mattress, or even home improvement essentials. It’s a powerful opportunity for brands to tap into seasonal mindset shifts, especially for products tied to life milestones or summer-driven behavior. Even in B2B, companies can tie into the season with thoughtful storytelling, thematic content, or creative brand building that aligns with the spirit of independence and momentum. ## Key Takeaways Fourth of July in the U.S. is a big summer holiday, and performance marketers can treat it like a one-day shopping event that can drive a spike in conversions and revenue. Build awareness ahead of time, test relevant creative, and use all available channels wisely to turn conversions around quickly for Independence Day purchases. ## Frequently Asked Questions (FAQs) ### Are there any specific demographic segments we should target? Within the U.S., there are a range of demographic segments to consider. Ideally, you should first use the audience data you’ve gathered to create segments that make sense for your business. But, broadly, in terms of age, younger generations like Gen-Z and millennials are more likely to more vigorously celebrate the Fourth. Other demographics to consider are high spenders, recent customers, and active loyal customers. Keep in mind that different regions of the U.S. may have different Fourth of July traditions, or be more likely to attend a parade or fireworks event. ### What kind of sales or promotions will be most appealing? Many industries can promote sales and promotions for the Fourth of July, including CPG, beverage, retail, automotive, and travel. Across all of these, consider sales or promotions that use patriotic themes, highlight summer essentials, and offer limited-time details to capture the short timeframe. You might try flash sales, patriotic or U.S.-themed promo codes, or free shipping, as some examples. Also consider bundles or a free giveaway or add-on for a large purchase like a grill or outdoor furniture set. ### Which marketing channels will best reach our target audience for this holiday? All your usual channels will come in handy as part of your marketing strategy for the Fourth of July — email, social media, paid ads, and text messages. Keep a close eye on local opportunities for this holiday, such as any tie-ins with parades, fireworks, or other events. You might consider print ads and QR codes, sponsorships, or offers that align with local celebrations. --- ### Insurance Marketing Trends for Digital Advertisers in 2026 URL: https://www.taboola.com/marketing-hub/insurance-marketing-trends/ Last Modified: 2026-03-16 12:01:10 While the insurance industry has always been competitive, the last few years have seen a rapid transformation of the landscape of digital advertising. This only complicates things further for insurance professionals. With rising customer expectations, tighter data privacy regulations, and soaring acquisition costs, insurance marketers must rethink the way they reach and convert audiences. In 2026, it’s important for insurance marketers to meet customers where they are. Here are some of the top trends shaping insurance marketing this year. What’s changed in our 2026 update: - All entries include updated and current information and advice. - All stats and figures updated with new and current information. - All information on data privacy laws updated in line with current information. ## Trend 1: The Rise of Personalized Customer Experiences in Insurance Personalization has gone from being a “nice to have” to a non-negotiable in insurance marketing. In 2026, personalization is no longer a differentiator on its own, though: Relevance, timing, and tone are what set brands apart. Customers have begun to expect tailored offers at each step of a long, multi-touchpoint buying journey. Here are some ways insurers are personalizing the customer experience: ### Hyper-Personalization Using Data Traditionally, advertisers have relied on demographics to personalize ads, but growing privacy concerns and restrictions around third-party data have accelerated a shift away from identity-based targeting. In 2026, leading insurance marketers prioritize privacy-safe, signal-based personalization instead. Instead of demographics, insurance marketers now harness: - Behavioral signals: These data points are based on user actions, such as page scroll depth and time on site. - Intent data: Brands can look at user behavior to predict when people may be actively considering something they sell. - Custom landing pages: Dynamic landing pages adjust based on the visitor. That means customers see information that matches their interests. Insurance can be an emotional purchase: Consumers want to know they can trust a company to follow through if they ever have to file a claim. Personalization helps you earn trust even with customers who are still in the early stages of shopping for plans. The power of personalization is increasingly showing up in consumer preference surveys, too. Here are a few stats that show the importance of personalizing the insurance experience: - 71% of consumers expect personalized interactions with brands, and 76% express frustration when personalization is missing. - 53% of customers say they’re willing to share personal information if it makes their experience smoother and more relevant. - 93% of customers say a brand will lose their trust if they mishandle their data. - Up to 88% of insurance customers indicate that they expect tailored experiences from insurers, and personalization correlates with higher loyalty and satisfaction. By personalizing your insurance marketing, you reduce friction for customers researching various options by offering only the most relevant information. Customers face a long, multi-touchpoint journey on their way to a purchase, and will shop around to find the best ideal, making it more important than ever to capture their attention quickly. You also have the ability to demonstrate empathy as a brand. For instance, if someone is shopping renter’s insurance for a first apartment, you have the opportunity to deliver messaging that mentions that user’s pain points. As valuable as personalization can be, it comes with an important caveat: Brands need to balance the value of personalization with the demand for data privacy. You can do this by having transparent data-use policies and making it easy for customers to opt out. ## Trend 2: Leveraging AI and Automation in Insurance Advertising Artificial intelligence (AI) is transforming the way insurance marketers operate. In 2026, AI has shifted from an experimental advantage to an operational necessity across insurance marketing teams. Instead of working for hours, crafting individual campaigns for different audiences, you can invest in innovative solutions that boost efficiency, sharpen your targeting, and unlock new creative potential for lead generation. Some top stats on AI and automation in insurance include: - 90% of insurers report being in some state of AI adoption, with 55% already in early or full implementation of generative AI technologies. - The majority of insurers are deploying AI, but only 7% have fully scaled systems across the enterprise. - 62% of survey respondents report experimenting with AI agents and 23% say they’re starting to scale them in some functions. - Across industries, AI-driven search ad spending in the U.S. is expected to grow to nearly $26 billion by 2029, up from just over $1 billion in 2025. Source: EMARKETER Here are some top ways insurance brands are using AI and automation to track down new customers. ### AI-Powered Automation Both insurers and marketers are increasingly incorporating AI into their lead generation processes. One of the biggest ways the technology can help you is in refining your audience targeting: Instead of targeting based on demographics and historical data, AI can base targeting on real-time browsing behavior to ensure you reach customers based on what they’re interested in now, not three months ago. In 2026, this targeting is increasingly powered by aggregated, privacy-safe behavioral and contextual signals, rather than individual identifiers. It can even predict future behavior, to a certain extent. AI can also supercharge your ad bids. By adjusting your bids in real time, AI can more effectively manage conversions and lower your acquisition cost, making your ads more efficient. French insurance company GMF, for example, utilized the AI-powered Maximize Conversions bidding solution, alongside the implementation of first-party data for targeting, to see an 82% increase in lead volume — 134% above their stated goal. Today, many insurers pair this kind of automation with cost and compliance guardrails to avoid over-optimization in regulated environments. Lastly, AI tools can help you seek out the best ad creatives. You can test various approaches, then optimize and scale based on how your creatives are performing. As a bonus, you’ll gather data that can help you create more effective ad creatives moving forward. One example of AI-powered testing is Progressive Insurance, which used generative AI to test ad variants, producing 96 audio variants in only two weeks. As a result, the company increased quote initiations by 31%. In the end, though, 96 variants proved to be excessive, making it tough for the AI to model data accurately. ### Chatbots and Virtual Assistants As your ads bring in leads, you’ll need representatives to respond and answer any questions people may have. Paying someone to do this job can cut into your budget, but technology has created an easier way. In 2026, chatbots and virtual assistants are commonly used to accelerate early-stage engagement, rather than replace human agents. These tools also come in handy for scheduling agent consultations and walking users through quote comparisons. While they’re no replacement for an experienced agent, technology can handle lower-level customer interactions, freeing up agents to focus on setting up coverage and interacting with existing policyholders. GEICO has hopped headfirst into the technology, building a chatbot into its mobile app. Policyholders can request information on coverage, view billing information, access documents like proof of insurance cards, and more. ### Predictive Analytics AI doesn’t just help with automation, it can also predict which users are most likely to convert and which offer will work best for a specific customer. Modern predictive tools increasingly provide probability-based insights rather than absolute predictions. In addition to knowing the “who” and “what” of ad placement, AI can also be used for timing, since the right tools will serve ads to a user at the optimal time to boost the chances of converting. As marketers shift from targeting by identity to targeting based on intent, it’s essential for insurance marketers to have the right tools in place. These tools will help you reach the right prospects without violating user privacy. Predictive analytics can also be used to assess your marketing campaigns. One example of this is Lemonade, which used the Bayesian marketing mix modeling to gather data on all of its marketing activities. This holistic approach allows the insurer to identify which marketing channels are most effective to help predict future performance. While AI-based tools have made marketing easier, they do come with some challenges. Insurers face both implementation cost and a learning curve in making the switch to these tools, as well as the extra steps involved in ensuring compliance with a regulated vertical like insurance. As they become more popular, marketers also risk over-relying on them. It’s essential to pair these solutions with human expertise to get the best results. It’s important to continuously monitor the performance of any AI-driven strategy. These ongoing audits will ensure the technology aligns with your goals and your customers’ expectations. ## Trend 3: The Growing Importance of Digital Channels and Omnichannel Strategies in Insurance Consistency is key when insurers are communicating with customers. The predictability serves as a comfort, whether they’re getting an insurance quote or comparing one provider against another. Paid reach on search and social platforms is more expensive than ever, and prices are only continuing to rise. As a result, insurers in 2026 are expanding into diversified digital discovery and performance display environments to improve efficiency. Some key stats impacting omnichannel strategies include: - 78% of insurers reported plans to increase budgets for tech spending in 2025. Priorities were, in order: AI, big data and analytics, and cloud and digital infrastructure. - 47% of all insurance policy purchases now occur through digital channels, significantly outpacing traditional agents and call centers. - In specific product segments like personal auto and homeowners insurance, a majority of policies are initiated online, reflecting growing digital discovery and purchase behavior. It’s important to note that even though insurance shoppers often start researching on mobile, the journey doesn’t end there. Top insurers are turning to omnichannel implementation, starting with digital ads and funneling users to personalized landing pages, then nurturing them through email and having an agent follow up to close the deal. Marketing doesn’t stop once you’ve won over a customer, though. By using tools like SMS reminders for policy renewals, you can ensure a policyholder sticks around from one year to the next. While many things about insurance marketing are standard to marketing in general, there are some factors that make it stand out. Here are some key touchpoints in the customer journey when shopping for a new policy: - Awareness: You want to make a good impression on new customers, and for that, few things work as well as display ads. The right ad creative can grab a prospect’s attention. - Consideration: Once a customer knows about your brand, it’s time to progress to the research phase. Comparison tools and dynamic ad creatives can help you educate consumers and convince them to make a purchase. - Decision: Sign-ups should be as easy as possible to keep interested customers moving forward. Seamless mobile quote forms and AI chat support can equip you with what you need to convert a prospect into a customer. - Retention: Keeping existing customers on board is important, too. Personalized renewal messages and loyalty messaging can maintain engagement with your current policyholders. Even the best marketing campaigns can fall apart. Disconnects between online and offline experiences aren’t uncommon, but there are some things brands can do to keep things consistent both online and off. Make sure you use consistent messaging across channels, and check that your online quotes include a save feature. This will allow a prospect to continue a quote on a different device. Don’t forget mobile: Consumers are increasingly browsing on mobile versus desktop, making it important that brands have mobile-friendly websites. Make sure your forms are both short and tap-friendly and that your websites have fast load times. If you encourage customers to call to get or finish an insurance quote, click-to-call is essential in a mobile-centric world. ## Trend 4: Content Marketing and Building Trust in the Insurance Sector Trust is essential in insurance. Customers need that extra layer of comfort before signing a contract, so for that reason, content marketing has become a staple in the industry. The right content helps educate customers and establish your brand as an authority. Here are some ways to build authority and foster trust for your insurance brand in 2026: - How-to guides. - Interactive quizzes. - Customer testimonials. - Premium and deductible calculators. - Thought leadership articles on industry changes. - Videos and blogs told from the perspective of licensed agents. A few trends are shaping the type of content that’s landing in 2026. Interactive carousels and dynamic question-and-answer content can capture short attention spans and make customers feel as though they’re part of your brand. Also of note: - About 75% of marketers reported diminishing returns on social media ad spend in Q1 2025, and more than 80% expanded beyond social media to additional channels. - A 2026 industry outlook revealed that customers increasingly expect clarity and transparency from insurance companies in online channels. Source: Taboola For insurance marketers, investing in diverse formats should be a priority. Interactive tools like premium calculators and educational articles can be ways to attract and engage customers. Search engine optimization (SEO) is another area of marketing that’s always changing. For insurance marketers, remaining competitive means tweaking your strategy for the techniques working best in 2026. That includes: - Long-tail queries: Phrases like “best health insurance for recent college graduates” will gain more traction than “best health insurance.” - Hyper-localization: If you can dominate a relevant local market or two, you’ll be able to capture more attention than going for a nationwide market. Keyphrases like “homeowners insurance San Diego” will help you gain visibility in those local markets. - Voice search optimization: Today’s customers search using voice. By structuring your pages for Google’s featured snippets and voice search, you can gain a competitive edge. - AI search: With growing numbers of people either searching directly through platforms like ChatGPT, or relying on Google’s AI snippets, tailoring content to be picked up by these tools is also critical for success. Another trend dominating content marketing in 2026 is video. Insurance can be an especially complex issue, and videos can help demystify various topics. DirectAsia used short-form video as part of a larger strategy that generated more than 2,000 leads in just a six-month period. Best of all, the company reduced its cost per lead by more than half. ## Trend 5: Data Privacy and Regulatory Changes Impacting Insurance Digital Advertising Privacy has become a hot topic in recent years, and in 2026, consumers are more aware than ever of their rights. A growing number of U.S. states now have comprehensive consumer privacy laws, creating a complex compliance landscape for insurance marketers. Insurers handle sensitive information, so it’s important to be aware of the key regulations and how they’re changing in 2026. - General Data Protection Regulation (GDPR): This data protection law applies to residents in the European Union (EU). It’s stricter than U.S. law, so it’s important that any insurers with EU operations stay on top of it. - California Consumer Privacy Act of 2018 (CCPA): California residents are protected under this law, which gives them control over how their data is collected and used. In 2026, increased fines and penalties continue to heighten pressure on businesses to remain compliant. - Gramm–Leach–Bliley Act (GLBA): This federal law ensures companies offering financial products — including insurance — disclose information-sharing practices and offer opt-out options for sharing personal information. Recent updates to the GLBA Safeguards Rule impose stricter cybersecurity and data protection requirements on insurers and their vendors. - New state privacy laws: 19 states now have privacy laws that impact marketers across all industries. Many states also impose comprehensive consumer privacy obligations, not just targeted or sector-specific rules. - American Privacy Rights Act: This proposed legislation will impose privacy regulations at the federal level. While it’s not in place yet, the APRA remains under active consideration and could significantly affect future advertising, so it’s important to keep an eye on it as you’re planning future data collection and advertising practices. As privacy laws continue to tighten, insurance marketers have been challenged to find new ways to target consumers. Here are some statistics on how privacy is affecting insurance marketers in 2026: - 19 U.S. states have comprehensive data privacy laws, with new laws taking effect in Kentucky, Indiana, and Rhode Island on January 1st. - As of early 2025, privacy and data protection laws are in effect in 144 countries. - Privacy laws protect an estimated 6.64 billion people, which is roughly 82% of the global population. Privacy regulations limit use of third-party cookies and cross-site tracking without consent. Advertisers also now struggle with retargeting, since these regulations limit the use of personal identifiers. Complicating matters is third-party cookie deprecation, which is now an operating condition rather than an upcoming disruption. This has driven marketers toward delivering ads based on a user’s current activity (contextual tracking) and actions on a site (behavioral intent signals). For insurance marketers, privacy protections are even more important. Consumers prioritize trust when looking for an insurance policy, and behavior that might be seen as shady or invasive can erode that trust. To thrive in a privacy-first world, insurers should: - Use clear, concise consent language. - Make it easy for consumers to choose which data to share. - Regularly audit data partners for compliance. - Lean into tools that predict instead of track. - Offer a gift in exchange for contact information (i.e., calculators, informational content). ## Key Takeaways Personalized insurance marketing is no longer optional, but in 2026, effectiveness depends on delivering relevance while respecting privacy. AI-powered tools make it easier for brands to use contextual tracking and behavioral intent signals to target prospects. With many consumers now starting their insurance search on mobile, a seamless omnichannel approach is crucial, but it’s important to build in privacy protections in everything you do. Insurance is one industry where trust-building is essential, and brands can do that through the content they create and present to consumers. ## Frequently Asked Questions (FAQs) ### What are the most effective digital advertising channels for insurance in 2026? In 2026, performance display advertising dominates, and brands are taking creatives to the next level. Carousel ads and vertical videos engage today’s users more than traditional native placements. While marketers still rely heavily on search and social, rising costs have driven marketers to look for more affordable ways to stand out in the crowded insurance market. Short-form video and mobile-optimized clicks are especially popular with insurance marketers in 2026, providing a way to capture user attention and boost their cost per acquisition (CPA). ### How can digital advertisers measure the ROI of their insurance marketing campaigns? Tracking return on investment (ROI) is an important part of optimizing your budget, but marketers need to look at a variety of metrics to get the full picture. CPA and cost per lead are important, along with overall customer acquisition cost, quote-to-policy conversion rates, and the lifetime value of each acquired customer. Top-performing insurers rely heavily on all-in-one platforms that can connect the dots between ad spend, user engagement, and policy enrollment, making it easier to keep an eye on the entire customer journey. ### What are some common mistakes to avoid in insurance digital advertising? The biggest mistakes in insurance digital advertising can be tied back to customer experience. Overcomplicated enrollment forms that aren’t mobile-friendly will lose many prospects almost as soon as you attract them. Another mistake brands make is failing to hyper-personalize advertising efforts, which can make ads feel generic. Since building trust is important in advertising, failing to comply with data regulations is a big no-no. Lastly, relying on third-party tracking data in an increasingly privacy-restricted environment can lead to outdated information that impacts your ROI. ### What are the key performance indicators (KPIs) for successful insurance digital marketing? In many ways, insurance marketing doesn’t differ from marketing in other industries. Marketers still need to focus on click-through rate, time spent on page, cost per lead, and return on ad spend. Insurance marketers should also monitor conversion rates on quote forms, policy signups, and policyholder retention. Platforms that track these metrics while building in AI optimization can help you identify which audiences and creatives bring the best return. ### How is the rise of insurtech companies influencing digital advertising strategies in the insurance industry? The global insurtech market is experiencing massive growth, and that’s good news for insurance marketers. Today’s top insurtech companies offer seamless, mobile-first experiences and highly educational content. To remain competitive, more traditional insurers have found they need to step up their game, investing in performance-focused tools that automate testing and targeting. These tools help level the playing field, allowing even smaller insurance providers to compete. --- ### How to Prepare for Amazon Prime Day 2025: Tips for SMB's URL: https://www.taboola.com/marketing-hub/amazon-prime-day-tips/ Last Modified: 2025-07-03 06:47:40 Amazon’s Prime Day has become a four-day online event during the height of summer. Millions of shoppers visit the site to find deals on items they’ve been considering, along with browsing to see what’s available at a great price. For SMBs, this event brings lots of opportunities to gain new customers and increase sales quickly. Last year, Prime members around the world bought more than ever before during Prime Day, and independent sellers, most of them small and midsized businesses, sold more than 200 million items. Prime Day is also fast-paced and competitive. While it has some similarities to Black Friday, marketers have to keep in mind that it’s a more time-sensitive buying journey, and more intent-driven. The funnel is a lot shorter on Prime Day, with research and buying happening in the same session. Here’s how to prepare and strategize to make these days successful, no matter what your budget and resources. ## Pre-Prime Day Preparation Start off strong with smart preparation ahead of the Prime Day event, including figuring out your campaign type, budgeting, A/B testing, audience targeting, and strategic bidding. Above all, make sure to have all your plans lined up in advance of the event. 58% of consumers wait until Prime Day to buy big-ticket items like electronics and home goods — it’s a captive audience, but for a very short period of time, so you want to make sure you’ve nailed all the details to make it easy for shoppers to find and buy what you’re offering, letting you focus on real-time tweaks as needed during the event itself. Also, make sure you have enough inventory, that your content is ready, and you’re advertising the right products. Here’s what else to consider as you’re putting your plan together. ### Campaign Types With multiple options and a huge potential audience, consider carefully which campaigns you’ll build and run for Prime Day. SMB marketers don’t always have massive resources, so it’s important to be smart about the placements you choose, and craft relevant messaging. In addition, make sure to diversify the channel mix for Prime Day campaigns: Place the right ads on the right channel based on what you know about your audience and prospects, and what your Prime Day goals are. You can create separate campaigns for each targeting type for more data-driven decisions at scale during the event. On top of this, you need to make sure to optimize your product detail pages and align them with your advertising around deals or coupons, especially any type of Prime-exclusive or Prime Day discounts. Product detail pages should have clear, concise messaging, bullet points on product benefits, and high-quality imagery and descriptions. Check ahead of time that the creative for Prime Day follows Amazon’s guidelines for special events. ### Budgeting For an event like Prime Day, it’s essential to be able to dynamically shift budget to top-performing creative to get the most out of it — while four days is longer than previous Prime Days, it’s still a condensed period of time, so wise spending will help you stay under budget. It’s possible within Amazon to create schedule-based rules, so that peak shopping hours receive more spending, and ideally higher conversion rates. ### A/B Test with Early Campaign Launch You can start with what you already know about which types of campaigns, content, and offers work for your audience as you prep for Prime Day. A/B testing ahead of the event can be very helpful, allowing you to start gathering data and understanding audience intent before you go live. Test your creative and be sure to retarget users, as well as calling out the discount, limited-edition, or other special Prime Day information. Amazon recommends starting automatic targeting at least five weeks ahead of Prime Day to give you time to identify any new shopping patterns or keywords that you can use. ### Audience Targeting Getting your audience targeting right will help you stand out to buyers in a sea of shopping opportunities. Consider these tips to get your strategy on track ahead of the event: - Focus on high-intent shopping segments, which might include deal seekers and category-specific enthusiasts. - Don’t exhaust channels where competition and prices are high, such as search, social, and Amazon itself — look for ways to prioritize visibility where competition is lower. - Use any predictive analytics tools and historical sales data to focus on users who are likely to convert based on past purchases. ### Bidding Tips SMB marketers have to optimize bidding wisely during Prime Day. You’ll want to have the most visibility possible within your budget, which can be challenging with so much competition. Your best bet is optimizing bids toward peak hours, then lowering them during slower parts of the day. Keep in mind the various time zones of shoppers as well, depending on what you know about your audiences. Again, Amazon offers features like schedule-based bidding rules and audience bid boosting, which might be useful, depending on your budget and goals. ## During Prime Day As the Prime Day event approaches, stay on top of all your marketing tactics throughout the four days. For SMBs in particular, agility and flexibility are key to success. ### Real-Time Analysis and Adjustments Performance marketers can count on Prime Day being a busy time — ideally with lots of purchases to show for it. Gather as much real-time data as you can with your marketing platform and make changes as needed. Ideally, your preparation will help you avoid too many surprises, but you can’t always predict every outcome, particularly when you’re dealing with lots of high-intent shoppers in a short time period. Focus on monitoring campaign performance, inventory levels, and competitor activity, such as price changes. Then, use insights to adjust targeting, bids, and budgets to get the most performance out of your campaigns (and the most sales possible). You might increase promotion of a product selling better than you expected, for example, or add cart recovery strategies if lots of shoppers are abandoning their carts. Some typical metrics to monitor on Prime Day include: - Inventory levels. - Sales volume. - Revenue. - Average order value. - Conversion rates. - Marketing campaign performance. - Web traffic and clickthroughs. - Customer service inquiries. ## Post-Prime Day Analysis After Prime Day is over, it’s time to see what went well and where to improve. Keep an eye on these areas: ### Data and Reporting Put together a comprehensive report after Prime Day is over. What performed well? What didn’t? What was much better or much worse than projections had anticipated? See what your top-selling products were, which promotions succeeded, and which customer segments drove the most revenue. Keep track of these metrics, plus any others that are relevant for your industry: - Overall sales performance compared to goals and previous events. - Product-level performance analysis. - Customer segment performance. - Marketing campaign effectiveness. - Inventory management efficiency. - Profitability analysis, including margins on discounted items. After doing a full report and debrief, performance marketers can make strategic changes accordingly. You might update customer segments based on Prime Day behavior, consider tech updates if there were roadblocks, or refine forecasting models for inventory, for example. The data from Prime Day, captured and analyzed well, can inform your performance marketing strategies going forward. ### Customer Sentiment What did customers think about your Prime Day promotions? Sales data and the number of first-time buyers can give you a baseline for how well the event went for your business. Ideally you also captured some useful first-party data on customers and gathered insights into popular products. Depending on your industry, you might consider engaging further with customers in these ways: - Follow-up social media posts. - Targeted emails, such as for product reviews or post-purchase surveys. - Analysis of customer service interactions. - Product reviews and ratings on Google or Yelp. - Focus groups or interviews if appropriate. ## Key Takeaways Amazon Prime Day can be a big revenue driver for businesses of all sizes. Making it a success for SMBs involves strategic audience targeting, budgeting, and bidding, along with real-time tracking and adjustments during the four-day event. ## Frequently asked questions (FAQs) ### How can I make my deals stand out? Standing out as an SMB on Amazon Prime Day takes some strategizing. Make your deals stand out as an SMB marketer by incorporating all the first-party data knowledge you can. The insights gleaned from first-party data can enable effective ad placement, so you reach the right audience through the right ad formats. ### Am I thinking full funnel, or should I be focused on my performance metrics? The traditional idea of full-funnel marketing still plays a role in buying journeys, but users are increasingly taking nonlinear routes to get to the conversion or purchase stage. The journey may look like a funnel, but it also may be a web of varying interactions across channels. On Amazon Prime Day, you may be working all stages of the funnel, from awareness to purchase. The time frame is so condensed, though, that performance marketing metrics are a better option to get results. Performance data should inform creative, targeting, and media spend decisions before and during the event. ### Is my inventory ready for Prime Day? A data-informed inventory strategy is the best way to be ready for Prime Day. There’s always a bit of uncertainty — will there be too little inventory for a popular product, or too much for another product that underperforms? If you’ve used your historical data and customer knowledge, along with Amazon’s inventory management tools, you’re as ready as you can be. Monitor inventory levels closely during the Prime Day event to make adjustments as needed to avoid customer dissatisfaction with out-of-stock issues. --- ### Affiliate Marketing Trends to Know URL: https://www.taboola.com/marketing-hub/affiliate-marketing-trends/ Last Modified: 2026-01-13 10:23:51 The affiliate marketing landscape is more competitive than ever in 2025, thanks to organic reach declining (in part to Google’s AI Overviews), as well as cookie-based tracking becoming less reliable. Still, there is plenty of opportunity to succeed if you can adapt to the changing environment. With that in mind, here are five affiliate marketing trends to be aware of right now. ## Trend 1: The Continued Emphasis on Transparency and Trust in Affiliate Marketing Trust and transparency have always been foundational to effective marketing. Consumers are far more likely to buy from brands they believe are honest and credible, and it doesn’t appear that trend will change any time soon. In fact, the need for trust is on the rise: - According to the 2023 Edelman Trust Barometer, 71% of consumers believe that it’s more important to trust brands today than it was in the past. - 79% of Gen Zers believe it’s more important than ever to trust the brands they buy from and use, per the same study, and 59% of consumers are more likely to purchase products from a brand they trust. - According to Trustpilot, 66% of consumers are often or very often influenced by customer reviews when making a purchase. Unfortunately, consumers are becoming more skeptical of overly positive product reviews, misleading content, and fake testimonials. Consider these statistics: - A 2023 Bazaarvoice study revealed that 75% of consumers worry about fake reviews. - As far back as 2020, 49% of global consumers already believed that many brands were guilty of manipulating customer reviews to boost their reputation. - 63% of consumers believe it's the brand’s responsibility to ensure that information presented in online reviews is accurate and complete The erosion of trust goes beyond shady marketing practices, with consumers being more exposed to fraud than ever before. In the United States, the Federal Trade Commission (FTC) reported that consumers lost more than $12.5 billion to fraud in 2024, a 25% increase over the previous year. So, how can advertisers and affiliates build long-term trust with their audiences in 2025 and beyond? For starters, only promote products you genuinely believe in, and look to add value rather than making a quick sale. Affiliates should share personal experiences and aim to provide unbiased comparisons. All parties must ensure they are following FTC guidelines by clearly disclosing affiliate links and partnerships. Many platforms offer tools to make this easier: For example, Instagram and YouTube offer built-in “Paid Partnership” tags to clearly label sponsored content (more on this in trend 3, below). The bottom line is that transparency and trust are no longer a nice-to-have in affiliate marketing, they are non-negotiables. And remember, authenticity matters more than perfection. ## Trend 2: The Growing Importance of Content Quality and Niche Authority for Affiliates As audiences place an increased importance on trust, it’s more crucial than ever for affiliate marketers to prioritize high-quality, valuable content. While this can require a significant amount of time and effort, affiliates who provide genuinely helpful, well-researched content will stand out, especially as distrust around the quality (and amount!) of AI-created content continues. Top marketers are aware of this. In 2024, WordPress VIP conducted an extensive survey of marketing executives and found the following insights: - 54% of organizations believe that the pressure to drive revenue with content has increased over the last 12 months. - Organizations saw creating high-quality content (33%) as their biggest challenge in 2024 (only 17% cited data privacy concerns). Source: WordPress VIP WordPress VIP found that the results were similar for both B2C and B2B businesses, indicating that content is key, regardless of whether you’re marketing to businesses or consumers. Building trust and authority through content goes beyond creating in-depth tutorials, reviews, or expert roundups that highlight your knowledge and experience, though. While those are highly effective content formats, you also have to engage with your community through social media, email newsletters, or online forums. It’s a long-term strategy, but over time, affiliates can establish themselves as trusted experts in their niche. Finally, don’t forget about the importance of SEO optimization in promoting your content. While AI tools are changing the way people search online, strong SEO practices continue to build trust and authority, not only with search engines but also with users. ## Trend 3: The Evolution of Influencer Marketing within Affiliate Programs Influencer marketing is no longer about the number of likes or overall reach: In 2024, brands partnered with more micro-influencers to increase engagement and conversions, since these smaller creators have established high levels of trust and authenticity with their audiences. - According to Hubspot’s 2025 State of Marketing & Trends Report, 44% of B2C marketers found the most success with micro-influencers (audiences of 10,000 to 99,999) in 2024. - Only 25% of marketers had the most success with macro-influencers (audiences of 100,000 - 999,999), per the same report, and only 7% had the most success with mega influencers (audiences of $1MM+). (Hubspot, 2025) Not only is there a greater focus on smaller influencers, but how influencers are compensated is also changing. Traditional sponsorships are still a thing, but more brands are providing influencers with unique affiliate links. This way, brands can track conversions directly from the influencer’s content, ensuring that payouts are tied more closely to performance. TikTok, YouTube Shorts, Instagram Reels, and even Threads have emerged as top platforms for influencer affiliate marketing because they provide high engagement and work well with product-focused content. Speaking of AI, many marketers believe that influencers will eventually be replaced by AI-generated influencers or social media avatars. In fact, Hubspot reported that 86% felt this shift would begin to happen by the end of 2025. But, that remains to be seen — as mentioned above, quality and authenticity are vital to success in affiliate marketing, and neither are currently terms strongly associated with AI-generated content. And, again, it’s clear that smaller influencers who have that built-in trust with their audience are finding increased success in affiliate marketing. ## Trend 4: The Strategic Use of Data and Analytics for Affiliate Optimization Data remains one of the most potent tools for affiliate marketers. The information you gain from measuring click-through rates (CTRs), conversions, and more can help you uncover what’s working and identify areas where you can improve. Marketers recognize the importance of data in 2025: When asked by Hubspot about the biggest marketing industry changes in the past year, data usage took two of the top five spots on the list (see chart below): Source: Hubspot So, what key metrics should affiliate marketers track? While it depends on your campaign goals, here are some key metrics to consider measuring: - Click-through rate (CTR): Measures the percentage of users who click on an affiliate link compared to the number who saw it. - Conversion rate: Measures the percentage of users who take the desired action, like an email sign-up, purchase, or download. - Earnings per click (EPC): Tracks the average earnings for every click on an affiliate link. - Traffic source performance: Breaks down where affiliate traffic is coming from and compares the conversion rates of each traffic source, such as social media, email, and search traffic. Explore 8 ways Taboola Can Maximize Your Affiliate Marketing Campaigns. Beyond standard performance metrics, many brands, especially retail brands, are placing an increased focus on incrementality. In other words, they want to know if the affiliate truly influenced a sale, or would the customer have purchased anyway? Brands are facing tighter profit margins, so naturally they want to know which affiliates are really driving incremental value, as they want to avoid paying commissions on sales that would have happened even without the affiliate’s involvement. Marketers can use key metrics, such as customer lifetime value (LTV), new vs. returning customers, and assisted conversions to measure incrementality. These and other metrics also help marketers identify high-performing affiliates to whom they can offer higher commission rates or exclusive deals. There are several tools available to track affiliate link performance and conversions, but the most suitable ones will depend on your specific use case. For example, Amazon Associates offers its own built-in dashboard, but it’s only compatible with Amazon links. For affiliates using Realize to track their performance, a wide range of metrics are available, including conversion rate, CTR, CPA, and more. If you’re promoting products across multiple networks (ShareASale, Impact), a tool like Affluent.io or WeCanTrack can consolidate all of your data in one place. If you’re managing paid traffic campaigns or require more in-depth attribution data, apps like Voluum or ClickMeter provide robust tracking, A/B testing, and conversion reporting. Speaking of A/B testing, marketers and affiliates should always A/B test their campaigns. Try different variations of everything from headlines to copy, CTAs, link placements, and landing pages to see which ones perform better. ## Trend 5: The Rise of New Platforms and Technologies in Affiliate Marketing While many tried-and-true affiliate strategies remain effective in 2025, affiliate marketing is evolving rapidly, thanks to the emergence of new social platforms and the growing popularity of short-form video formats, such as TikTok, Threads, Instagram, and YouTube Shorts. AI tools also make it easier than ever to build a website, although driving traffic to sites is still a challenge, one best met through a combination of digital advertising, social media, and SEO. According to Hubspot, marketers today are reporting the highest ROI from short-form video: Source: Hubspot, 2025 State of Marketing Report AI tools are also transforming how affiliates create content and optimize their campaigns. AI content generation is already widespread, thanks to platforms like Jasper, Claude, ChatGPT, and many more, and according to a 2024 study, nearly 80% of affiliate marketers reported embracing AI-driven content creation. When it comes to creating ad campaigns, ad networks like Realize, Google Ads, and Facebook Ads now offer AI-powered ad targeting and bidding, making it easier for advertisers to reach their target audience. Web3 applications, built on blockchain technology, are being used to automate payouts, improve tracking accuracy, and create more trust between advertisers and affiliates. While the technology is still early and somewhat experimental, use cases like tokenized incentives and smart contract-based affiliate programs show promise. Privacy-related changes have created challenges in affiliate marketing, too, including the loss of third-party cookies. This has made traditional affiliate tracking much less reliable. To counter this, affiliates should try to use first-party tracking whenever possible and leverage affiliate platforms that support cookieless tracking, like Impact or Partnerize. Other options include focusing on email marketing, which is not affected by browser-based tracking limitations. Cross-device tracking has become increasingly valuable in 2025 as users are constantly moving between phones, tablets, and desktop computers. To keep up, affiliate platforms are using first-party data, user logins, and device fingerprinting to track a user’s actions across devices. Even major affiliate networks, such as Impact and Partnerize, are supporting cross-device attribution to help affiliates get proper credit and improve commission accuracy. ## Key Takeaways To succeed as an affiliate marketer in 2025 and beyond, you need to be able to adapt to a rapidly changing environment that’s being impacted by AI-powered search (and content creation), ever-changing privacy laws, and evolving consumer expectations. The most successful affiliates will understand how to create high-quality content that builds trust and transparency with their audiences. They will also realize that performance marketing that leverages data is no longer optional. If you can embrace these trends and leverage the right tools and strategies, you can stay ahead of the curve and continue to grow in a highly competitive industry. ## Frequently Asked Questions (FAQs) ### What are the most profitable affiliate marketing niches in 2025? In 2024, Authority Hacker surveyed over 2,000 professional affiliate marketers and found that the most profitable affiliate niches were Education & E-Learning, Travel, Beauty & Skincare, Finance, and Technology. Health & Fitness, E-Commerce, and Home & Garden were also very profitable. ### How can beginners get started with affiliate marketing? While the path to becoming an affiliate marketer can vary, here are five key steps to follow: - Choose a profitable niche you understand. - Build a platform to sell on, e.g., YouTube channel, TikTok account, website, email newsletter, etc. - Sign up with one or more affiliate networks, like Amazon Associates, ShareASale, Impact, Skimlinks, etc. - Create valuable content, such as detailed product reviews or comparisons, using written copy and/or video. - Measure your performance. Use analytics software and AI tools to monitor clicks, conversions, and earnings. Remember to A/B test and double down on what works. ### What are some common mistakes to avoid in affiliate marketing? Inexperienced affiliate marketers often struggle with the same mistakes. This can include creating low-value posts that are stuffed with affiliate links and, as a result, potentially penalised by search engines; failing to include proper legal disclosures; and not measuring performance by tracking click-through and conversion rates. ### How can advertisers find and recruit high-quality affiliates? To ensure you work with high-quality affiliates, leverage analytics to identify top performers and target niche creators who have established a high level of trust and transparency with their audience. Always remember to keep the lines of communication open between you and your affiliates. ### What are the key legal and regulatory considerations for affiliate marketing? One of the challenges affiliate marketers face in 2025 is complying with the ever-changing regulatory landscape. You must comply with global privacy laws, such as the GDPR, inform users when cookies are used, avoid any deceptive sales practices, and satisfy FTC disclosure requirements (U.S.). Additionally, many social media platforms, search engines, and affiliate networks have their own rules regarding affiliate links. If you don’t comply, your account may be suspended. --- ### User Acquisition (UA): How to Get Prospects to Convert URL: https://www.taboola.com/marketing-hub/user-acquisition/ Last Modified: 2025-06-30 09:18:35 User acquisition, or UA, is getting new users to view a website, download an app, or sign up for a new service. For some industries, this looks more like customer acquisition, but for digital products and services — like websites and apps — it’s called user acquisition. ## Understanding User Acquisition Companies measure UA in a few different ways, depending on the industry and other factors. For instance, an app may measure UA through how many downloads it gets, while a website might track UA by how many clicks brought in new visitors. Both approaches have a similar goal, i.e., to gain new users at a lower cost than the user’s lifetime value (LTV). ### What Are the Main Goals of User Acquisition Efforts? Regardless of the UA measurement, many companies have similar user acquisition goals, such as: - Monetizing your site, app, or service. - Increasing attention and recognition for your company or brand. - Figuring out which marketing and promotional methods work for converting new customers. - Determining how users navigate the service, site, or app from initial visit or download, through their entire experience. ### How Does User Acquisition Relate to the Overall Growth Strategy of a Company? UA is an ever-evolving measurement. It’s a tool that takes you through the entire lifecycle of a user, from their first impression of your product or service to ongoing engagement. Without a user acquisition model, it’s difficult to figure out what works (and doesn’t) for building up a user base. Even as goals, ideas, and plans change over time, building and tweaking user acquisition measurements is a normal part of the process. ## Key Channels and Strategies for User Acquisition ### What Are Organic User Acquisition Channels? Organic UA attempts to reach users through non-paid, organic means, like search engine optimization (SEO), social media posts, and content marketing. These options provide free (or less costly) resources you can capitalize on to attract new users. For instance, you can write forward-thinking articles on your site, optimize the page for SEO, and share that article on your site without paying much in the way of additional costs to spread that content. ### What Are Paid User Acquisition Channels? While organic user acquisition gains users through largely free avenues, paid UA is when a company pays to get new users to visit, download, or otherwise find a new site, app, or service. These methods include paid search, social media promotions, or display ads — all content that costs money to execute. Affiliate marketing is one way to build users through paid user acquisition. When an affiliate shares a business’ ad, product, or service on their website or social channels, it reaches their audience. When one of them clicks and converts, the affiliate gets paid a commission. Influencer marketing operates similarly: Companies pay influencers to share a product, service, site, or app with their fanbase and followers. Sometimes, influencers promote companies with a discount code or special link. Every time one of those followers uses that code or clicks a link, it counts as a conversion and influencers get paid. Some companies offer referral programs to boost user acquisition. These options vary based on the company, offerings, and benefits: For instance, a bank might offer a cash incentive for every referral that signs up for a new account. Other companies may reward existing customers with free products, store credits, and upgrades without an additional charge. ## The User Acquisition Funnel ### What Are the Different Stages of the User Acquisition Funnel? User acquisition funnels follow a similar strategy to other types of marketing funnels. At the top of the funnel, there’s awareness. This is the largest part of the funnel, where sites and apps are trying to reach the most amount of users. In the middle of the funnel, there’s consideration — a smaller group of users who are starting to learn more about the product. At the end of the funnel is the smallest stage: decision. This stage is the group who will hopefully take the final step to conversion. ### What Key Metrics Are Tracked at Each Stage of the Funnel? #### Stage 1: Awareness Impressions and clicks measure the awareness stage. Look at specific metrics, like unique visitors, where users are coming from, and the bounce rate. Measure all UA outlets, like native advertising, video ads, and search engine optimization (SEO) content. #### Stage 2: Consideration Think about how users and readers are considering your site or product. For ads, evaluate the cost per click (CPC) and cost per lead (CPL). Also look at the click-through rate (CTR) and session times to see how many users are going from initial awareness to consideration. #### Stage 3: Decision For this stage, you’re tracking retention rates, cost per acquisition (CPA), and how many people are starting to trust your app or site. These measurements help you bring users back as repeat users and readers. ### How Can You Optimize Each Stage to Improve UA Efficiency? Start by setting clear goals for your business or site. These goals are different for everyone, depending on the type of business you have and what you’re looking for out of users. During the awareness stage, you’ll want to reach your wider target audience. You can do this through hyper-targeting and segmentation, where you send out specific ads and messaging to specific groups of people. Try A/B testing on your site, app, and emails, switching up call to action (CTA) messages and visuals to see what sticks. As you’re evaluating who is coming from where, continue to target and encourage those users in the consideration stage. Take advantage of remarketing tools to get your messaging back in front of people who’ve seen your product, but not yet converted. At the decision stage, make conversion as easy as possible. Signing up, downloading, check-out, or any other onboarding should be seamless. Keep users engaged all the way through by offering incentives, sign-up bonuses, free trial periods, or other methods to encourage immediate action. ## Measuring and Analyzing User Acquisition Performance ### What are the key performance indicators (KPIs) for user acquisition? KPIs vary depending on your app or site. Generally, you should think about conversion rate, i.e., the number of users that are actually converting on your site. Another vital metric is customer acquisition cost (CAC), which measures how much you’re spending on average to acquire new customers. To calculate this, add up all your sales and marketing spend, and divide it by the number of new customers you’ve acquired during a designated period. ### What tools and platforms are used to track and analyze UA data? Depending on what you’re measuring, you might need a few different tools in your arsenal to track and monitor user acquisition data, such as Google Analytics, Meta Ads Manager, TikTok Ads, or Pinterest Advertising. Use them to discover where users found you, then track their customer journey. You can group users together through these discoveries, using cohort analysis to monitor user behavior throughout their journeys. ## Optimizing User Acquisition Efforts ### How can A/B testing be used to improve UA campaigns? A/B testing can improve CPC and CTR results as you learn how different users interact with your site, app, and advertising efforts, so you should constantly be testing everything from email newsletter subject lines to different images or visuals and various CTAs. Take advantage of advertising platforms that offer AI-powered A/B testing, which will let you test and optimize in real time. As an example, A/B testing is a great way to optimize landing pages to boost conversion rates, lower your bounce rate, and analyze the users you’re intending to target. You should test different layouts, styles, product offers, promotions, and CTAs. For instance, does the landing page with user testimonials perform better than the version without it? A/B testing will give you a concrete answer. ### How can you leverage data and analytics to refine your UA strategies? Data should be at the forefront of your strategy, guiding your decisions and letting you figure out for certain what works and what doesn’t. It’s important to remember that data changes regularly, though — there’s no set-it-and-forget-it outlook. Use the data you have to see how you’re reaching, acquiring, and retaining users, then figure out how to continually build off of that momentum. ### What is the importance of understanding user intent in UA? User intent is everything — without understanding a user’s intentions when clicking through to your landing page, you can’t effectively capture their interest and convert them. As a result, you’ll see high bounce rates and low conversion rates, leading to an overall unsuccessful campaign. ## Key Takeaways User acquisition is the process of targeting, converting, and retaining users. It’s best achieved through a mix of organic and paid efforts, and as with any other kind of customer acquisition, it’s vital to both understanding user intent, and your own data. ## Frequently Asked Questions (FAQs) ### What is the difference between user acquisition and customer acquisition? User acquisition is usually for apps and websites, while customer acquisition is for people buying a product. They are technically interchangeable, depending on your business model and what you offer in a user experience. ### How do you choose the right UA channels for your business? First, learn your target audience, goals, and what you can offer users. Then, look at what your competitors use and what works best for them. Think about your budget and identify two or three different channels where you can make initial investments. But, don't forget organic user acquisition options, as these don't require much investment to start. ### How important is mobile user acquisition? While it depends on your company's overall mission, strategy, and offerings, mobile UA is vital for most companies. The vast majority of users are browsing for their needs on their phones. If you want to maximize UA, you should develop a mobile strategy. ### What are some emerging trends in user acquisition? Hyper-personalization is at the forefront of UA. Artificial intelligence, or AI, is a leading tool in the hyper-personalization of UA, whether delivering targeted messages during the impression or interest stages, or triggering an email if a potential user doesn’t make a purchase. --- ### Google Ads: How Do They Work? URL: https://www.taboola.com/marketing-hub/google-ads/ Last Modified: 2025-08-21 10:58:30 Whether you realize it or not, you most likely see Google Ads every single day, either on your desktop or on your mobile, even when you don’t use the Google search engine itself. Google Ads — which until 2018 was called Google Adwords — is one of the most popular online advertising platforms, enabling businesses to showcase ads across Google’s ecosystem. You’ve seen Google Ads on search engine results pages (SERPs), of course, but they’re not limited to just the search engine: They appear on YouTube (owned by Google), Google Display Network partner websites, and apps. Businesses can create targeted campaigns to reach potential customers based on keywords, demographics, interests, or browsing behavior. Just as Google’s search engine constantly evolves in ways it doesn’t publicize, so does Google Ads. And, no, you can’t pick and choose which merchant’s sites you want to advertise on — that’s up to Google. With constant changes that even the best SEO team can’t be fully aware of, it’s vitally important that marketers get familiar with the mechanics of Google Ads to optimize placement and get the most bang for your buck. Here’s the rundown on what you need to know. ## Types of Google Ads ### Search Ads These ads appear on SERP pages. In the earliest days of Google, search ads were text-only ads that appeared in the right-hand sidebar next to search engine results; later, thumbnail photos were added. It was pretty cluttered, if you recall, with a typical SERP page displaying up to 11 ads. The type of ad shown would be determined by the specific keywords users queried: For example, a user searching for “Halloween costumes for cats” would be shown ads for exactly that, as well as adjacent products — pet food, pet toys, and so forth. Google removed sidebar ads in 2016, in favor of a model where ads appeared above the search results (up to four ads) and at the bottom (up to three ads), bringing the total ads shown per page down to seven from the previous 11. Most likely, this decision was based on the fact that, in 2015, mobile searches started to exceed desktop for the first time, and sidebar ads are not optimized for mobile. In those days, Google SERP page ads were clearly labeled “Ad,” but otherwise, the ads sometimes resembled regular search results. So, for example, a search for “Halloween costumes for cats” might feature four ads at the top with titles like, “How to make your cat tolerate clothing without getting bitten.” If you weren’t paying attention, you might think it was an article — only by clicking on it would you see that the link’s purpose was to lead to a page selling taco or princess costumes for your feline. In 2022, however, Google completely overhauled this practice. As mentioned above, this is the year that Google Ads on SERP got rid of the word “Ad:” Instead, Google made it obvious that ad content was of a commercial nature by clearly naming the company running the ad and even incorporating the company’s official logo. This ended up being a boon for advertisers and their brand recognition. A recent, significant change came in 2025. Previously, companies could control where on the SERP page the ad appeared, but as of 2025, Google made the process more complicated. As Google’s official Google Ad support section explains: “When someone searches on Google, we run different auctions for each ad location where we show Search ads — for example top ads are selected by a different Search ad auction from ads that show in other ad locations. Until now, Search ads from a given advertiser were generally restricted to a single ad location on a given page.” ### Display Ads Visual ads (images or rich media) appear across the Google Display Network (GDN), reaching users as they browse millions of sites, including news sites. Here is the Daily Mail site, for example. You perhaps didn’t realize it, but these tiny blue triangles that dot the page indicate that these ads come from the Google Ads network. If you were to click on one of these, it would give you information about the individual advertiser and explain why you are being shown this ad. The image below is what you’d see if you were to click on a blue triangle corresponding to an ad for the clothing subscription company ThreadUp: You can also block the advertiser by clicking the “Block” button at the top of the page, or report questionable content by clicking the “Report” button. Unfortunately, advertisers cannot pick and choose which sites they want to be listed on within the GDN: That’s up to the Google algorithm. All the more reason to make sure your ad copy is continually optimized. ### Shopping Ads These show product images, prices, and store names directly in search results, ideal for e-commerce businesses. This is what might come up if you were to search “Halloween costumes for cats.” The user can choose by size, cost, color, and any number of other parameters. ### Video Ads Video ads typically run on YouTube and can appear as skippable or non-skippable videos, in-stream or in-feed. Users complain about these all the time, but the completion rate tends to be pretty high, so keep that in mind. ### App Promotion Ads App campaigns can help you promote your apps across Google’s largest properties, including Search, Google Play, YouTube, Discover on Google Search, and the Google Display Network. You can simply add a few lines of text, a bid, some assets, and App campaigns automatically optimize your ads for you. ### Performance Max Campaigns This is a newer ad type, powered by Google AI, that combines machine learning and automation to serve across all Google properties, optimizing performance in real time. According to Google: “Performance Max is a goal-based campaign type that allows performance advertisers to access all of their Google Ads inventory from a single campaign. It's designed to complement your keyword-based Search campaigns to help you find more converting customers across all of Google's channels like YouTube, Display, Search, Discover, Gmail, and Maps.” ## How Does Google Ads Work? Google Ads — again, formerly Google AdWords — operates through a real-time auction system. Advertisers select keywords, set bid amounts, create ads, and target specific demographics. Google then evaluates: - Bid amount. - Quality Score (based on ad relevance, landing page experience, and expected click-through rate ). - Ad extensions. The combination determines your Ad Rank, which dictates your ad's position on the results page. ## Why Advertisers Use Google Ads - Intent targeting: Unlike other ad formats, search ads reach users who are actively looking for specific products or information. - Flexible budgeting: From $5 to millions, campaigns can be customized to fit almost any budget. - Measurable results: Track impressions, clicks, conversions, and return on ad spend (ROAS) in real time. - Automation tools: Smart Bidding and responsive search ads help streamline campaign management. - Integration with first-party data: Similar to performance advertising platforms like Realize, advanced optimization is provided using proprietary first-party behavioral signals, improving ad performance and allowing targeting in cookie-constrained environments. ## What is Google Ads Keyword Planner? Google Ads Keyword Planner is a free tool that helps advertisers research and analyze keywords. It provides search volume, competition level, and cost-per-click (CPC) estimates. This tool is foundational for keyword planning and budget forecasting. ### How to Access and Use Keyword Planner 1. Navigate to Google Ads and log in to your account. 2. Click on the Tools & Settings icon (wrench) in the top menu. Under the Planning section, select Keyword Planner. 3. Choose an option: Get search volume and forecasts: View historical metrics and forecasts for your keyword list. Discover new keywords: Find new keyword ideas based on terms related to your products or services. Here, for example, I have entered “Halloween costumes for your cat.” The field beneath allows you to “Enter a site to filter unrelated keywords.” Here, you would enter the advertiser’s URL in order to optimize keywords by allowing Google Ads to ascertain what they do and don’t sell. So, if your advertiser’s URL is https://www.pet-halloween-store.com (fictitious), you’d enter it there. This way, the algorithm knows not to include words like “superhero costumes for elementary school kids” or other keywords irrelevant to the vendor. 4. Review the list of suggested keywords along with their average monthly searches, competition, and CPC estimates. In the screenshot below, you can see that the Google Ad algorithm has suggested such keywords as “cheshire cat” and “cat in the hat.” Use filters to narrow down the list based on location, language, and other parameters. Notice that you can get quite granular here: On the Refine Keywords drop-down menu on the right, you can even include related species, such as lions or leopards. 5. Select desired keywords and add them to your plan to estimate performance and costs. ## What are Negative Keywords? Negative keywords are terms that prevent your ads from being triggered by certain words or phrases, ensuring your ads reach the most relevant audience. By excluding specific search terms, you can reduce irrelevant traffic and improve your campaign's return on investment (ROI). ### How to Add Negative Keywords 1. In your Google Ads account, click on the campaign or ad group you wish to modify. From the left-hand menu, select Keywords. 2. Click on the Negative Keywords tab. 3. Click the + button to add new negative keywords, then choose to add them at the campaign or ad group level. Enter the keywords you want to exclude, then save. ## What are Google Ads Extensions? Google Ads extensions are additional pieces of information that expand your ad, providing users with more reasons to click. They can include extra links, contact information, or promotional details, enhancing the visibility and effectiveness of your ads. ### How to Set Up Ad Extensions 1. In your Google Ads account, click on Ads & Extensions from the left-hand menu, then select the Extensions tab at the top. 2. Click the + button to add a new extension, then choose the type of extension you want to add (e.g., Sitelink, Callout, Structured Snippet). 3. Fill in the required details for the selected extension type and assign the extension to specific campaigns or ad groups as needed. Save your changes. ## What is Google Ads Remarketing? Google Ads Remarketing allows you to re-engage users who have previously interacted with your website or app. By showing targeted ads to past visitors, you can increase brand recall and encourage conversions. ### How to Set Up Remarketing 1. In Google Ads, click on the Tools & Settings icon. Under Shared Library, select Audience Manager, and click the + button to create a new audience list based on website visitors, app users, or customer lists. 2. Navigate to Audience Sources within the Audience Manager. Choose your data source (e.g., Google Ads tag, Google Analytics) and install the remarketing tag on your website or app. 3. To create a remarketing campaign, click on Campaigns in the left-hand menu. Click the + button to create a new campaign, then select a campaign goal (e.g., Sales, Leads) and choose Display or Search as the campaign type. In the Audiences section, add your previously created remarketing audience and configure the rest of your campaign settings, then launch the campaign. ## Google Ads Best Practices ### For Campaigns A well-structured Google Ads campaign begins with a clear alignment between keywords, ad copy, and landing pages. Ensure your keywords reflect user intent and match the content users will land on. Avoid keyword stuffing and opt for tight ad groups for better quality scores and more relevant ad experiences. Landing pages should deliver on the ad promise, be mobile friendly, and load quickly. ### For Performance and Optimization Continuous optimization is crucial. Use A/B testing to compare variations of headlines, calls to action (CTAs), and visuals. Test not just ad creatives but also different keyword match types, audience segments, and bidding strategies. ### For Budgeting and Bidding Set realistic daily and monthly budgets based on historical performance and customer lifetime value. Use automated bidding strategies such as Target CPA or ROAS, which leverage machine learning to adjust bids for optimal results. Make room for flexibility, especially during peak campaign periods. ### For Choosing Ad Formats Choose ad formats based on your goals — text ads for lead generation, responsive ads for A/B testing creative combinations, and shopping ads for e-commerce. The latest AI-based marketing tools can help with this. ### For Local and Geo-targeting Use geo-targeting to tailor ads to specific regions, cities, or even zip codes. This is particularly useful for local businesses or region-specific offers. Layer this with demographic data to improve personalization. ### For Remarketing Use remarketing to re-engage past visitors who didn’t convert. Segment these audiences by behavior (e.g., abandoned cart vs. product page view) and tailor your messaging. Modern, AI-based marketing tools allow for dynamic remarketing using first-party insights. ## Is Google Ads Worth It? Google Ads can be a powerful tool for digital marketers, but it comes with significant challenges. Rising competition, higher CPCs, and the increasing complexity of campaign management make it harder for advertisers to achieve a strong ROI. All of that said, Google Ads are worth it when certain conditions apply: ### If the Company Has Cash to Spend With Google search ads, the average CPM is just under $40 at the time of this writing. Google can get your ads seen by a lot of people, but at a cost that starts high and can get more expensive if you’re in a competitive field. ### If a Brand Already Has a Strong Online Presence A well-known brand is far more likely to attract clicks because users already recognize and trust it. In contrast, a lesser-known small business, unfamiliar to most, struggles to gain attention, even if its ads appear prominently at the top of search results. ### If the Company Has a Well-Defined Target Audience Google Ads is an effective marketing tool for brands with a well-defined target audience. Its extensive customization options allow advertisers to tailor ads to a specific, highly targeted market segment, potentially boosting return on investment. For instance, a luxury cosmetics company can target affluent users searching for “high-end eye cream,” while a local ice cream parlor can target “best ice cream in Bethesda.” On the other hand, Google Ads might be less effective for other types of businesses, for the following reasons: ### Highly Competitive Fields In a crowded market, brands must differentiate themselves with innovative advertising strategies and distinctive performance marketing tactics. For example, a horoscope app could employ bold social media campaigns featuring user-generated content (“Why I gave up on Scorpio men”) to build engagement and stand out from competitors. ### The Relevant Keywords are Too Common Marketing everyday products like cookies or shoes through Google Ads can be challenging, with low returns on investment. For example, a small bakery targeting “chocolate chip cookies” may struggle against major brands like Nestlé, while a local shoe store bidding on “running shoes” could be outbid by giants like Adidas. ## Do Marketers Have Alternatives to Google Ads? Yes, there are several strong alternatives to Google Ads that can help marketers diversify their ad strategies and reduce reliance on walled gardens by advertising on the Open Internet. Platforms leveraging the open web, like Realize, allow advertisers to target users across thousands of premium publisher sites. These platforms offer transparency, control, and access to niche audiences while also supporting performance-driven campaigns with AI optimization. This makes them a viable and often more cost-effective choice. ## Key Takeaways Google Ads is a powerful, widely used digital advertising platform offering search, display, video, shopping, and app promotion ads. It operates on a pay-per-click model, allowing businesses to target users based on intent, demographics, and browsing behavior. ## Frequently Asked Questions (FAQs) ### What are the different keyword match types? Google Ads offers four main ways to match your ads to user searches: broad match, phrase match, exact match, and negative match. These determine how closely a user’s search term needs to align with your chosen keywords. Broad match casts a wide net, showing your ad for related searches, while phrase match narrows it to specific phrases. Exact match is precise, targeting exact terms, and negative match blocks irrelevant searches, helping refine your audience. ### How is Ad Rank determined? Ad Rank is calculated based on your bid amount, ad quality (CTR, relevance, and landing page experience), the Ad Rank thresholds, and the expected impact of extensions and other ad formats. ### Is $5 a day enough for Google Ads? It depends on your industry and goals. While $5 a day can generate data for testing or hyper-local targeting, most competitive industries require a higher daily budget to see meaningful results. ### What’s the difference between Search, Display, and YouTube ads? Google Ads come in three main flavors. Search ads pop up when someone types a query into Google, like “best dive bars near me.” Display ads are eye-catching banners or images shown on websites in Google’s vast Display Network, such as news sites or blogs. YouTube ads, meanwhile, are videos that play before or during content, perfect for storytelling or brand awareness. Each type serves a unique purpose, from driving immediate clicks to building broader recognition. ### What’s a good CTR for Google Ads? A good CTR varies by industry, and depending on whom you ask, anywhere from 2%-7% is considered healthy for search ads. Display ads tend to have lower CTRs. ### How do I set daily and monthly budgets effectively? Start with your monthly budget, divide by 30.4 to get a daily average, and use performance data to adjust bids. (Tip: Platforms like Realize can automate budget allocation to the best-performing ads.) ### How do I know which keywords are driving sales? To pinpoint which keywords spark sales, set up conversion tracking in Google Ads and link it to Google Analytics. This setup tracks user actions — like purchases or sign-ups — showing exactly which keywords deliver results. ### Why is my landing page experience hurting my Quality Score? Your landing page — the webpage users land on after clicking your ad — can drag down your ad’s Quality Score if it’s slow to load, irrelevant to the ad, or hard to navigate. A confusing layout or content that doesn’t match the ad’s promise (say, an ad for sneakers linking to a page about socks) can frustrate users and lower your score. Make sure your page is fast, relevant, and user-friendly to keep your ads competitive. --- ### Paid Search: Getting Found When It Matters URL: https://www.taboola.com/marketing-hub/paid-search/ Last Modified: 2025-08-21 10:59:08 When you type a query into a search engine like Google, take a look at what pops up. Beyond the organic results, you'll almost always see those little "Ad" labels: That’s paid search in action. While they may not be as flashy as an animated banner ad, they’re still a core part of the digital marketing landscape, allowing businesses to appear at the top of search engine results pages (SERPs) right when someone is looking for something specific. For anyone just starting to step into online advertising, understanding paid search isn’t just an option, it's a necessity. Paid search lets you meet your potential customers exactly where they are — and at the exact moment, too. As an advertising copywriter I can tell you firsthand that little adjustments can make all the difference in the end. ## Understanding Paid Search Advertising Paid search is a digital advertising model where advertisers bid on keywords for their clickable ads to appear in search engine results. That’s the high-level overview, but there’s still a lot more to cover, and there’s multiple layers to learn. It's often referred to as Search Engine Marketing (SEM), though SEM really pulls in a broader range of activities, including search engine optimization (SEO). For clarity’s sake, when I talk about paid search here, I’m focusing on those ads you see at the top (and sometimes bottom) of the SERP. They may sometimes appear similar to the organic search results, but there’s a key difference: Organic results are what the search engine determines to be the most relevant, high-quality content for a user's query, achieved through SEO efforts over time. You don't pay directly for clicks on organic results. Paid search, conversely (and hence the name), is a direct payment model: you pay when someone clicks on your ad. So why is paid search such a popular and effective digital advertising channel? It’s partly due to its ability to capture high-intent traffic. Someone typing "buy running shoes online" into Google is likely pretty close to making a purchase, and paid search allows you to be there, at that exact second, with a relevant offer. It's all about being seen by the right people at the right time, and the immediate visibility, precise targeting capabilities, and robust measurability make paid search a game-changing tool for businesses. ### How Does Paid Search Work on Search Engine Results Pages (SERPs)? I’ve covered some confusing and complex processes in digital marketing, but rest assured, paid search is actually pretty easy to understand. When a user types something into a search engine, that search engine's algorithms work fast to determine which ads are going to be most relevant and valuable to display. This all happens in a matter of milliseconds, through a complex auction system. Advertisers start by bidding on keywords, and if their bid and ad quality are high enough, their ad appears. These ads are typically marked with a small "Ad" or "Sponsored" or something similar, and placed above the organic listings, or sometimes below them or even off to the side. ### What Are the Key Components of a Paid Search Ad? A standard paid search text ad is pretty concise, but still packed with information for the user, such as: - Headline: Usually displayed in blue text, the headline is your big chance to capture attention, and often includes keywords the user searched for. You usually get a few headline fields to work with, allowing for more real estate. - Description: This provides a little more detail about your product, service, or offer, and it’s where you can expand on your headline, highlight benefits, and include a call to action (CTA). You’ll generally have a couple of longer description lines to work with here, but you’ll still need to be brief. - Display URL: This shows the user the web address of your landing page. The actual landing page URL might be a lot longer, but the display URL is usually simplified for readability and trustworthiness. - Ad Extensions: Additional pieces of information that can appear with your ad, like phone numbers, site links, structured snippets, or location details. They don't always show, but when they do, they can be a great way to enhance your ad's visibility. Crafting these elements well is crucial. Each word should earn its place, so choose them carefully. I've seen campaigns perform wildly differently just by tweaking a single headline. ## The Auction System and Ad Rank Paid search operates on an auction system, but it's not what you may be picturing, where it’s all about who bids the highest: Instead, it's a more nuanced system designed to provide the best possible experience for both advertisers and search users alike. ### How Does the Paid Search Auction Work? As soon as a search query is made, the auction kicks off, and multiple advertisers might be bidding on the same keywords. In milliseconds, the search engine takes a look at all the eligible ads to determine which ones will show up and in what order. This is your Ad Rank, which is determined by multiple factors: - Your bid: This is the maximum amount you're willing to pay per click. It's definitely a major piece of the puzzle, but not the only one. - Quality score: Another important metric that measures the overall quality and relevance of your keywords, ads, and landing pages. A higher Quality Score means your ad is more suited and useful to the user. - Ad extensions: Including relevant ad extensions (like enhanced sitelinks, ratings, app links, price, and promotions) can all help boost your Ad Rank. - Context of the search: Factors like the user's location, time of day, device, and other signals all play a role here. ### How Does Quality Score Influence Your Ad Rank and Cost per Click (CPC)? Quality Score is Google's (and other search engines’) way of rewarding advertisers for relevance. It's typically scored on a scale of 1 to 10, and a higher Quality Score means your ads are seen as more relevant to what users are searching for. Besides bragging rights, it also leads to several other benefits too, such as: #### Higher Ad Rank When I mentioned before that a higher bid isn’t everything, here’s an example of why. Even with a lower bid, a high Quality Score can allow your ad to rank higher than competitors who have higher bids but lower Quality Scores. #### Lower Cost Per Click (CPC) Another area where it gets interesting for your budget. A higher Quality Score often translates to a lower CPC, meaning you pay less for each click. This is Google's incentive for you to provide a better user experience. Basically, Google wants satisfied users who find what they're looking for quickly, and happy advertisers who get good results. A high Quality Score helps achieve both of those. If your Quality Score is low, you'll essentially be paying a premium for your clicks, and that’s a clear signal that something in your campaign needs optimization. ### How Can You Improve Your Ad Rank? Improving your Ad Rank really comes down to optimizing all the factors that influence it. Here are a few strategies that you should focus on: - Increasing your bids (strategically!): Again, it’s not solely about bids, but sometimes a slight increase can give you the edge over competitors. - Improving your Quality Score: For improving your Ad Rank, increasing your Quality Score might have the most impact. Ensure that your keywords, ad copy, and landing pages are highly relevant to each other, and to the user's search intent. - Using relevant ad extensions: These provide more information, and also take up more SERP real estate, boosting visibility of your ad. ## Keyword Targeting and Matching Options in Paid Search Don’t underestimate the power of keywords — they’re the backbone of paid search. Having accurate terms or phrases that users type into search engines is vitally important, and those keywords are also what you’ll bid on. ### How Do You Select Relevant Keywords for Your Paid Search Campaigns? Selecting relevant keywords is the foundation of everything else, and begins with thorough keyword research. Don’t just guess — utilize the data and findings as you go. Remember that you want to identify terms that your target audience would use when searching for your products or services. Here are a few places to start: - Brainstorm: Think like your customer and get into their mindset. What would they type into the search bar? What would they avoid? - Use keyword research tools: There are plenty of these tools out there. Google Keyword Planner, SEMrush, and Ahrefs can show you search volume, competition, and related terms. - Analyze competitors: Take a look at what your competitors are doing. What keywords are they bidding on? - Consider intent: Are users in the "information-gathering" phase or "ready-to-buy" phase? Tailor your keywords to match their intent and where they are on their customer journey. Remember that relevance is king. Don't chase high-volume keywords if they aren't truly relevant to what you offer — you'll just blow through your budget with less to show for it. ### What Are the Different Keyword Match Types and How Do They Work? Different keyword match types give you control over how closely a user's search query needs to match your keyword for your ad to show, and it’s critical for managing spend and relevance. Choosing the right match type should be an ongoing optimization process. I advise starting with a mix of these, then refining as you gather more and more data: #### Broad Match This is probably the most flexible match type, since it allows your ad to show up for searches that include misspellings, synonyms, related searches, and other relevant variations. For example, if your keyword is "women's shoes," your ad might show up for "ladies sneakers" or "buy footwear." The good part is that broad match offers wide reach, but it can be less precise and lead to irrelevant clicks if it isn’t managed carefully. #### Phrase Match Your ad will show for searches that include your exact keyword phrase, but can also include other words before or after it. For example, if your keyword is "winter coats," your ad might still pop up for "best winter coats" or "waterproof winter coats for sale," but not "coats for winter." This offers a balance of reach and relevance and it’s a good middle ground to make sure you’re still getting in front of the right users. #### Exact match This is the most precise one — with exact match, your ad will only show up for searches that are the exact keyword phrase or really close variants of it (like plurals or misspellings). It offers the most control and highest relevance, but limits reach. ### How Do Negative Keywords Help Refine Your Targeting? Negative keywords are terms you add to your campaigns to prevent your ads from showing for irrelevant searches, and they’re just as important as your regular keywords. For example, if you sell "luxury watches," you might add "free," "cheap," or "repair" as negative keywords. This ensures your ad doesn't show for someone looking for a "free watch repair" or cheaper models. They also significantly reduce wasted ad spend and improve the quality of your traffic by excluding searches that are clearly not looking for what you offer. Don’t sleep on these! ## Ad Formats and Extensions ### What Are the Different Types of Paid Search Ad Formats Available? While text ads are the most common, they’re not the only types. Other formats can include: #### Responsive Search Ads (RSAs) With RSAs, you provide multiple headlines and descriptions, and the system automatically tests different combinations to find the best-performing ones. This is Google's current default and highly recommended (especially by the copywriting side of me). #### Dynamic Search Ads (DSAs) These ads are generated automatically, based on the content of your website. The search engine scans your site and then matches relevant queries to dynamically generated headlines. It’s great for websites with large inventories, and for smaller ones, too. #### Shopping Ads (Product Listing Ads) Shopping ads are a little different: These rely heavily on visuals for e-commerce, showing product images, prices, and merchant names directly in search results. #### Call-only Ads You probably won’t use these, but it’s worth mentioning them just for reference. Call-only ads are for businesses that primarily want phone calls, and allow users to call your business directly from the search result. ### How Do Ad Extensions Enhance Your Paid Search Ads? You may already have a solid ad format written up and ready to go, but ad extensions take it up a notch. These are additional pieces of information that can appear alongside your main text ad, and can give your ad a glow-up in multiple ways: - Increased visibility: The biggest one, literally, is that they make your ad bigger, taking up more space on the SERP, which naturally draws more attention. - More information: They provide users with more relevant details, helping them decide if your ad is what they need, potentially leading to higher purchases. - Improved click-through rate (CTR): More visibility and information often lead to higher CTRs. - Higher Ad Rank: Google factors the expected impact of extensions into its Ad Rank calculation for you. Maybe the best reason to use extensions: They're basically free upgrades to your ad real estate, so always use them where they’re going to be relevant. ### What Are Some Common and Effective Ad Extensions? Ad extensions are pieces of information that can be tacked on to your ads to provide searchers with more details, and make your ads more interesting and inviting, such as: - Sitelink extensions: These provide additional links to specific pages on your website directly from the ad, and help connect users with exactly what they’re looking for. - Callout extensions: Short, non-clickable phrases that highlight unique selling points or features. For example: "Free Shipping," "24/7 Support," or "No Contract Required." - Structured snippet extensions: Display specific, pre-defined categories of information, like "Types" (e.g., Laptop, Desktop, Tablet) or "Services" (e.g., Oil Change, Tire Rotation). - Call extensions: Add a phone number to your ad, allowing users to call you directly. - Location extensions: Show your business address, map, and distance to nearby users. - Price extensions: Display prices for different products or services directly in the ad. - Promotion extensions: Highlight specific sales or promotions that the user might find interesting. ### How Can You Use Ad Extensions to Improve CTR and Ad Visibility? Ad extensions are easy to add on, but to max out their impact, use relevant extensions that genuinely add value, and always ensure that the information is accurate and up-to-date. Google automatically chooses which extensions to show based on what it thinks is most relevant, but by providing more options, you increase your chances of better visibility and a higher CTR. It’s all about giving the user more reasons to click, and extensions can help do just that. ## Measuring and Optimizing Paid Search Performance One of the great things about paid search is its measurability. If your ads are working, you know, and often know why (or not). ### What Are the Key Metrics to Track in Paid Search? - Impressions: The number of times your ad was displayed. This indicates your reach and visibility. - Clicks: The number of times users clicked on your ad. - Click-through rate (CTR): Clicks divided by impressions (as a percentage). A strong CTR indicates ad relevance and appeal. - Conversions: The number of desired actions taken after an ad click (like a purchase, lead form submission, download, etc). - Cost per acquisition (CPA): The average cost to generate one conversion. This is vital for understanding your profitability. - Return on ad spend (ROAS): Revenue generated from your ads divided by the cost of those ads. It shows how much revenue you're getting back for every dollar spent and is essential for e-commerce. Don’t forget about conversion tracking in paid search, either: Without it, you’re basically flying blind. Setting up conversion tracking correctly (via Google Ads conversion tags or Google Analytics) is the first, non-negotiable step to understanding the actual business impact of your ads. You need to know what actions users are taking after they click, otherwise, you're just optimizing for clicks, not the real results. ### How Do You Analyze Your Paid Search Performance Data? Analyzing data means looking for trends and anomalies, and it’s your key to constantly improving. Start by comparing performance over different time periods, across different keywords, ad groups, and campaigns. Also look out for: - Underperforming keywords/ads: High CPCs or low conversion rates might indicate an issue. - Well-performing keywords/ads: If something’s working, double down on it, even increasing bids or budget for them. - Audience insights: Are certain demographics or locations converting better? Dig deeper into that. - Device performance: Does your mobile performance match desktop, or does it need optimization? People are more likely to quit when a mobile website is all skewed and broken. ### How to Optimize Your Bids, Keywords, and Ad Copy for Better Results Optimization is a continuous, ongoing cycle. You need to constantly be looking for ways to improve from all angles, such as: - Bids: Adjust your bids based on performance, increasing them for high-converting keywords/ad groups, and decreasing for underperformers. Also consider automated bidding strategies once you have sufficient conversion data. - Keywords: Regularly review your search terms report to find new negative keywords and new positive keywords to add, and refine match types. - Ad Copy: Continuously A/B test different headlines, descriptions, and calls to action. Even small changes can bring big, impactful improvements. Continuous A/B testing (or split testing) — running two or more variations of an ad, landing page, or targeting setting simultaneously to see which performs better — is also a vital, data-driven approach that removes guesswork, as well as some of the pressure of choosing one line of copy, allowing you to systematically improve your campaigns. In terms of things to avoid, the biggest ones are not doing thorough keyword research, neglecting negative keywords, failing to set up conversion tracking, having irrelevant landing pages, and not continuously optimizing. And, don't ever just copy what your competitors are doing without understanding why they're doing it. ## Advanced Paid Search Strategies ### What Are Remarketing Lists for Search Ads (RLSA)? RLSA allows you to customize your search ad campaigns for people who have already visited your website. You can bid higher for these users, show them different, more tailored ads, or even target them with keywords you wouldn't normally bid on for cold audiences. They already know your brand, so they’re often much more likely to convert. ### How Can You Use Dynamic Search Ads (DSAs)? DSAs are better for businesses with large inventories or frequently updated websites, since instead of bidding on individual keywords, they target searches relevant to the content of your website. Google will automatically generate headlines for your ads based on the user's query and your website content. The benefit here is that it saves a lot of time on keyword management and ensures you're covering relevant, long-tail searches you might miss otherwise. ### What Are Shopping Campaigns and How Do They Work? Shopping campaigns are specifically designed for e-commerce businesses. They use product data feeds (your website’s product catalog) to generate visual ads that appear directly in the search results. Unlike text ads, you don't bid on keywords directly in shopping campaigns. Instead, you optimize your product feed and set bids for product groups, which can be effective for driving online sales. ### What Are Local Service Ads (LSAs)? Local Service Ads (LSAs) are primarily for service-based businesses (like plumbers, electricians, locksmiths) that operate locally. These ads feature your business at the very top of Google search results, often with a "Google Guaranteed" badge, and users can directly call or message you through the ad. The cost is per lead, not per click, so you often only pay for qualified leads. These ads have been a game-changer for local businesses. ### How Can Automation and AI Enhance Paid Search Management? Automation and AI are transforming paid search at a rate that’s hard to keep up with, even for those of us in the industry. But, smart bidding strategies (like Target CPA, Target ROAS, Maximize Conversions) can use all the capabilities of machine learning, and leverage them to optimize bids in real time to reach specific goals. AI can be a big help with tasks like generating the creative for ads, audience segmentation, and even anomaly detection. While human oversight is something that’s needed at this point, these types of tools can still help improve efficiency and performance, especially for large and complex sites where it’s too much for a human team to always be tracking. ## Key Takeaways Paid search is a dynamic and essential digital channel, offering instant visibility and the potential to capture high-intent traffic. This method has a lot of different strategies and subsets to it, but it’s overwhelmingly a positive thing for advertisers. You’ll still need to put the work in, though, so start getting a deeper understanding of keyword targeting, the auction system, ad formats, and a relentless focus on data-driven optimization. When you continually refine your bids, keywords, ad copy, and strategies, you can drive measurable results, and achieve your marketing goals, too. Paid search is a truly powerful tool, and when mastered, can deliver significant returns. ## Frequently Asked Questions (FAQs) ### What is a good click-through rate (CTR) in paid search? It’s all relative, and a "good" click-through rate can vary a lot based on several factors. First off, your industry plays a big role. Niche industries might see higher CTRs because their ads are specific to a small, engaged audience, but broader ones might have lower CTRs because of more competition and less specific intent. Your position on the search results page also matters; location and placement are both key factors. Ads at the very top generally achieve higher CTRs, but not always, and the quality and relevance of your ad copy and headlines play into it as well. Don’t overthink it, either — a compelling message that resonates with the search query will always perform better. The specific keyword match type you're using influences CTR, too. Exact match keywords are often going to have higher CTRs because of their precise targeting, while broad match can have lower CTRs, but greater reach. Honestly, don’t set your hopes too high when starting out, as anything above 2-3% is often considered decent for search ads, but top-performing campaigns in certain industries can see CTRs much higher, even into the double digits. Try not to compare your new venture to them, though: It's best to benchmark against your own past performance and industry averages rather than chasing a single, universal number. ### How much should I budget for a paid search campaign? It varies, since there’s endless types of businesses, budgets, and ways to run a campaign. But, in general, determining a budget for a paid search campaign is less about a fixed amount and more about a strategic allocation — one that’s based on your business objectives and takes the competitive landscape into consideration. A good place to start is considering your desired cost per acquisition (CPA) or cost per lead (CPL) and how many conversions you’re looking to achieve from there. So, if your average customer value is high, you're aiming for ten conversions per month, and your target CPA is $50, you'd need at least a $500 budget. Competitive industries will have higher click costs, though, meaning your budget will need to be larger if you want to achieve meaningful visibility. For those advertisers just beginning, I’d advise starting out with a small, conservative budget, even a few hundred dollars per month. This will help you gather initial data, identify high-performing keywords, and optimize ad copy and landing pages. Then, once you've reached a point of positive return on investment (ROI) and understand your metrics, you can scale up your budget to capture more market share. ### How does SEO work with paid search? SEO (search engine optimization) and paid search are two separate things, but have a lot of overlapping, complementary strategies that work together to maximize your visibility on SERP. SEO focuses on earning organic, unpaid rankings by optimizing your website's content, structure, and authority to be boosted by search engine algorithms. It’s a long-term play, but builds sustainable, free traffic. Paid search provides immediate visibility by allowing you to bid on keywords and display ads at the top of the SERP. This option offers instant presence for keywords where your organic ranking might be weak, or for highly competitive terms. That’s why, together, they create a formidable search presence — and you should utilize them both. ### What are some emerging trends in paid search advertising? The paid search landscape is always evolving and growing, with new trends popping up constantly, some because of technological advancements and others because of shifting user behaviors. One of the most significant emerging trends is the integration of artificial intelligence (AI) and machine learning — and it’s happening rapidly. You can see it in some of the advanced smart bidding strategies out there now, which use AI to optimize bids in real time based on a multitude of signals, aiming to hit specific conversion goals more efficiently. AI has also made its way into the creative side of things, aiding in ad creation and optimization, and automatically generating and testing combinations of ad copy to see what resonates best. Another trend, which isn’t exactly new but also isn’t going away, is the continued emphasis on audience-first targeting. Keywords still remain crucial, but advertisers are increasingly leveraging audience signals (like demographics, interests, behaviors, custom segments) to layer on top of keyword targeting, helping ads reach the right people and not just those searching for a specific term. The rise of visual search and alternative search interfaces, like voice search, image search, and even conversational AI, is changing the game and pushing advertisers to think beyond traditional text ads. Finally, there's a growing focus on privacy-centric measurement. As third-party cookies phase out, more reliance on first-party data and privacy-preserving solutions for tracking conversions and user journeys are coming up. Platforms like Realize are a useful tool, leveraging unique first-party data integrations with publishers to help advertisers achieve performance goals across the open web, even beyond the traditional search ecosystem. --- ### Last-Touch Attribution: What Is It, How Is It Best Used? URL: https://www.taboola.com/marketing-hub/last-touch-attribution/ Last Modified: 2025-07-15 09:59:07 What finally made a customer convert? That’s the big question that attribution models and analytics data aim to answer, helping marketing teams identify which touchpoints worked along the buyer’s journey. With multiple attribution models available, let’s see if last-touch attribution is the right one for your business. ## Defining Last-Touch Attribution When your team analyzes their marketing efficacy, they look to the touchpoints along the buyer’s journey that helped move the customer to the next step. The final place your prospect interacts with your business is their “last touch” before a sale, so last-touch attribution credits that interaction with the sale. For example, the last touch might be downloading a rate card or brochure from your website, or seeing an ad on social media. ### How Does the Last-Touch Attribution Model Assign Credit for Conversions? Other attribution models break down percentages, assigning a certain amount of credit to the interaction for each touchpoint. However, with last-touch attribution, 100% of the credit is given to the last touch, or the last engagement the potential customer had with the business. ### What Are the Key Characteristics and Assumptions of This Model? Last-touch attribution is based on a few key concepts, which can help you determine if it aligns with your brand’s philosophies and customer journey: - The last touchpoint is the biggest and most important factor in the sale. - Previous touchpoints warm up a prospect to the idea of purchasing, but don’t hold too much purchase power: Rather, they are increasing awareness of the company. - People make decisions impulsively, often with just one click, which happens in that last touchpoint. - The entire customer journey doesn’t have to be long, elaborate, or complicated. ## Advantages of Last-Touch Attribution ### In What Scenarios Might Last-Touch Attribution Be a Suitable Model? If you’re new to attribution models, last-touch attribution is one of the simplest and easiest to understand and can serve as an excellent starting point for evaluating the customer experience. This model is also ideal if you’re looking to understand what converted the customer to action, and can provide valuable insight to help drive future conversions. ### How Can Last-Touch Attribution Help in Quickly Identifying the Last Marketing Effort Before a Conversion? Last-touch attribution can quickly show you the data on exactly what converted customers — because it doesn’t share attribution with any other touchpoints, there’s little math involved in assigning credit. In addition, this model eliminates all of the leads that didn’t convert along the way, just focusing on the ones that did. That can help you narrow your research down to people who said “yes” and ultimately became customers, versus those who said “maybe” or “probably” along the journey, but didn’t take the final steps to make a purchase. ## Disadvantages and Limitations of Last-Touch Attribution ### How Does Last-Touch Attribution Potentially Undervalue Earlier Touchpoints and Their Influence? The last-touch attribution model has its flaws, particularly in not recognizing earlier touchpoints’ role in the customer journey. Since marketing is often a series of touchpoints, removing those first relationship-building impressions can undervalue the work that has gone into capturing a prospect’s attention at the outset. In short, it can be simplistic to assume the conversion that took place resulted entirely due to the last touch. ### Can Last-Touch Attribution Lead to Misallocation of Marketing Budget and Resources? Marketing teams should be aware that the most simple strategy isn’t always the best one. If you’re using a last-touch attribution model and placing all of your resources into that final conversion metric, you might overspend on the wrong touchpoint. While the final ad inside an offer that really pushes people to convert can be flashy and exciting, it doesn’t replace building trust with leads through multiple touchpoints leading up to it — a sometimes slow and deliberate process that’s just as worthy of budget and resources. ### Are There Specific Types of Conversions Where Last-Touch Is Particularly Misleading? Longer and more complex buying journeys, or products and services that are expensive or involve a long-term engagement with your company, might not be best choices for last-touch attribution. That’s because lead nurturing is an essential part of these more involved buyer’s journeys, including multiple interactions like demos or discovery calls, that prepare a lead for the opportunity at that last touchpoint. Additionally, some types of buyer journeys involve multiple platforms, such as thought leadership content, social media ads, blogs or webinars, product or service demos, and more, so only crediting a retargeting ad at the end of all that investment is incorrect. In addition, last-touch attribution doesn’t reflect the impact of each of those investments in lead nurturing. ## Alternatives to Last-Touch Attribution ### What Are Some Common Alternative Attribution Models? If you find that the last-touch attribution model isn’t the best fit, there are several alternatives, including: - First-touch attribution: All credit for the conversion goes to the first touchpoint, such as the first ad someone clicks on, or the first email they open. - Linear attribution: All touchpoints play an equal role in the customer journey, including the first and last. - Time-decay: Helpful for businesses with slower sales cycles, this attribution model emphasizes recent and bottom-of-the funnel interactions. - Lead conversion: If you’re looking to optimize your advertising budget, this model gives priority and credit to “milestones” along the way, and focuses on the channels that turn leads into customers. - Position-based: This model credits all touchpoints, but allocates a bit more credit to the beginning and end of the journey. ### When Might It Be More Appropriate to Use a Different Attribution Model? When determining the best attribution model for your marketing, it’s important to remember your goals. While the overall aim is to convert the prospect, the right attribution model can help you better understand your customers and what leads them to action — this knowledge can help you make better use of your marketing budget and resources for campaigns. ## Key Takeaways Last-touch attribution is a framework for crediting the final part of the customer journey with conversion. It is not always a complete picture of all the efforts that go into a full customer journey ahead of conversion, but can be a useful metric when determining resources to allocate to final touchpoints. ## Frequently Asked Questions (FAQs) ### Is last-touch attribution ever the best model to use? Last-touch attribution can be helpful if you’re looking for a simple attribution model and one with fewer errors. It can also be helpful in exceptionally simple marketing campaigns where there aren’t many touchpoints, and most of your proven value to leads is in the last touchpoint anyway. ### How does last-touch attribution compare to multi-touch attribution? Last-touch attribution focuses on a single touchpoint within the buyer’s journey, whereas multi-touch attribution examines the entire action process to determine which touchpoint led to the conversion. Several attribution models, such as position-based or lead conversion, are considered multi-touch because they credit multiple touchpoints with conversion. ### What are the implications of using only last-touch for reporting? While last-touch attribution is generally less prone to errors in comparison to attribution models that incorporate several touchpoints, last-touch eliminates the overall buyer’s journey and other factors that may have led to the customer’s conversion. It can ignore certain marketing touchpoints, such as TV ads, that might not be as easily measurable for conversion as an email newsletter or online ad. ### How can I implement last-touch attribution in my analytics platform? First, determine if your analytics platform’s default is last-touch attribution, as is sometimes the case. From there, you can often choose the type of attribution you want to use in settings. For some, you might need to manually assign credit as percentages, or otherwise follow the prompts to implement the attribution model you prefer. --- ### Negative Keywords: Definition, Benefits, Match Types URL: https://www.taboola.com/marketing-hub/negative-keywords/ Last Modified: 2025-06-30 08:12:11 As a copywriter for almost 20 years, keywords have played a major part in lots of the online advertising and writing I’ve done. Every click costs money in today’s online world, so it's more important than ever to show your ads to the right people. To that end, we spend a lot of time picking the best words for our ads, writing catchy messages, and setting bids with the ultimate goal of finding that perfect customer. But, knowing which words to avoid can be just as important as knowing which words to use. That's exactly what negative keywords do. They're like your ad campaign's helpful gatekeepers: They let in the people who are truly interested in what you offer, and politely turn away anyone who isn't. Without them, your ads could pop up for searches that have nothing to do with your business, causing you to waste money on clicks from people who won't buy, and meaning your ads just won't work as well. From my years in advertising, I can tell you that ignoring negative keywords is like leaving money on the table. These keywords are a key part of any good pay-per-click (PPC) plan, whether you're using Google Ads, Realize, or any other platform that can help you use and monitor your keywords. Getting this right isn't only about saving cash, either — it's about making your message clearer, giving people a better experience, and building ads that work smarter and harder for you. ## Understanding Negative Keywords A negative keyword in online advertising is a word or phrase that stops your ad from showing up when someone searches for that term. While your regular keywords tell the ad platform when to show your ad, negative keywords tell it when not to. It's a big difference, and it's what makes your ads hit the right target. If you don't use them, your carefully made ads might be seen by a huge group of people who aren't really interested, which wastes your money, your time and efforts, and makes your ads less effective. Search engines and ad platforms are smart, but they don't always know exactly what someone means. The same word can have many different meanings. "Apple," for example, could be a fruit, a tech company, or the Beatles’ iconic record label, and if you’re trying to sell fresh fruit while your ad shows up for someone trying to buy an iPad, that's a lost opportunity. A negative keyword like "iPhone" or "MacBook" would stop that impression (when your ad is seen) and, more importantly, hit the brakes on that useless click. This is why negative keywords are so essential, since they tackle two big challenges in online advertising: wasted spending and ad relevance. ### How Do Negative Keywords Prevent Your Ads From Showing for Irrelevant Searches? Negative keywords work by filtering things out. When a user types something into a search engine or looks at content on a website, the ad platform first checks what they typed or what the page is about against your regular keywords. Then, before your ad even thinks about showing up, it checks that same search or content against your negative keyword list. If there's a match with a negative keyword, your ad just won't show. It's a simple but super powerful way to avoid showing your ad to someone who has no intention of buying. I've seen campaigns completely turn around just by taking the time to build a strong list of negative keywords — it's an easy fix with big potential. ### What Are the Different Match Types for Negative Keywords? Just like regular keywords, negative keywords have different match types, which give you different levels of control over when your ads are stopped, and getting to know them is important for precise targeting. As an example, if you sell fancy clothes and don't want your ads to show for searches like "cheap clothes," just adding "cheap" as a broad negative might block too much, and you’ll most likely want a more specific match type. I'll get into it a bit more later on, but for now, understand that they give you the fine-tuned control you need to reach the audience you’re aiming for. ## Benefits of Using Negative Keywords There’s more to negative keywords than just keeping costs down: They’re essential for all-around efficiency, and for making your advertising truly relevant. When your ads are more relevant, more people who see them and want what they’re showing will click through. A higher CTR tells ad platforms that your ads are valuable and useful, which in turn can improve your Quality Score, too. ### How Do Negative Keywords Help Improve Campaign Relevance? Negative keywords are a powerful option in your online advertising toolbox for getting your ads in front of the right eyes. Let’s say you're running ads for your high-quality, custom-made furniture. Without negative keywords, your ads might appear in searches like "cheap furniture assembly," which, while including "furniture," is clearly not a search aligned with the pieces you’re making and selling. By adding "assembly" as a negative keyword, you make sure your ads only show to people truly looking for your level of quality furniture. This directly makes your campaign more relevant because you're reaching people whose needs perfectly match what you offer. Don’t think of your negative keyword list as a one-and-done, though: You should be refining it every so often based on the data and trends you’re seeing. Every time you clean up your negative keyword list, you're sharpening your campaign's focus. ### How Can Negative Keywords Reduce Wasted Ad Spend on Irrelevant Clicks? This is often the first and most obvious benefit. Every time someone clicks your ad who isn't a potential customer, you're paying for nothing, and it's basically throwing money away. Negative keywords act like a shield for your budget, stopping those costly, irrelevant clicks. By filtering out people who aren't qualified, you ensure your ad money goes only to users who are truly interested, making your spending that much more efficient. ## Identifying and Implementing Negative Keywords ### How Do You Identify Potential Negative Keywords for Your Campaigns? One of your best tools is going to be the search term report in whatever advertising platform you’re using. Looking at this report shows you the exact words people typed that made your ads appear, and it’s a goldmine of insight. It runs deeper than that, too, with key clues on what your next move to improve your campaign should be. It's not about what you think people are searching for, but what they actually searched for, so check it every week, especially when campaigns are new. Here are a few things to look for: - Completely unrelated terms: These are low hanging fruit, and easy wins to block. - Terms with low sales: If a term gets clicks but no sales, it might mean the person wasn't really looking for what you offer. - Terms that show low buying intent: Such as "How to build a website for free" for a web design company. Besides the search term report, regular keyword research tools (like Google Keyword Planner, Semrush, Ahrefs, Realize, etc.) can also help. When you're looking for good keywords, pay attention to related terms that pop up, especially those that might suggest a different goal. Another approach is to do what we creatives do: brainstorm. Get into character as your perfect customer, and then think like someone who isn't your customer but might use similar words. Think about common first-searched terms like "free," "cheap," "jobs," "reviews," "DIY," or even specific brand names you don't sell. ### What Are Some Common Categories of Negative Keywords? Whatever product or service you’re offering is going to have its own set of niche words you’ll want to include and exclude, but here's a quick list of common negative keyword types that businesses in general can use: - Free/low-cost words: "free," "cheap," "discount," "coupon," "bargain." Unless you actually offer free services, these often bring in people who aren't serious buyers. - Job seekers: "jobs," "careers," "employment," "hiring." If you're selling a product and not offering employment, you don't want people who are looking for work clicking your ads. - Research/information: "what is," "how to," "examples," "pictures," "reviews." While some research can lead to a sale, often these users are just gathering info and aren't ready to buy. - Competitors: If you don't want your ads to show up when people search for your competitors' names, make sure to include them on your negative list. - Wrong product types: For example, if you sell men's shoes, you'd want to block "women's shoes." - Adult content: If your product or service has nothing to do with adult content, it's smart to block terms that might be related to it. Just in case. ### How do you add negative keywords to your advertising platform (e.g., Google Ads)? Most advertising platforms make this pretty easy. In Google Ads, you'll usually find a "Negative keywords" section within your campaigns or ad groups. Simply type in the words or phrases you want to block and track the results. As far as using negative keywords at the campaign or ad group level, it really depends on how you've set up your campaign. - Campaign level: Use negative keywords here if they’re completely useless across all the ad groups in that campaign. Going back to the example above, if you only sell new cars, "used" would be a negative keyword for the whole campaign. - Ad group level: Use this when a negative keyword should only stop ads in one specific ad group, but might be fine for another. My general advice: Start at the campaign level for big blocks, then get more specific at the ad group level. ## Negative Keyword Match Types in Detail ### How Do Broad Match Negative Keywords Work? A broad match stops your ad from showing if the search query contains all the words in your negative keyword, no matter the order, and often includes the similar words around it too (like plurals, synonyms, and even misspellings). For example, if you add "free download" as a broad match negative, your ad won't show for: "free music download," "download free software," "downloading free games." But, your ad might still show up for: "free music," and "software download" (because "download" and "free" aren't together, or the meaning is interpreted differently.) Use broad match for the really obvious, universally unhelpful terms you want to block aggressively across all variations. ### What Are the Nuances of Phrase Match Negative Keywords? A phrase match negative keyword stops your ad from showing if the search query includes the exact phrase of your negative keyword, in the exact order — it can still have other words before or after the phrase, too. So, if you add "running shoes" as a phrase match negative, your ad won't show for: "best running shoes for sale," "discount running shoes." But, it will show for "shoes for running" since the words are in a different order, and "running fast shoes" because a word is stuck in the middle. These are really useful tools that give you good control without being too strict. I suggest using them for the specific irrelevant phrases you find in your search term reports. ### When Should You Use Exact Match Negative Keywords? An exact match negative keyword only stops your ad from showing if the search query is exactly the same as your negative keyword, with no other words before, after, or in between. So, as an example, if you add "" as an exact match negative, your ad won't show for "cheap watches,” but will for "buy cheap watches," and "cheap watch." These are good for when you need super precise blocking. They're also great for stopping the very specific, unhelpful searches you've found, making sure you don't accidentally block good searches that have slight changes within them. ### How Do Different Match Types Impact Ad Triggering? Different match types directly affect when your ads show (or don't show). Broad match negatives cast the widest net, but the problem is that they may block more than you mean to if you're not careful. Phrase match gets more precise, since that blocks only specific word combinations, but exact match is the most precise since they block only the exact search. Choosing the right match type for each negative keyword is crucial to making your ads work better. Remember that it's always about finding the right balance between saving money and making sure you don't miss out on potential customers. ## Best Practices For Negative Keywords ### How Often Should You Review and Update Your Negative Keyword Lists? Often! Don’t get too obsessive about it, but I highly suggest checking your search term reports at least once a week if you’re just starting out or it’s during a busy time. For campaigns that have been running for a while, once a month or every two weeks might be enough. The online world changes and evolves all the time, and so do the ways people search: What was relevant last month might not be now, so make sure to keep up with your keywords frequently. ### What Are Some Advanced Strategies For Negative Keyword Implementation? Once you’ve got the hang of managing the basics, take a deeper dive with these methods: - Negative keyword lists: Create shared lists of common negative words (like "free" or "jobs") and use them across multiple campaigns. Doing this saves you time and keeps things consistent. - Competitive negatives: If you don't want to show up in searches for your competitors' names, make sure those are strong exact and phrase match negatives. This prevents accidental views or clicks if people search for "(your brand) vs. (competitor’s name)." - The "zero conversions" rule: If a search term has gotten lots of clicks but zero sales over a long time, it's a strong candidate to become a negative, even if it seems a little related. What people intend to do matters more than just the words they use. ### How to Avoid Accidentally Blocking Relevant Searches With Negative Keywords This is where it gets tricky, since you don't want to go too far and block good customers by mistake. Start with a phrase or exact match for anything you're not sure about — you can always make it broader later. Also, be sure to watch your ad performance closely: If your website traffic suddenly drops, or you're getting fewer sales, check your recently added negatives and make adjustments. Look into using the "Conflicts" tool — some platforms will warn you if a negative keyword might be blocking one of your good keywords. ### How to handle plurals, misspellings, and variations of negative keywords Unlike regular keywords, negative keywords don't always automatically cover similar words like plurals or misspellings for broad or phrase match. If you add "shoe" as a broad match negative, there’s a chance it might block "shoes," but it's not guaranteed for every situation or platform. So, for really important negative terms, it's often smart to include common plurals and misspellings (e.g., "cheep"). For exact match negatives, you’ll need to be really precise; for example, "" won't block "." ## Key Takeaways Negative keywords are much more than just a list of words to avoid — they’re a smart way to make your advertising more efficient and effective, and one that you can learn to perfect based on what matters to you and your campaign. By carefully finding, using, and updating them, you can help ensure that your ad money is going to the most valuable audience. This is what’s going to improve your CTR, Quality Score, and most importantly, what you get back from your ad spending. It's an ongoing process, but one that definitely pays off in the long run. ## Frequently Asked Questions (FAQs) ### What is the difference between a regular keyword and a negative keyword? The big difference between a regular keyword and a negative keyword is the job they do in your ad campaign. A regular keyword is a word or phrase that you want your ad to show up for when someone searches for it or looks at related content. Negative keywords are the flipside of this, a keyword or phrase that tells the ad platform to not show your ad. The goals of a negative keyword are to block unhelpful traffic, save money that would be wasted on clicks from people who won't buy, and make sure your ads are only seen by those most likely to become customers. ### Can negative keywords hurt my campaign performance? Yes, they can! While negative keywords are incredibly useful for making your ads better and more precise, using them incorrectly (or too much) can definitely hurt your campaign. The biggest risk is being too aggressive and adding negative keywords that accidentally block searches from genuinely interested people out there. Another way they can cause harm is by narrowing your audience too much. It’s good to be targeting the right users, but if you have too many specific negative keywords, you might unnecessarily reduce how often your ads can be seen and clicked. ### How many negative keywords should I use? There's really no magic number for how many negative keywords you should use, and it ultimately depends on your business, how specific your product or service is, and how broad your initial keywords are. A very niche product might need fewer negative keywords than a general service that could be searched for in many irrelevant ways. Some campaigns might work perfectly fine with just a few dozen negative keywords, while others could benefit from hundreds, or even thousands. What's even more important than the number is the quality and relevance of your negative keywords. Focus on finding and using the right ones — those that consistently appear in your search term reports as irrelevant, or those you can expect will attract people who aren't qualified buyers. ### What are some examples of effective negative keywords for different industries? Finding effective negative keywords can depend a lot on your specific business, but here’s a list of some common examples across different industries: #### Online shops (selling products): - "Free," "cheap," "discount," "used," "second hand," "DIY": Unless you offer these, users searching terms like this usually means they’re looking for something free, very cheap, used, or want to build it themselves, not buy a new product. - "Review," "forum," "blog," "wiki": These often mean someone is just looking for information, not ready to buy. - Specific competitor brands: If you don't sell those brands, blocking them stops mistaken clicks. - "Jobs," "career," "hiring": For ads selling products, these are irrelevant because they mean someone is looking for work. #### Service businesses (e.g., plumbers, web designers): - "Jobs," "careers," "training," "certificate": These block people looking for work or courses in your field. - "DIY," "how to fix," "tutorial," "guide": Blocks people trying to do the service themselves instead of hiring a professional. - "Free," "pro bono": If your services cost money, these block people looking for free help. - "Reviews," "opinions," "forum": Again, this often means someone is gathering info, not looking to hire. #### Software/SaaS companies: - "Free trial" (use with caution): Only use this if you specifically want to avoid people mainly looking for free access, or if your free trial isn't meant to bring in sales. - "Crack," "torrent," "keygen," "pirated": Essential for blocking searches related to illegal software downloads. - "Jobs," "developer," "API": Unless your campaign is specifically for developers or hiring, these terms are usually not going to be relevant. - "Login," "support," "help": These indicate existing customers looking for assistance, not new leads. #### Real estate: - "Rental," "lease," "apartment," "for rent": If you only sell properties, these terms are irrelevant. - "Foreclosure," "auction," "bank owned": If you don't deal with these types of properties, block them. - "Jobs," "agent training": Blocks people looking for careers or training in real estate. - "Free appraisal," "estimate value": Unless these are what you use to get leads, they can attract people not ready to buy or sell. This is all just a jumping-off point — the best negative keywords will always come from carefully checking your own campaign's search term reports and really understanding your target audience and how they search. --- ### Fashion and Beauty Marketing Trends in 2026 URL: https://www.taboola.com/marketing-hub/fashion-beauty-marketing-trends/ Last Modified: 2026-03-08 13:38:58 Those in the fashion and beauty marketing world are all too aware of how fast the industry moves, and with all kinds of technical changes happening all the time, it’s only getting faster. As a result, the pressure to stay visible to consumers across platforms is higher than ever. In 2026, winning brands are not those with the biggest budgets or flashiest launches, but those that can tell cohesive stories across social, commerce, and the open web while meeting rising expectations around transparency, personalization, and value. Consumers are savvier and more selective, more conscious of how brands show up in their feeds, inboxes, and content environments. This means marketing strategies must be both emotionally resonant and operationally sophisticated to succeed. ### What’s changed in our 2026 update: - All entries include updated and current information, figures, and stats. ## Trend 1: Visual Storytelling Evolves From “Authentic” to Trust-Driven Content While content is still vital for marketing efforts in 2026, it doesn’t look the same as it once did. Fashion and beauty especially must be relatable, particularly when appealing to a Gen Z and millennial audience who now make up, in many cases, the largest proportion of the customer base. “Authenticity” remains essential in fashion and beauty marketing, but now, it’s evolved from a creative style into a credibility requirement. Consumers, particularly Gen Z and younger millennials, are increasingly skeptical of overly polished or transactional brand messaging. Instead, they gravitate toward content that feels informative, transparent, and rooted in real experiences rather than aspirational fantasy alone. An uncertain economic outlook, along with the growth of “real life” social media, has seen a shift from perfection in fashion and beauty advertising to more realistic and transparent campaigns. Documentary-style, high-quality content has become increasingly popular in everything from campaign billboards to TikTok posts from fashion and beauty brands. Influencer marketing has also matured significantly. While macro-influencers still play a role in brand awareness, micro and nano creators are now core performance drivers, particularly when paired with affiliate models and paid amplification. Many creators are no longer just endorsers but are distribution partners, running ads, hosting live shopping events, and contributing creative assets directly into brand performance funnels. Short form video remains dominant, but longer-form, educational video and immersive native formats are gaining renewed importance in 2026, particularly on the open web. Fashion and beauty brands are increasingly using native and display placements to extend storytelling beyond social feeds, meeting consumers in high-attention environments where discovery and consideration happen more deliberately. Native advertising continues to be a powerful growth tool for visual storytelling, allowing brands to scale trusted content formats in editorial-style environments. Campaigns that blend creator-led visuals with native placements are proving effective at driving both engagement and conversion, especially when paired with retargeting and sequential messaging. Stats you should know about authentic marketing in fashion and beauty: - 85% of Gen Z consumers say that authenticity is important when choosing a brand to support. - 71% of Gen Z consumers have been influenced to make a beauty purchase because of a TikTok video. ## Trend 2: The Dominance of Social Commerce and Live Shopping in Fashion and Beauty Social commerce has become the new storefront in recent years, with platforms like TikTok Shop and Instagram Checkout blending content and commerce seamlessly. These channels have become as important as brand websites for those in the fashion and beauty industries: Live shopping events have become major revenue drivers, offering an interactive experience for customers to ask questions and make purchases in real time. For businesses looking to integrate a social commerce aspect into their marketing this year, it’s vital that the infrastructure is put in place to make this successful. Product tags should be included in all posts, directing users straight to the social shop to make a purchase. Livestreams — already popular in the Asia-Pacific region, but emerging elsewhere too — should become exclusive events, offering scheduled product drops or a livestream-exclusive discount or bonus. Using platform-native formats, like TikTok’s live cart integration, can also boost revenue when launching a social commerce aspect to your business. Digital advertisers can also blend both organic and paid posts as part of a hybrid social commerce marketing strategy. Running boosted live sessions or retargeting via certain platforms can reduce customer acquisition costs (CAC) while increasing return-on-ad-spend (ROAS). Outside of social platforms themselves, display and native advertising placements are supporting these social-first strategies through promoting live events, new launches, and social shopping opportunities. Social commerce stats you should know for 2026: - Brands integrating paid amplification with social commerce see up to 25% lower CAC compared to organic-only approaches. - 76% of customers will buy from a brand they feel connected to on social media. - Nearly 44% of U.S. TikTok users bought something from TikTok Shop within the last year. ## Trend 3: Personalization and Inclusive Marketing in Fashion and Beauty Consumers have come to expect a high level of personalization in the marketing they see — gone are the days of simply “Hi ,” in an email. With dynamic tools that enable data to be shared and leveraged across different platforms, product recommendations, landing pages, and even predictive design can be incorporated into a personalized marketing campaign. In 2026, personalization has become more about experience design than anything else. Tools like AI-powered quizzes, along with more traditional digital marketing data sources like website and social media traffic and behavior, can be used to map demographics with purchase history to ensure that recommendations are as personalized as possible. Not only is personalization important, but so is inclusivity. This is no longer accepted as an afterthought or a campaign gimmick, but instead a baseline expectation from fashion and beauty customers. Brands that show a diverse range of ethnicities, body types, gender identities, and appearances are typically seeing stronger engagement and a deeper sense of brand loyalty among customers. It’s essential that this is reflected across both ad creative and messaging for a successful campaign. Those who fail to do so are increasingly perceived as out of touch by consumers. The numbers don’t lie when it comes to inclusivity and diversity in fashion and beauty marketing in 2026: - Brands that prioritize diversity outperform their peers by 36%. - Brands using advanced AI-driven personalization report revenue lifts of 10–15% year over year. - Sales increased by 15% for small to medium businesses who used AI-personalization in their fashion and beauty brands in 2024, over 2023. - 75% of customers said they’re more likely to shop with inclusive brands than those that aren’t. ## Trend 4: Leveraging Augmented Reality and Virtual Try-Ons in Fashion and Beauty When it comes to new technology, augmented and virtual reality (AR and VR) have changed the shopping experience for many customers. Beauty brands are using these tools to let customers virtually try on lipstick shades or foundation in real time, helping them make purchases that suit them best. The fashion industry is also experimenting with virtual fitting rooms, where customers can see how an outfit would look on their body without ever trying anything on. Platforms like Instagram, Shopify, and BigCommerce are all making this possible with easy access to AR tools, with many offering integration directly onto product pages. Not only are these fun ways for customers to play with products in a way that works for them, but it also reduces return rates anywhere from 5% or more when customers know for sure that a product is what they’re looking for. Advertisers can also use AR try-on functionality in ads, through tools like shoppable AR lenses, to boost ad engagement. Not only that, but one of the biggest benefits to this type of technology is the sustainability practices it can help support (more on that later). Immersive experiences housed within native placements are sparking curiosity among consumers at the top of the funnel, where discovery is a critical goal. AR and VR allow brands to put together virtual product displays for customers, rather than having a physical display where test products are used before a purchase decision is made. This prevents excess waste while helping to keep costs lower and focused on inventory for sale. AR shopping stats you need to know in 2026: - 61% of shoppers prefer to use retailers that have an AR experience. - AR is predicted to grow to over $9 billion by 2032. - AR and VR in the beauty industry is projected to grow by over 25% before 2027. ## Trend 5: The Role of Sustainability and Ethical Practices in Fashion and Beauty In an ongoing climate crisis and concerns around the sourcing of the goods we bring into our homes, consumers are increasingly voting with their wallets and choosing to support brands that align with their values. Resale, refill, and recycle are all buzzwords being used in the fashion and beauty industries more regularly, with movements toward slow fashion and clean beauty picking up traction in the last few years. How do brands communicate their efforts toward being more sustainable and ethical, though? Transparency is essential: This means clear web pages dedicated to sustainability and product origins, along with reports demonstrating clear intent to be more ethical and sustainable. Some beauty and fashion brands have taken steps to become Fair Trade or certified B Corps, meeting strict standards around ethics and sustainability over both their products and how their businesses operate. In 2026, consumers are far more discerning about how brands communicate their efforts. Vague claims or surface-level messaging are increasingly met with skepticism, pushing brands toward clearer, data-backed sustainability narratives. Visual branding is one of the best ways to quickly communicate this in stores and online. Leaping Bunny certification for beauty brands can be added to product packaging, along with B Corp status. QR codes linking back to sustainability web pages can also be a helpful tool for keeping customers informed and aware. Sustainability stats for beauty and fashion businesses in 2025: - 62% of Gen Z shoppers prefer to buy from sustainable brands. - 73% of customers are willing to pay more for sustainable products. - 60% of beauty shoppers say that eco-friendly packaging has influenced a purchasing decision. ## Key Takeaways The goal for fashion and beauty brands in 2026 shouldn’t be to chase every new trend, but instead take intentional steps to promote their products to an ever-changing customer base. Authenticity, inclusivity, interactivity, and sustainability should be the foundation to any marketing campaign, helping you drive greater ROI and build long-term trust with your community. ## Frequently Asked Questions (FAQs) ### What are the most effective digital advertising channels for fashion and beauty in 2026? Visual channels such as display, video, and social are some of the most effective channels for marketing in these industries in 2026. This is demonstrated by the explosive growth in TikTok, Instagram Reels, and Pinterest. Live video commerce platforms like TikTok can also help drive both traffic and revenue. Display ads are also proving to be a powerful tool for brands to interact with consumers between discovery and purchase. ### How can fashion and beauty brands build brand loyalty online? By creating engaging, personalized content with inclusive models and transparent values, fashion and beauty brands can build a strong customer base and long-term loyalty online. ### What are some successful examples of digital fashion and beauty campaigns? Campaigns such as Fenty Beauty’s UGC foundation challenge and Glossier’s customer-driven storytelling are two of the best recent examples of fashion and beauty brands creating authentic, community-first marketing content. Using both platform-first and advertiser-first data, businesses in these industries are able to effectively segment their campaign audiences to target more specifically and upsell across different audience groupings by factors such as gender identity, and retargeting for abandoned carts or based on browsing history. ### How is AI impacting the fashion and beauty marketing landscape? AI is powering tools that allow businesses in these industries to further personalize content and give customers the opportunity to try items before making a purchase. Chatbots are supporting customer service teams, while predictive analytics and customer segmentation from gathering data is helping marketing teams tailor campaigns and boost performance. ### What are the key metrics for measuring success in fashion and beauty digital marketing? Return on ad spend (ROAS), customer acquisition cost (CAC), conversion rates, product return rates, and customer lifetime value are some of the most important metrics to track in fashion and beauty. --- ### Six E-commerce Marketing Trends in 2026 URL: https://www.taboola.com/marketing-hub/ecomm-marketing-trends/ Last Modified: 2026-03-08 13:52:26 When you work in e-commerce marketing, you have to be ready to constantly embrace new tech. In recent years, the industry has gotten to grips with everything from augmented reality (AR) and virtual reality (VR) to sophisticated AI chatbots and live-streamed shopping experiences. Despite this, the goal is always the same: Attract more customers and keep your current ones coming back for more. It’s clear that growth is continuing amid these maturing tech platforms and channels. In 2023, researchers predicted that e-commerce would pass the $1 trillion mark: now, new data shows that number was more than 5x higher than expected, at $5.6 trillion. The same research shows that e-commerce is expected to grow to 22.9% of total sales worldwide by 2028. With that in mind, there’s a lot to consider in attracting and keeping e-commerce customers. There are six trends in particular to be aware of, which I’ll get into below, but first, some memorable stats: - Nearly 86% of people made an online purchase within the last month, with 36% spending less than $200 and 36% spending between $200 and $400. - Revenue in the e-commerce market is projected to reach $3.89 trillion in 2026. - The highest total e-commerce revenue is in fashion, with a total of $990 billion. - There are about 2.71 billion online shoppers as of 2024. ### What’s changed in our 2026 update: - All entries include updated and current information, figures, and stats. - Key e-commerce seasons and dates updated for 2026. ## Trend 1: The Continued Growth of Personalization and Customer Experience The convergence of big data, powerful analytics tools, AI capabilities, and advances in digital advertising technology all mean that modern e-commerce companies have the tools to offer interesting, personalized experiences. Here’s more detail on how businesses are moving personalization and customer experience forward: ### Using Data and AI for Personalization With rich, up-to-date first-party data, e-commerce businesses can create personalized customer experiences. Relevant data points include customer behavior and history, along with demographic and other gathered information, to segment audiences and build tailored buyer journeys. E-commerce businesses are also using predictive analytics and AI-powered recommendations for faster, automated customer recommendations. E-commerce companies have to pay close attention to price, too, particularly in economically challenging times. Last year, 60% of consumers switched from a brand they were loyal to because of cost considerations. Marketers can use all this information to surface relevant product suggestions, show unique website content in real time, send personalized messaging, use email strategically, and highlight price drops, special offers, or better cost vs. competitors. Plus, incorporating AI capabilities into e-commerce campaigns can help marketers scale and change directions quickly as needed, to stay ahead of market shifts or consumer preferences. ### Creating Exceptional Digital Experiences In addition to personalized experiences, modern consumers also assume they’ll get nicely designed, dependable shopping and buying journeys on any site or platform. Shoppers will quickly navigate away if a site doesn’t use responsive design for mobile or loads too slowly. Creating pleasant browsing experiences can also help stand out from the competition, whether that’s curated collections, advice available via chat, virtual try-on, outfit or accessory suggestions, and more. E-commerce users also require intuitive navigation and one-stop checkout. Marketers should make sure to reduce the number of steps required of a shopper to increase conversions, sales, and returning customers. The whole experience should be easy and intuitive: that includes transparent shipping and delivery data, return options, multiple ways to reach customer service, and offering as many payment methods as possible to make it easy for users to buy. Clear communication and an easy purchasing process also helps build trust and authority for e-commerce brands. Here’s what else to know: - Conversion rates can increase by 17% for every one second faster that a site loads. - $260 billion of lost e-commerce orders can be recovered by creating a better checkout design. - 39% of users abandoned their cart during checkout because extra costs were too high. - Personalized content can increase revenue more than 25% and improve a customer’s chances of spending more than planned by more than 40%. ## Trend 2: The Dominance of Mobile Commerce and Shoppable Content Mobile commerce has overtaken desktop browsing, and continues to grow. E-commerce keeps pushing the limit on what’s possible for shoppers and buyers, too, so here’s what to keep an eye on. ### Thinking Mobile-First E-commerce companies should put mobile browsing and shopping first, but also keep in mind the broader omnichannel market, and the more complex buyer’s journey that mixes online and offline browsing. A customer might buy products online and pick up in-store, or buy online and return in-store, or make part of a purchase in-store and the rest online if a product isn’t available in-person. Convenience is everything for modern shoppers. As consumers move quickly across search, social, and other platforms, it’s important to make sure the brand and the products are represented consistently and positively. ### Going Live Mobile e-commerce has become commonplace, and live-streamed, real-time shopping experiences are likely to be the next big step. It takes some creativity and innovation to stand out from the crowd with live e-commerce, such as with product demos, exclusive product offerings, and partnerships with influencers to reach more viewers. ### Tech Tools Combining audience knowledge with available technology can lead to some positive outcomes. An app, for example, acts as an entirely owned channel to reach and convert prospects: That could include app-only offers, push notifications, and other techniques and tactics that make sense. Another new technology in circulation with e-commerce brands is AR, which can bring prospects closer to the products virtually. Here’s what else to know: - Nearly 80% of all retail website visits were on smartphones worldwide in 2025. - 40% of consumers would pay more for a product they could customize in augmented reality. ## Trend 3: The Power of Social Commerce and Influencer Marketing in E-commerce Lots of e-commerce innovations over the past few years have been related to social media and entities outside of the traditional brick-and-mortar businesses. Consider the following: ### Social Media Evolves E-commerce businesses have found new, varied ways to monetize social media platforms. The platforms themselves have added features so that e-commerce companies can set up direct-sale shops inside of Instagram, TikTok, and others. To make this go smoothly, make sure there are resources to support this type of selling. In addition, use the available features on the particular channel, like shoppable ads, along with contests or live shopping options to engage audiences. ### Influencers Take the Spotlight E-commerce marketers and advertisers might be driving most of a business’ tactics, but influencers can be hugely important for brands to attract new prospects and drive conversions. It’s important to choose the right influencer for the brand, and pick the partnership type (e.g. sponsored content, product collaborations, brand ambassadorship, etc.) that best fits the business, too. Influencers can help build trust with testimonials, photos, videos, and other authentic, credible content. Consider how upcoming generations factor into this, too. Recent data shows that Gen Alpha may already be impacting the way in which purchases are made, with 70% of parents saying they buy products for their children based on their child’s favorite character or show fairly often. “For Gen Alpha, what's going to be really imperative for e-commerce brands is kind of shifting back to the 1950s, where you're sponsoring content and having native placements,” points out Lauren Petrullo, CEO and founder of Mongoose Media and co-founder of Asian Beauty Essentials. “You're doing product placement inside the programming, because most Alphas think, ‘How dare you assault me with a commercial. The audacity to steal my attention for your business!’ But, if you do it well, and you can interrupt that space, there's going to be such a good market for you.” Also keep in mind that Gen Alphas can be business owners themselves, and the distribution of generational wealth is different than previous generations. “Gen Alpha includes millionaires before 14, because these are YouTubers,” says Petrullo. “These content creators are doing live shopping to provide a hybrid experiential type of environment for those that are virtual and can't be physically in there, and for those that don't have stores anymore, because retail is a hard experience for a lot of people. This is the first time that we have generational wealth established before someone becomes an adult — they have employees and they have massive purchasing power.” Measuring the ROI of social and influencers can include attribution and channel-specific numbers like reach, engagement, and conversions. Here’s what else to know: - More than 50% of e-commerce brands spend at least 20% of their marketing budget on social media influencers. - The top social media app for e-commerce purchases is Facebook. - 74% of consumers have purchased a product because an influencer recommended it. ## Trend 4: The Importance of Data Analytics and Marketing Automation in E-commerce Without trusted data and accompanying technology, e-commerce businesses likely can’t keep up. These broad areas of focus can help companies build successful pipelines. ### Data Analytics With an ongoing shift away from third-party cookies, e-commerce businesses are gathering first-party data from customers — generally more accurate than third-party, and a way to build deeper relationships. Analyzing this data alongside all the other information a business collects offers a window into what’s working, and what isn’t. Data points can include customer interactions, purchase history, and demographics; individual product performance, average order value, and customer lifetime value; and conversion rate along with traffic data from web, social, and other channels. The results of analyzing this data can lead to personalization and better journeys, but also cost savings, reduced overlap or redundancy, improved customer retention, better cart completion, and more efficient campaigns. E-commerce marketing teams already do A/B testing to improve their metrics, but data analytics can be applied to areas like inventory management, as well, and at an advanced level can provide predictions on sales trends and other areas. It’s these types of approaches that help e-commerce businesses to truly make data-driven decisions. ### Marketing Automation and Other Tools E-commerce marketing isn’t a task that can be done manually in 2026: For even startups or very small businesses, the work of reaching and converting prospects intelligently and efficiently across multiple channels is simply too much. AI and other automation advances in online marketing mean that marketers can now perform data analytics and build a customer journey accordingly, then deliver and manage communications across channels, including email and social. Using marketing automation for e-commerce can streamline customer communication to eliminate excess contact, too, and many platforms incorporate AI for chatbots and assistants. Here’s what else to know: - 70% of e-commerce brands will rely on real-time analytics by this year. - 51% of shoppers prefer e-commerce businesses with live chat support. - Data-driven e-commerce companies are 23x more likely to acquire customers and 19x more likely to be profitable. ## Trend 5: The Evolving Role of SEO and Content Marketing in E-commerce SEO remains a crucial piece of e-commerce customer acquisition in 2026. As the discipline continues to evolve, e-commerce marketers should build a solid strategy with new opportunities in mind. ### Using SEO for Acquisition E-commerce teams can use SEO and AEO or GEO (answer or generative engine optimization, i.e., content designed to be picked up by ChatGPT or Google’s AI Overview) to attract targeted traffic, boost conversions and engagement, and build and maintain brand visibility. Paid search is also still a viable tactic, depending on budget and market saturation, but organic search serves as a low-cost, always-on channel that can steadily drive traffic. Good e-commerce SEO and AEO/GEO practices include optimizing web product pages, keyword targeting, and creating useful content for prospects at all funnel stages. Content marketers in e-commerce companies should use rich keywords and aim to inform and educate readers about products and trends on the path to conversion. Content marketing tactics in e-commerce range from reviews and user-generated content to video and audio. ### Voice Search Voice search is an emerging trend in e-commerce, as smart speakers and phone assistants become more sophisticated and useful. Consumers can shop easily (then purchase and track orders) with these hands-off options, so putting resources into voice search optimization is something marketers should consider. Get to know the tenets of voice search, such as optimizing for natural language and using long-tail keywords. Here’s what else to know: - 32% of consumers globally are using voice assistants weekly, and voice assistant users are 33% more likely to have made an online purchase in the past week. - Product pages that used the exact keyword in the meta description ranked better than those that didn’t. - 43% of e-commerce traffic originates from organic Google searches. - 57% of users won’t buy from a business online if it has less than a four-star rating. ## Trend 6: The Rise of Re-Commerce As sustainability continues to grow to top of mind in shoppers' priority lists, re-commerce — i.e., buying and selling used products online — continues to gain popularity, with 65% of shoppers aged 18-24 likely to have bought preowned goods within the last 12 months. Brands are having fun with it, and customers are enjoying reduced prices: For example, children’s clothing brand Hanna Andersson has “Hanna-Me-Downs,” reflecting the demand in the secondhand fashion market, which is projected to reach $367 billion by 2029. ## Key E-commerce Seasons and Dates in 2026 New Year: January provides a chance to clear holiday inventory while capitalizing on the “new year, new me” mindset for fitness and seasonal products. Super Bowl: The Super Bowl (February 8, 2026) often boosts demand for campaigns centered on snacks, drinks, and home entertainment. Marketing teams can use this as a comeback moment for customers who haven’t engaged in the buying cycle after a leaner post-holiday spending month in January. Valentine’s Day: This couple-focused day offers a prime moment to focus on personalized gifts, luxury items, and memorable experiences. Black History Month: February is ideal for spotlighting Black-owned businesses, creators, and cultural contributions. Authentic storytelling, along with curated collections and guides, can help brands connect with value-driven consumers. Tax season, wellness, and Presidents’ Day: Presidents’ Day (Feb. 16, 2026) is a strategic time for promotions on home goods, appliances, and tech. This is also when tax refunds start encouraging big-ticket purchases. Spring break and spring cleaning: Spring break planning might boost travel-related purchases, in addition to spring cleaning content and product opportunities. Memorial Day: With summer officially here, promotions around outdoor gear, travel, and family gatherings will all resonate well. Fourth of July: Focus on summer fun here, with bold, festive visuals. Highlight limited-time offers on BBQ supplies, summer- and holiday-friendly apparel, and outdoor products for back yards, beaches, and pools. Amazon Prime Day: Now several days in July, use this time to optimize your Amazon listings and ads to capitalize on high-intent shoppers and increased traffic, emphasizing urgency and exclusivity, especially for deals on tech and home products. Back to School: Beginning in August, marketers should plan to target parents and students with budget-friendly bundles, organizational tools, and time-saving solutions, using messaging around preparedness and new beginnings. Halloween: Spooky season is the time to have fun with your marketing, while promoting costumes, candy, and decorations. You can also experiment with limited-time products and UGC to drive engagement. Black Friday: Deep discounts and doorbuster deals are the order of the day here: Play up the urgency and scarcity angles, and try to create advance buzz with email teasers and VIP early access events. Cyber Monday: Ensure everything is optimized for smooth, high-traffic experiences and focus on online-only convenience, flash sales, and extended weekend deals. Christmas: Round off the year with warm, family-oriented messaging, and promote gift guides, shipping cutoff dates, and bundle deals to help people feel more at ease amidst the holiday stress. ## How to Determine If an E-commerce Trend is Right for Your Business While it might be tempting to tackle all of the latest trends, it’s important to keep in mind that some might not be the perfect fit for your business or client. Petrullo gives the example that not all businesses have the resources to do live shopping. “If it causes you too much stress and it's not something you're comfortable with, then don't submit,” she says. “There's a thousand and a half other things you can do.” Petrullo does encourage businesses to be willing to look bigger and consider under-utilized resources, though — even if it makes them uncomfortable. “If you have a TikTok shop and you can tap into the affiliate program, you can unlock a herd of sales people to promote your business and your products,” she says. “We've had clients that do $30,000 days overnight by launching a TikTok shop and having a strategic affiliate marketing plan.” ## Key Takeaways E-commerce marketing has already started to popularize live and social shopping, and offers a testing ground for emerging technology like AR and voice search. E-commerce marketing trends in 2026 include the use of sophisticated data analytics, incorporation of AI, automation wherever possible, and personalization. Understanding the different roles generations play in your sales can also help you strategically plan your promotions and the growth of your e-commerce business through social media channels. ## Frequently Asked Questions (FAQs) ### What are the most effective digital advertising channels for e-commerce in 2026? E-commerce digital advertising will continue to span a range of channels. The best channels in 2026 are a combination of paid and organic, all focused on targeting prospects and turning them into repeat customers. Successful e-commerce marketing channels also include user-generated content, live streams, performance marketing channels (e.g. PPC and online advertising) and social platforms. ### How can e-commerce businesses reduce cart abandonment? Cart abandonment remains a typical challenge for e-commerce businesses. With lots of digital tools available, businesses can use data to understand why customers abandon their carts. Teams may try streamlining the checkout process, including shipping and fee information up front, opening a guest checkout option, and testing personalized retargeting to re-engage customers. ### What are some successful examples of e-commerce digital marketing campaigns? Successful examples of e-commerce digital campaigns include the Dove Campaign for Real Beauty, which built brand and messaging alongside memorable photos, and Spotify’s Wrapped campaign, which sparks lots of copycats and conversation among users each year, and has inspired countless other brands to create their own year-in-review content. ### How is AI impacting the e-commerce marketing landscape? AI is playing a big role in e-commerce marketing, particularly in the areas of personalized experiences. Even small marketing teams can get help from AI in analyzing tons of customer behavior and predicting trends. In addition, AI can help teams automate tasks to become more efficient without adding resources. ### What are the key metrics for measuring success in e-commerce digital marketing? E-commerce digital marketing tracks similar metrics to other industries, such as customer acquisition cost, web traffic, and customer lifetime value. Metrics specific to measuring success in e-commerce include the sales conversion rate to measure how many site visitors make a purchase, the cart abandonment rate, average order value, and return rate. --- ### Cookies: Types, Use, How They Affect Privacy URL: https://www.taboola.com/marketing-hub/cookie/ Last Modified: 2025-06-22 11:52:19 It probably happens to you every day: There you are, trying to read an article, and you can’t even scroll down because a privacy notice has hijacked the page. It warns, “This site collects cookies,” prompting you to accept, deny, or accept “only necessary” cookies. What are these cookies? What do they do? Are they good or bad? What connection do they have with your privacy? Are they called biscuits in the UK? (Not this kind, no.) This article breaks it all down for you. ## What Is an Internet Cookie? An internet cookie, often just called a cookie, is a small text file that a website stores on your device (like a computer or smartphone) when you visit that site. The term was coined in 1994 by Netscape chief architect Lou Montulli: Think of it as a digital tag that helps websites remember you and your preferences. Cookies are created by the website’s server and sent to your browser, where they’re stored for future use. They’re essential for making the web more user-friendly, but can also raise privacy concerns if not managed properly — hence the prompts to accept or reject cookies. ### “Magic Cookies” The term cookie in computer-speak actually predates the advent of the World Wide Web. Pre-1994, it was called a “magic cookie.” It was used in the context of UNIX operating systems — the basis for most modern operating systems, including Linux and MacOS — to refer to a type of tag or message between users and computers, and back again. According to a Reddit AMA (“ask me anything”) session given by none other than Montulli himself: “It’s based on a fortune cookie, a message wrapped in a container. The name ‘cookies’ comes from a software trick from an old operating systems manual I read a few years earlier, a technique for passing information back and forth between the user and the system. For some reason, the small piece of data exchanged had been called a ‘magic cookie.’ Inspired by that earlier model, sketched out an architecture for a web-based ‘cookie’ that would give the medium a sense of memory without compromising privacy.” ### HTTP Cookies HTTP cookies operate through HTTP (Hypertext Transfer Protocol), the system that powers web communication. When you visit a website, the server sends a cookie to your browser via an HTTP response header called Set-Cookie. Your browser stores this cookie and sends it back to the server with every subsequent request using the Cookie header. This back-and-forth lets the server recognize you and tailor the experience, like keeping you logged in or showing relevant content. Below, you’ll see what a cookie looks like behind the scenes. This is an example of a “persistent” cookie (I’ll explain what that means in a bit) that your computer might make when you access the site for fictitious cosmetics retailer, Schlefora.com. You won’t see this at all from the front end, but the cookie script, which is saved in your browser, would look like this: Name: session_id Value: a9f3e9a8c1234d76a9f567bb0f8912df Domain: schlefora.com Path: / Expires: Wed, 25 June 2025 10:00:00 GMT Secure: True HttpOnly: True The text above is a way of identifying that particular visit you made to Schlefora; here, “session” is another term for visit. This is what each component means: - Name/Value: This pairs a label with a piece of data (like a unique session ID). - Domain: The website that set the cookie. - Path: The specific page or folder it applies to. (It’s blank here because it’s a fake site and there is no path.) - Expires: When the cookie will be deleted by the browser. - Secure: Only send the cookie over HTTPS. - HttpOnly: Makes it inaccessible to JavaScript (for security). ### Where Are Cookies Stored? Cookies are stored on your device by your web browser, typically in a dedicated folder. For example, in Google Chrome, cookies are saved within the browser’s user profile directory (e.g., ~/Library/Application Support/Google/Chrome/Default/Cookies on a Mac). Each browser has its own storage system. You can view or delete these cookies through your browser’s settings, usually under the “privacy” or “history” section. From a user point of view, this is what the drop-down menu looks like in Chrome, giving you the option of which cookies you want to delete: This is what it looks like in Safari (first, find “system settings” under the apple icon at the top left of your Safari browser): ## What Are Cookies Used For? Cookies serve a range of purposes, from improving website functionality to enabling targeted advertising. Here are some common uses: ### Session Management Cookies keep track of your activity during a single browsing session. For instance, when you log into a website for your bank account, a session cookie stores your authentication details so you don’t have to log in again on every page. These cookies typically expire when you close your browser, ensuring temporary, secure access. ### Personalization Ever notice how websites suggest products or content based on your past visits? Cookies store data about your preferences, like your location, language, or browsing history, to customize your experience. For example, an e-commerce site might show you items similar to ones you’ve viewed, making your shopping more relevant. ### Advertising Cookies are a cornerstone of online advertising. They track your behavior across sites — like which pages you visit or what you search for — to build a profile of your interests. Ad networks use this data to serve targeted ads, increasing the chances you’ll click on them. While this can make ads more relevant, it also raises privacy questions. To address these privacy concerns, advertisers are pivoting toward platforms that can leverage first-party data, rather than third-party data. ## What Data Do Cookies Collect? Cookies can collect various types of data, depending on their purpose and the website’s setup. While they don’t store large amounts of information directly, the data they hold can be powerful when combined with other tracking methods. Here’s a look at what they typically gather: ### User Identifiers Cookies often contain unique IDs that link your device to a website or ad network. These IDs don’t include your name, but act like a digital fingerprint, letting sites recognize you across visits. For example, a login cookie might store a user ID to keep you signed in. ### Browsing Behavior Cookies can track which pages you visit, how long you stay, and which links you click. This data helps websites understand user behavior and optimize content. Ad cookies might also record your searches or viewed products to tailor ads across different sites. ### Device and Settings Information Some cookies collect details about your device, like your browser type, operating system, or screen resolution. They might also store your preferences, such as language or font size, to ensure the website displays correctly and feels personalized. ## What Are the Different Types of HTTP Cookies? Not all cookies are the same — they vary in purpose, lifespan, and origin. Understanding the types of cookies helps you grasp their role and impact. Here are the main categories: ### First-Party Cookies First-party cookies are created and used by the website you’re visiting. They’re typically used for essential functions like remembering your login, storing cart items, or saving your preferences. These cookies are generally safer and less privacy-invasive since they’re limited to one domain (the site you’re on). ### Third-Party Cookies Third-party cookies are set by domains other than the one you’re visiting, often by ad networks or analytics providers. They track your activity across multiple sites to build a profile for targeted ads or analytics. For example, a third-party cookie from an ad platform might follow you from a news site to a shopping site, serving relevant ads. These cookies are more controversial due to privacy concerns. ### Session Cookies Session cookies are temporary and expire when you close your browser. They’re used for short-term tasks, like keeping you logged in during a single visit or tracking your progress through a multi-step form. They don’t store data long-term, making them less intrusive. ### Persistent Cookies Persistent cookies stay on your device for a set period (days, months, or years) or until you delete them. They’re used for long-term personalization, like remembering your login details or preferences across visits. While convenient, they can collect more data over time. ## Why Are Third-Party Cookies Being Phased Out? Over the past several years, major browsers and regulators have taken steps to phase out third-party cookies in response to growing concerns about user privacy. Third-party cookies — set by domains other than the one a user is actively visiting — have been widely used for tracking users across sites and building detailed behavioral profiles, often without their explicit consent. This cross-site tracking has drawn scrutiny from consumers and regulators alike, prompting legislation such as Europe’s GDPR and California’s CCPA. In response, browsers like Safari and Firefox began blocking third-party cookies by default, and Google Chrome, which commands over 60% of browser market share, announced in 2020 that it would phase them out entirely (although this plan changed somewhat). This shift marks a fundamental change in how digital advertising and personalization are handled, pushing the industry toward privacy-preserving alternatives such as first-party data, contextual advertising, and predictive AI models. Some performance advertising platforms combine rich first-party behavioral signals with contextual analysis to optimize ad performance without compromising user privacy. ## How Cookies Affect User Privacy Cookies are a double-edged sword: They make the web more convenient, but can compromise your privacy if misused. As websites and advertisers rely on cookies to track behavior, users face growing concerns about data collection and control. ### Tracking and Profiling Third-party cookies, in particular, enable cross-site tracking, allowing companies to build detailed profiles of your online habits. These profiles can include your interests, location, and even inferred demographics, which are often shared with advertisers. Without transparency, you might not know who’s collecting your data or how it’s used. ### Data Security Risks Cookies themselves don’t contain viruses, but they can be vulnerable to attacks like cross-site scripting (XSS), where hackers steal cookie data to impersonate you. If a website doesn’t secure its cookies properly (e.g., with HTTPS or the Secure attribute), your data could be at risk. Many users don’t realize how much data cookies collect or how to manage them: Websites often use vague cookie consent notices, making it hard to opt out of tracking. This lack of clarity can leave you feeling powerless over your personal information. ### GDPR vs. U.S. Legal regulations about cookies differ by region. The GDPR (used in the EU) requires websites to obtain explicit, informed consent before placing any non-essential cookies on a user’s device. In contrast, U.S. laws like the CCPA allow websites to use cookies by default and require only that users be given the option to opt out. The GDPR emphasizes prior user control, while U.S. rules generally focus on transparency and the right to refuse. Here is an example of a cookie alert that a user might see if logging in to a website from a U.S.-based computer or phone. This is taken from the atlantic.com news site. Notice that the default is that if you “accept,” you are agreeing to accept third-party cookies. You have to take an additional step to opt out. This would not be considered GDPR-compliant. By contrast, here is a cookie alert you might see if logging into a website from an EU-based device. This is taken from the EU parliament official site. Notice that you get the choice to refuse or accept up front, without any additional steps. This is what makes it GDPR-compliant. ## How to Enable and Remove Cookies Managing cookies gives you control over your online experience and privacy. Most browsers let you enable, disable, or delete cookies through their settings, and you can also use tools to fine-tune how cookies work. ### Enabling Cookies To enable cookies, go to your browser’s settings (usually under “Privacy” or “Security”). For example, in Chrome, navigate to Settings > Privacy and security > Cookies and other site data and select “Allow all cookies” or “Block third-party cookies.” Enabling cookies is often necessary for websites to function properly, like for logins or shopping carts. ### Deleting Cookies You can delete cookies to clear stored data and start fresh. In most browsers, find the option under Settings > Privacy > Clear browsing data. Select “Cookies and other site data” and choose a time range (e.g., last hour, or all time). Be aware that deleting cookies will log you out of sites and reset preferences. ### Using Browser Extensions For more control, consider extensions like uBlock Origin or Privacy Badger, which block trackers and unwanted cookies. You can also use “Incognito” or “Private” browsing modes, which prevent cookies from being stored after your session ends, though they don’t block cookies entirely. ## Key Takeaways Cookies are small text files that websites use to remember your preferences, manage sessions, and enable features like personalization and advertising. They collect data like user IDs, browsing behavior, and device details, which can enhance your experience but also raise privacy concerns. Types of cookies include first-party (site-specific, safer) and third-party (cross-site, used for ads), as well as session (temporary) and persistent (long-term). Cookies can track your online activity, sometimes without clear consent, posing privacy and security risks if not managed. You can enable or remove cookies via browser settings or use extensions for more control, balancing functionality with privacy. Third-party cookies are being phased out due to privacy concerns. ## Frequently Asked Questions (FAQs) ### What’s the difference between first-party and third-party cookies? First-party cookies are set by the website you’re visiting and only work for that site, handling tasks like logins or preferences. Third-party cookies come from external domains (e.g., ad networks) and track you across multiple sites for ads or analytics. First-party cookies are less privacy-invasive, while third-party cookies are often restricted by modern browsers due to tracking concerns. ### What is a session cookie vs. a persistent cookie? A session cookie is temporary, stored only until you close your browser, and used for short-term tasks like keeping you logged in during a visit. A persistent cookie remains on your device for a set period, enabling long-term features like remembering your login or preferences across visits. ### How do I delete cookies? To delete cookies, go to your browser’s settings (e.g., Settings > Privacy > Clear browsing data in Chrome). Select “Cookies and other site data,” choose a time range, and confirm. This will log you out of sites and clear stored preferences. You can also delete specific cookies via developer tools. ### How will marketers track users without cookies? With third-party cookies being largely phased out, marketers are shifting to alternatives like first-party data (collected directly from users), contextual advertising (based on page content), and privacy-preserving technologies like Google’s Privacy Sandbox. These methods aim to balance targeting with user privacy. ### Can cookies be used for analytics tracking? Yes, cookies are widely used for analytics, tracking metrics like page views, time spent, and user journeys. Tools like Google Analytics use cookies to identify returning users and measure site performance, helping website owners optimize content and user experience. ### How do you make GA4 cookie-compliant? To make Google Analytics 4 (GA4) cookie-compliant, implement a cookie consent management platform (CMP) to get user consent before setting analytics cookies. Use GA4’s consent mode to adjust tracking based on user preferences (e.g., disabling cookies if consent is denied). Ensure your privacy policy discloses cookie use and complies with regulations like GDPR or CCPA. ### How are cookies different from app tracking? Cookies are browser-based text files that track web activity, while app tracking uses device identifiers (e.g., IDFA on iOS) or SDKs to monitor in-app behavior. Cookies are limited to web browsers, while app tracking spans mobile apps and can collect more device-specific data. Both raise privacy concerns but operate in different ecosystems. ### How do I implement a secure cookie? To implement a secure cookie, use the following attributes in the Set-Cookie header: - Secure: Ensures the cookie is only sent over HTTPS. - HttpOnly: Prevents access via JavaScript, reducing XSS risks. - SameSite=Strict or SameSite=Lax: Limits cross-site requests to prevent CSRF attacks. - Set a short expiration time for sensitive cookies and encrypt their contents if storing sensitive data. Always test cookies on a secure server. --- ### Target Audience: The First Step in Defining Your ICP URL: https://www.taboola.com/marketing-hub/target-audience/ Last Modified: 2025-06-22 11:22:34 You want to strategically and carefully spend your marketing budget, and doing so starts with understanding your target audience. Broad marketing messages designed to appeal to “everyone” are rarely effective, so defining your target audience will allow you to customize and refine your marketing message to appeal to those individuals most likely to convert into paying customers. ## Defining and Understanding Target Audience Your target audience is a group of people that you want to reach with your marketing, so they will buy your products or services. A target audience shares certain characteristics, like their demographics and interests, that make them a good fit for your business. Once you determine your target audience, you can better focus your marketing efforts to reach these individuals. ### What Are the Key Characteristics Used to Define a Target Audience? Most companies use demographics as a key characteristic to define a target audience. Demographics, or statistical data, might include characteristics like age, location, gender, occupation, income level, and relationship status. Psychographics, including behaviors, interests, and thought processes, are also key characteristics: You might, e.g., define your target audience based on their interests, pain points, values, and spending behavior. For a practical example, picture a business selling running shoes for men. This brand might define their target audience as being male athletes between ages 18 and 40, with an income level from $50,000 to $120,000, who may participate in competitive running and want an innovative shoe designed to increase foot turnover, and who may be willing to spend more on shoes for a competitive advantage. ## Identifying Your Target Audience Identifying your target audience helps you to better refine and focus your marketing efforts. You can also generate higher quality leads, which can help improve conversions. Identifying your target audience is a process, and you’ll need to refine your target audience as your business and its offerings evolve. ### What Are the Initial Steps in Defining Your Target Audience? When you’re first working to define your target audience, it’s helpful to use market research to better understand who your customers are. You can poll customers and ask about factors like their pain points, demographics, and why they chose the products or services they did. Reviews of your business can also reveal information about who your buyers are, why they made a purchase, and what pain points your products are solving. Looking at competitors’ target audiences is helpful, too. Observe their marketing, including their messaging, and consider who they’re trying to reach and what pain points they’re addressing. Take the information that you’ve collected and make a list of common characteristics your audience shares. You can continue refining your target audience as you start to better understand your customers. ### How Can Market Research Help in Understanding Your Ideal Customer? Market research is a valuable tool in understanding your ideal customer, and you can perform market research in several ways. The fastest and simplest way to do market research into your target audience is to review existing sources, such as industry publications (which publish information on common trends), demographics, and household incomes within your target industry. You can also turn to consumers themselves, though this method can be expensive and time-consuming. If you already have a robust social media following or customer base, you can poll your customers and followers to learn more about them. Asking questions about their pain points, demographics, and what they like or would change about your products and services can give you a deeper, more detailed understanding of your audience. In addition to surveys and polls, focus groups and interviews can also help you collect detailed information. ### What Are Buyer Personas and How Do They Help in Visualizing Your Target Audience? Buyer personas are profiles that help you visualize your ideal customer. You can create buyer personas based on your target audience’s common characteristics. For example, let’s say that your clothing brand’s target audience is women aged 30 through 40 who want stylish work clothes that are also highly comfortable. They value products made in the USA and want to avoid harmful chemicals and undesirable manufacturing practices. Using those characteristics, you could create a buyer persona of Sarah, a 35-year-old communications professional who needs a suitable wardrobe that can keep her comfortable during long workdays. She makes $80,000 per year, often uses Instagram and LinkedIn, and lives outside of Chicago. You can add other details, like the types of publications Sarah reads, the movies and music she enjoys, and what motivates her to make a purchase. This persona can help you better envision a very specific individual within your target audience, so you can more easily tailor your messaging and marketing for maximum effectiveness. ## Key Characteristics to Consider As you define your target audience, you’ll want to include several types of key characteristics to paint a complete picture of your potential and ideal customers. ### What Demographic Factors Are Important to Consider? Demographic factors are easily definable characteristics that you can use to identify your target audience. Some key demographic characteristics include: - Age. - Gender. - Location. - Income. - Education. - Marital status. - Gender identity. ### What Psychographic Factors Influence Your Audience? Psychographic factors are also important. These factors help define how your target audience thinks, like: - Values. - Beliefs. - Interests. - Lifestyle. - Attitudes. - Morals. - Expectations. - Pain points. ### What Behavioral Factors Are Relevant? Your audience’s behavioral factors are essential, since they can affect how your audience decides to make purchases, how they engage with your business, and more. Include factors such as: - Purchase history. - Interactions with your brand. - Website history. - Purchase intention. ### How Do Needs and Pain Points Define Your Target Audience? If your product or service addresses the audience’s pain points, they’re more likely to be motivated to make a purchase from your business. If your product or service doesn’t address those pain points, though, you’re not helping them to solve a problem, so it will be more difficult to motivate them to buy. Chances are, if your offerings don’t align with their needs and pain points, you’re looking at the wrong target audience. ## Reaching Your Target Audience ### How Does Understanding Your Target Audience Inform Your Choice of Advertising Channels? Defining and understanding your target audience gives you valuable information that you can use to strategically choose your advertising channels. Different demographics tend to use different advertising channels, such as social media marketing, influencer marketing, radio, direct mail, and email marketing. For example, according to the Pew Research Center, 93% of U.S. adults ages 18 through 29 use YouTube, but just 38% of that same demographic uses X. When you know the specific demographic you’re trying to reach, you can focus on the advertising channels they use most often, so your marketing is more likely to be successful. Understanding your target audience can also help you determine your timing and frequency of communication. A younger demographic like adults ages 18 through 29 are used to seeing frequent marketing messaging, so you might increase your frequency to help your messaging stand out from all of the other ads. But, if you’re marketing to seniors, they could see such frequent messaging as being overwhelming and pushy. You can adjust your frequency based on what you know about your target audience and how they’re likely to receive your messaging. ### How Can You Use Audience Insights for Personalization? Having detailed audience insights is also helpful when personalizing your marketing messaging. Once you’ve identified your target audience’s interests, pain points, and buying motivation, you can craft more personalized marketing messages that address and reflect those characteristics. Such personalized marketing can make a strong impression on your audience, increasing its effectiveness and prompting your audience to take action, whether that’s following your brand on social media or learning more about a product by visiting your website. ## Evolving Your Understanding of Your Target Audience Your job isn’t done once you’ve defined your target audience — it’s important to continuously develop and evolve your understanding of them. ### Why Is It Important to Continuously Refine Your Understanding of Your Target Audience? Consumer behaviors and market dynamics continuously change, so your target audience’s interests, buying motivations, and pain points may have dramatically changed from what they were a year ago. If you release new products or services, they may appeal to a slightly or even entirely different target audience. Rebranding can also mean it’s time to change your target audience. Continuously refining your understanding of your target audience helps ensure that you’re truly marketing to the right potential customers. View your target audience as a continuously evolving group of people to keep your strategy accurate and effective. ### How Can You Use Data and Analytics to Track Changes in Your Audience? Analytics and data inform your initial understanding of your target audience, and they’re also useful in tracking changes in your audience. Your social media analytics can help you spot changes and new trends in who is following your pages, which can be an important source of demographic information. Web analytics tools, like Google Analytics, can provide you with information on website visitor demographics, brand interactions, and purchase behaviors. You can also use AI to analyze large amounts of data, and this technology may be able to spot trends or changes that you haven’t yet identified. Don’t forget that your current audience of social media followers and customers is also a valuable resource. You can gather feedback from your audience through polls, focus groups, and surveys, which can help you improve your products and services, your marketing messaging, and your understanding of what matters most to your audience. ## Key Takeaways Understanding your target audience is essential for effectively marketing your business. Many tools can help you determine the key characteristics that define your target audience, and you can then use that information to craft effective messaging on the channels they use. Since your target audience can evolve over time, it’s important to continuously expand your understanding of them to maximize the effectiveness of your marketing. ## Frequently Asked Questions (FAQs) ### Can a business have more than one target audience? Yes, a business can have multiple target audiences. That’s particularly true for businesses that offer a range of products and services which may appeal to different audiences. ### What are some common mistakes in defining a target audience? Many businesses make their target audience too broad and try to appeal to nearly everyone. Doing so can cause your marketing to miss the mark, since it’s not personalized to a particular demographic. Other common errors include not performing enough research and not refining and evolving the target audience over time. ### How does a target audience differ from a marketing persona? Your target audience is the large, broad group of people you want to reach with your marketing; hopefully these people will buy your products or services. A marketing persona is a highly detailed profile of a hypothetical person that could be in your target audience. The marketing persona helps you to better visualize an individual in your target audience, which can make it easier to understand whom you’re marketing to and how to refine your marketing messages. ### What tools can help me identify and analyze my target audience? There are many tools to help identify and analyze your target audience. Social media analytics can provide valuable information on your follower demographics and interests, while your own website analytics can help you better understand website visitor demographics, interactions, and buying behavior. --- ### Ad Formats: How to Choose the Right One? URL: https://www.taboola.com/marketing-hub/ad-format/ Last Modified: 2025-06-30 08:58:18 Choosing the correct ad formats — i.e., the way in which your ad is presented to your target audience — is an important part of any advertising campaign. It’s vital to choose the right format for your product and your consumer, but how do you know which one will work best for your goals? ## What Is an Ad Format? An ad format refers to the way in which an advertisement is presented. This can mean videos, display ads, or other ways of reaching consumers. Advertisers can choose the type they think will help them most effectively tell their story. ## Importance of Ad Formats As with any kind of storytelling or communication, the appropriate presentation makes a world of difference. Different ad formats will work better for different products or platforms, based on what people want to know about a product, or why they come to a particular platform. ## Types of Ad Formats These are some of the most common ad formats: ### Video Ads Video ads appear on various platforms, e.g., social media and YouTube. Assuming a viewer has the time and is willing to watch the video, these are incredibly effective ways to engage and bring the consumer into your world. ### Display Ads Display ads are your classic banner ads, skyscraper ads, and the like. They can be dynamic or static, and you’ll see them on all kinds of apps and websites. ### Sticky Ads These are similar in appearance to display ads, but they move with the reader down the page, rather than being visible in only one place. ### Native Ads Native ads, as the name implies, appear to be part of the content the user is consuming, matching the already existing content in tone and topic. (They will, however, be labeled as “sponsored post” or similar.) ### Search Ads These sponsored posts live at the top of search engine results pages and target users who are searching topics directly related to your product or service. ## Platform-Specific Ad Formats If you think of a platform as a room where you’re giving a talk, then choosing the right format is about reading the room before speaking. “Each platform has its own rhythm, and if you’re not in tune with that, then your ad gets ignored,” says Taron Flood, growth marketer and newsletter operator. ### Meta Sometimes the smartest move is to make an ad feel like part of the conversation. “On Meta, the best ads look like they belong,” says Flood. “Think lo-fi, in-feed creative that doesn’t scream ‘ad.’ Carousels still work for driving clicks, but short-form videos win attention, especially with Reels.” ### Google Ads An effective Google Ad is all about intent, says Flood: “Search is where people already know what they want, so your job is basically to just not to get in their way. Clarity and straightforwardness beat clever every time.” ### YouTube When you’re asking viewers to stay for a bit, the ad should be instantly engaging. “YouTube rewards narrative, but only if you manage to earn those first five seconds of attention,” says Flood. “Even skippable ads need to hook hard and early.” ### LinkedIn Advertisers on LinkedIn should aim to teach, not pitch, says Flood: “Since LinkedIn is a platform where people tend to flex their expertise, sponsored content that isn't overly sales-y, as well as insight-driven carousels, are probably a marketer's best bet there.” ### TikTok In-feed video ads are ideal on TikTok, says Flood. “Native advertising is really king there, the kind of advertising that flows seamlessly in users' For You feed,” he says. “The rule of thumb on TikTok is to make your ad feel as little like an ad as possible.” ### Realize This performance advertising platform is ideal for advertisers who care about thoughtful, well-crafted campaigns, says Flood: “Realize allows for advertisers to automatically deliver personalized creative that matches the right headlines and visuals to the right audiences.” ## Which Ad Formats Are Best for Driving Engagement? There’s a difference between flashing an ad in front of a viewer and actually gaining their interest and attention. “Short-form video almost always wins, especially when it’s native to the platform and gets to the point fast,” says Flood. However, he notes, carousels and polls deserve more credit, and their value is increasingly being noted. “They invite interaction, not just consumption. And everyone knows that that extra second of engagement is everything.” ## Creative Formats Specs “If you ignore the specs, you’re ignoring how people actually use the platform,” says Flood, who adds that specs are more like behavioral cues than tech details. “A 9:16 video with sound on hits completely different than a muted 1:1 scroll.” Every platform will have its own preferred image size. Whether you’re looking to make a carousel ad for Facebook, video ads for LinkedIn, or native ads for Google, you’ll want to learn about the specifics for each platform. These details are in accordance with each platform’s rules, and they’ll help your ads look their best within the surrounding content, which benefits you as a marketer. ### Image Sizes per Type Platform Ad Type / Ratio Dimensions (Pixels) Notes Meta Facebook Feed (1:1) 1440 x 1440 Facebook Feed (4:5) 1440 x 1800 Facebook Carousels 1080 x 1080 (min) FB Right Column, MP 1200 x 1200 Instagram Feed 1080 x 1080 Instagram Stories, Reels 1080 x 1920 Audience Network 398 x 208 (min), 1200 x 628 (rec) Native, Banner, Interstitial Google Images Horizontal 1200 x 628 (min 600 x 314) Vertical 960 x 1200 (min 480 x 600) Square 1200 x 1200 (min 300 x 300) YouTube Horizontal 1920 x 1080 Vertical 1080 x 1920 Square 1080 x 1080 LinkedIn Horizontal 640 x 360 (min) - 7680 x 4320 (max) Vertical 360 x 640 (min) - 2430 x 4320 (max) Square 360 x 360 (min) - 4320 x 4320 (max) TikTok In-Feed Ads 540 x 960 (min), 720 x 1280 (rec) File size up to 500 MB Realize 1200 x 674 (min 400 x 350) Smaller images will serve in fewer placements ### Headline Character Limit per Platform Platform Ad Type Headline Length (Characters) Notes Meta Facebook Feed 27 (recommended) Instagram 40 (recommended) Google Text Ads 30 (max) YouTube In-Feed Video Ads 100 (max) Text longer than 25 characters may be shortened on some devices. LinkedIn ~70 (avoid truncation), 200 (max) Use around 70 characters to avoid truncation on most devices. TikTok 20 (max) Realize 35-45 (recommended), 60 (max) ### Aspect Ratios per Platform Platform Ad Type / Orientation Aspect Ratio Meta Facebook Feed 4:5 Facebook Right Column 1:1 Facebook Stories 9:16 - 1.91:1 Instagram Stories 9:16 Google Images Horizontal 1.91:1 Vertical 9:16 Square 1:1 YouTube Horizontal 16:9 Vertical 9:16 Square 1:1 LinkedIn Horizontal 1.91:1 Vertical 4:5 Square 1:1 TikTok Horizontal 16:9 Vertical 9:16 Square 1:1 Realize 16:9, 4:3, or 1:1 ## Some Ad Format Selection Strategies When you’re considering an ad format, the first thing is to think about what you want to achieve. “Always start with a goal, not the format,” says Flood. “If you want awareness, go big with video. For leads, try low-friction formats like Meta’s instant forms.” ### For Retargeting While there is no specific ad format for retargeting, you’ll want to use whichever format will connect with your viewer and meet them where they are. If you’re a newer brand, prioritizing formats that build familiarity is key. If you’re more established, test sequential storytelling or deeper engagement formats like quizzes or carousels. “Sequential storytelling for established brands is my favorite because when it’s done well, it can be an engagement goldmine,” says Flood. ### For Funnel Stages Consumers bring different attitudes and levels of interest at each stage of the marketing funnel. There will be some overlap, with the main difference being that the ads should get more detailed and specific as they move further down the funnel. At the awareness stage, visually engaging formats like display ads, short videos, and educational native ads can be most inviting. During the consideration stage, viewers are more open to blog posts, webinars, and interactive ads where they can spend more time learning about the product. In the conversion stage, advertisers can deliver content that will tip the consumer into buying, such as free trials, limited-time offers, and retargeting ads. ### For Industries Whether your ad is for SaaS, fashion houses, or educational organizations, you’ll want to use the format that helps you tell your story as clearly and effectively as possible on any given site. These can be text-based, video-based, or image-based. Each format can work for each type of industry, but it’s easy to see how video-based and image-based ads will have an edge for a visual industry like fashion, whereas text-based and video-based ads can work better for Saas and educational organizations, where the consumers will want to know all the finer details. ## Format Testing and Strategies Testing formats side-by-side under the same objective is a good strategy. “For instance, on Meta, you can run carousels, video, and static in one ad set and let performance guide your budget allocation,” says Flood. “It's always important to keep an eye on what’s performing on your organic channels. That’s usually where some of our best ideas live.” ## Key Takeaways Ad formats are the ways ads are presented to your viewer. Different platforms offer different specs and guidelines for advertising on their sites, and it’s important to stay aware of the details. It’s also key to test how your ads perform with different ad formats so you can be efficient and effective in your campaigns. ## Frequently Asked Questions (FAQs) ### What is the most popular ad format? “Single image and video ads still lead in volume, and on Meta especially. They're easy to produce and plug-and-play. But, what’s popular isn’t always what’s effective,” says Flood. ### Should I test different ad formats in the same campaign? You should test different ad formats in the same campaign, but don’t just focus on format. “Message testing is just as key as testing the layout,” says Flood. “Gains come from testing how quickly you get to the point, how you frame the hook, and whether your CTA actually lands.” ### What kind of content works best in a slideshow or collection ad? For a slideshow or collection ad, you’ll want to focus on high-quality images and charts. If you do use text, it should be short, to-the-point, and it should visually pop. ### What’s the best format for Google Display ads? While there’s no official best format, responsive display ads can help brands optimize their campaigns, achieve a broader reach, and save time. Google might use asset enhancement and AI-generated assets, and Google’s AI can generate various ad combinations. ### What is a display ad format? A display ad format means a video, photo, illustration, or GIF where the display is meant to catch the viewer’s eye. The format can also be understood as the combination of size and aspect ratio. ### What format should I use for LinkedIn lead-gen ads? LinkedIn offers a variety of formats for lead-gen ads, including carousel, video, and single image ads. The platform provides pre-filled lead-generating forms. ### What ad format should a SaaS company use on LinkedIn? Sponsored content of all kinds can be effective, as people come to LinkedIn to learn more about their industries. Blog posts, videos, and thought leadership posts can speak to the LinkedIn visitor and engage them while they’re already in a learning stage. --- ### Paid Media: Exploring the Core of Online Ads URL: https://www.taboola.com/marketing-hub/paid-media/ Last Modified: 2025-07-09 06:43:19 Finding your way in the world of digital marketing is basically like learning a new language. The terms, acronyms, and concepts are tough enough to grasp, not to mention that they shift and evolve all the time, too. But, one of the foundational concepts you need to know, and maybe the one that holds the most immediate power to drive results, is paid media. As a copywriter for many years, I’ve done my share of creative work for paid media, and I’ve also seen it from the other side, too, when advertising my own services. So, let's get into it: What exactly paid media is, why it's so crucial, and how you can start using it to your advantage. ## Understanding Paid Media Paid media generally refers to any marketing channel or tactic that you pay for to promote your content, product, or service. It's advertising in its purest digital form, where you exchange money for exposure and attention. That may sound pretty basic, but paid media is important, as it offers immediacy and scalability, allowing you to reach a large audience quickly. It's also more about actively seeking your audience out, rather than waiting for them to find you organically. ### How Does Paid Media Differ From Owned and Earned Media? Owned media is any channel you fully control, like your website, blog, or social media profiles. While it offers full control and cost efficiency, the tough part is that building an audience takes lots of time. Earned media, meanwhile, is when others talk about you without direct payment — think media mentions, reviews, or even social media shares. It can be really valuable for credibility, but it’s also the least controllable or predictable. That’s where paid media comes in. It fuels owned media by driving traffic, and amplifies earned media by broadening its reach, giving you direct access to your target audience and bypassing the slow build of owned media and the unpredictable nature of earned media. In my experience, having a well-rounded strategy integrates all three, and all of them mutually benefit each other. ### What Is the Primary Goal of Utilizing Paid Media Channels? In a nutshell: Driving specific, measurable actions. While brand awareness is undoubtedly a benefit, paid media really focuses on performance and conversions, whether that's leads, sales, app downloads, or website traffic. But, unlike traditional advertising, digital paid media (when used correctly) offers some truly incredible precision. You can track every click, impression, and conversion, allowing for continuous optimization, and ensuring that your ad dollars are working as hard as possible. Paid media is a crucial component of many marketing strategies thanks to a combination of speed, scalability, and precise control that other types of marketing just can’t match. You also get back deep, actionable data on performance, which gives you the info needed for continuous optimization and ensuring you’re spending your advertising budget efficiently. ### What Are the Key Advantages and Disadvantages of Using Paid Media? Nothing in marketing is going to be a magic bullet, as you’ll quickly find out, and paid media is no exception. Still, understanding its pros and cons are vital before you go all-in. Pros Cons Speed & Scale Cost Precise Targeting Ad Fatigue Measurable & Controllable Platform Dependence Predictable Results Advantages - Speed and scale: Paid media can get near-instant results and reach large audiences fast, which is especially ideal when you’re running time-sensitive promos. - Targeting precision: Better targeting (like demographics, interests, behaviors), leads to less wasted ad spend. - Measurability and control: Extensive data for real-time tracking, testing, and optimization brings a better and clearer understanding of ROI. - Predictability: Once optimized, campaigns can offer fairly predictable results, allowing for consistent scaling. #### Disadvantages - Cost: Requires budget and spending which can get expensive quickly, particularly in competitive markets, if it’s not carefully managed. - Ad fatigue: Audiences can get burnt out on seeing repetitive ads, requiring constant creative refreshes. - Dependence on platforms: Performance can be impacted by changes in a platform’s algorithms or policies. ## Types of Paid Media Channels The digital advertising landscape is large, but these can be classified into a few main categories, each with unique strengths. ### What Are the Different Types of Paid Search Advertising? Paid search advertising focuses on appearing at the top of search engine results when users search for specific keywords. There are different types and subsets which I’ll explore more later on, but a couple of common types include: - Text ads: These are the standard text-based ads in search results. They’re not flashy, but they’re highly effective. - Shopping ads (product listing ads): These ones feature images with product details, and are excellent for e-commerce sales. Strategy plays a big part, too, but paid media’s power lies in its intent-based targeting, when users are actively looking for something. ### What Are the Various Forms of Paid Social Media Advertising? Paid social media advertising means running ads on social platforms like Facebook, Instagram, LinkedIn, and TikTok, since they offer rich targeting based on demographics, interests, and behaviors. Common forms include: - Image and video ads: Visually engaging formats for brand awareness and direct response. - Carousel ads: These showcase multiple images or videos and are great for product features. - Lead ads: In-platform forms for easy lead generation. Social media advertising is a top choice for reaching people where they spend significant online time, and it’s also ideal for building brand loyalty and sparking discovery. ### What Are Display Ads and How Do They Work? Display ads are visual advertisements on websites and apps. They come in various sizes and formats and work by placing your ad on sites within display ad networks or via ad exchanges. Targeting can be broad or specific and is based on things like demographics and interests, but they’re great for brand awareness, driving traffic, and remarketing. From my years in this business, I've always found them to be an underappreciated powerhouse for getting your brand seen. ### What Is Programmatic Advertising and How Does It Fit Into Paid Media? Programmatic advertising is the automated buying and selling of ad inventory using algorithms and real-time bidding. The automated part is usually what sounds scary and off-putting to newcomers, since you don’t have direct control, but it’s an efficiently managed auction for ad space, and includes display, video, and native ads. Programmatic advertising leverages all sorts of data to optimize ad placements across lots of sites and apps, providing you with a scalable and efficient way to buy digital ads. ### What Are Other Forms Of Paid Media, Such as Influencer Marketing and Native Advertising? Beyond the major platforms and traditional formats, other paid media forms can be: - Influencer marketing: Paying people with significant followings to promote your offering. The value comes from the influencer’s credibility with their audience, something you can’t replicate in an ad by itself. - Native advertising: These are ads designed to blend seamlessly with surrounding content, appearing as sponsored articles or recommended links. They’re much less intrusive and more engaging to users. ## Planning and Executing Paid Media Campaigns Launching a successful paid media campaign requires careful planning, not just throwing money at ads. Here’s what to know: ### What Are the Essential Steps Involved in Planning a Paid Media Campaign? Some of the key planning steps include: - Clearly define your goals: What do you want to achieve? Is it awareness? Leads? Sales? - Understand your audience: Who are you targeting? Know their demographics, interests, and behaviors. - Do your research: Learn about keywords, audience, and competitive analysis. - Allocate your budget: How much will you spend? And more importantly, where? - Select your channels: Choose platforms based on goals, audience, and budget. - Develop your creative: Design compelling ad creative (images, videos, copy). - Set up your campaign: Configure targeting, bidding, and scheduling. - Set up tracking and measurement: Ensure proper tracking (pixels, conversion tags). ### How Do You Define Your Target Audience and Campaign Objectives for Paid Media? Defining your target audience is crucial — it’s the heart of your campaign, so be specific beyond "everyone." Create buyer personas covering demographics, psychographics, and behaviors. In terms of selecting the most appropriate media channel for your goals, consider the following: If you’re targeting immediate sales, paid search is strong; for brand awareness, social media or display ads may be the way; for content-driven conversions, native advertising is your best bet. For paid media budget setting, consider your cost per acquisition (CPA) goals, customer lifetime value (CLTV), and competition. Start comfortably and adjust based on performance — it’s a continuous calibration. Developing effective ad creative and messaging means tailoring content to each platform and audience mindset. ## Measuring and Analyzing Paid Media Performance Without proper measurement, your paid media efforts are just a shot in the dark. Track everything and learn from what you find. ### What Are the Key Metrics to Track for Paid Media Campaigns? Crucial metrics include: - Impressions: Ad display count (reach). - Clicks: User interactions with your ad. - Click-Through Rate (CTR): Clicks divided by impressions (ad engagement). - Conversions: Desired actions taken (purchases, leads). - Cost Per Click (CPC): Average cost per click. - Cost Per Acquisition (CPA) / Cost Per Lead (CPL): Average cost to acquire a conversion/lead (profitability). - Return on Ad Spend (ROAS): Revenue from ads divided by cost. ### How Do You Measure the ROI of Your Paid Media Investments? Measuring ROI compares campaign revenue/value against costs. Here’s how to do the math: ROI = (Revenue from Paid Media - Cost of Paid Media) ÷ Cost of Paid Media x 100 For e-commerce, it’s pretty direct. With lead gen, assign a monetary value to leads based on historical conversion rates and customer lifetime value. This data is what’s important for true profitability. ### What Tools and Platforms Can Be Used for Paid Media Analytics? Most major paid media platforms offer built-in analytics. Beyond those, you can leverage: - Google Analytics: For understanding user behavior on your website, post-click. - Conversion Tracking Pixels: Code on your site to track specific user actions from ads. Interpreting data for optimization is something that’s always ongoing. Keep your eyes open for identifying trends, pause underperforming ads, adjust bids, and always be refining targeting. A/B test creatives and landing pages, too: The goal here is continuous improvement. ## Best Practices for Paid Media ### How Do You Ensure Your Paid Media Campaigns Are Cost-Effective? This is an important one, and there’s a bunch of different approaches you can take toward cost-effectiveness, such as: - Targeting precision: Accurate targeting minimizes wasted ad spend. - Relevant ad creative: Ads need to resonate with people to drive clicks and conversions. - Optimized landing pages: Clear, concise, conversion-focused landing pages maximize ad spend. - Continuous optimization: Regularly review data, adjust bids, and refine targeting. - Leverage negative keywords (for search): Exclude irrelevant terms to avoid wasted clicks. ### What Are the Best Practices for Targeting and Segmentation in Paid Media? Since targeting and segmentation are foundational cores of a paid ad campaign, here’s what’s recommended: - Start broad, then refine: Begin wider to gather data, then narrow based on converters. - Leverage custom audiences: Use existing customer data for remarketing or similar audiences. - Utilize behavioral and interest targeting: Go beyond basic demographics to target based on online interests and behaviors. - Test and iterate: Continuously test targeting parameters for best results. You should review and optimize your paid media campaigns frequently (daily for high-volume, weekly/bi-weekly for smaller). Consistency and data responsiveness are key. Be aware of some common mistakes, too, which include not setting clear objectives, poor targeting, ignoring tracking, bad landing pages, and "set it and forget it" approaches. Finally, never underestimate a compelling headline — I've seen too many otherwise good campaigns tank because of bland copy. ### How to Leverage Data for Personalization in Your Paid Media Efforts Personalization is increasingly vital in this digital landscape, since data is what allows you to tailor ads to specific audience segments. Here are a few ways to go about it: - Dynamic creative optimization: Serve different ad variations based on user data. - Remarketing/retargeting: Show ads to past website visitors, utilizing their familiarity for conversion-focused messages. - Audience segmentation: Create unique ad messages and offers for smaller, specific groups. ## Key Takeaways Paid media is a helpful and necessary tool for digital marketers right now, offering a reach and level of precision in targeting that’s hard to match. When you have a better understanding of channels, meticulous planning, and continuous optimization, you can harness its power to achieve your marketing goals. It’s an ongoing process of learning, testing, and refining. ## Frequently Asked Questions (FAQs) ### What is a good starting budget for paid media? There’s no one-size-fits-all answer, as a "good" starting budget for paid media depends on several factors that are unique to you and your business. Consider your industry's competitiveness: Highly competitive sectors like finance or e-commerce for popular products will naturally have higher advertising costs due to more advertisers bidding for attention. Your target audience also plays a role, as niche audiences might be less expensive to reach than broad ones. Most importantly, your specific marketing goals will help dictate your budget. A common approach for beginners is to start small, maybe a few hundred dollars a month. That’s what I did when advertising my copywriting services and music composition. This allows you to learn how the platforms work, understand your audience's response, and identify what creative and targeting strategies perform best — all without breaking the bank. Once you start seeing positive results (and a clear return on your initial investment) you can incrementally increase your budget. It’s a process of learning, optimizing, and scaling based on data, rather than guessing. ### How do you know which paid media channels to invest in? It’s a strategic decision, and one that should be driven by your audience and your objectives, not by what’s popular at the moment. The way to start is by understanding where your target audience spends their time online. Are they actively searching for solutions on Google, or more likely to discover new products while scrolling through social media? Consuming content on news sites and blogs? Do the research and find out. Then, align the channel's strengths with your campaign goals. If immediate conversions are your priority and your product or service solves an urgent need, paid search (like Google Ads) is a strong choice because it captures high intent. But, if you’re aiming for building brand awareness, or targeting consumers with more visually-based products, social media platforms like Instagram or TikTok are likely to be a better choice. ### Can paid media help with brand building? It sure can! Paid media is a powerful tool when it comes to brand building, and it’s a misconception to think that it’s only good for things like direct response or immediate sales. Plenty of paid media campaigns are designed for performance (like getting clicks, leads, and sales), but the consistent exposure and controlled messaging that it offers are really what’s invaluable for increasing brand awareness, recognition, and recall. When your ads appear consistently across various platforms and relevant websites, they help reinforce your brand identity, clearly communicate your values, and make your brand more familiar to a wider audience at a glance. Methods like display advertising are a proven way to win if used correctly, along with video ads on social media, or even certain native ad placements. These types of formats are perfect for showcasing your brand's personality to the right audience, telling your story, and creating an emotional connection with potential customers. The more you target specific demographics and interests, the more you can ensure your brand message reaches the right people, and builds a strong foundation of awareness before they’re even ready to make a purchase. It’s this consistent brand exposure which can lead to increased trust, loyalty, and ultimately, a stronger market position. Essentially, it’s all about planting seeds for future growth, and you can start doing that right now. ### How does paid media work with SEO? Paid media and SEO work together to help your website get seen online. With paid media, you have instant visibility by putting your ads at the top, which is great for quick results or testing new keywords. SEO builds a more long-term visibility by making your website naturally appealing to search engines, which attracts free traffic over time. They can share information, like which keywords perform best, so both strategies become even stronger. Using both helps your brand show up more often in search results, giving you a better chance to connect with people looking for what you offer. --- ### Third-Party Data: How It Works, Why It Matters URL: https://www.taboola.com/marketing-hub/third-party-data/ Last Modified: 2026-03-09 08:00:47 As an advertising copywriter, I know firsthand how understanding data helps create strong ad campaigns. Launching into the creative side of things is my first instinct when starting a new project, but taking some time to dive deep into the data often provides the missing insights I need. That goes for the client side of things, too. Even if you’re the advertiser, understanding different types of data is key to making your campaigns work. Third-party data has been a major part of how online ads are delivered and shown to people for years, and it’s this information, collected and shared by companies separate from the advertiser or website owner, that’s helped brands reach people they haven’t interacted with directly. Third-party data is also a bit of a third-rail topic, and we’ll get into that later. But, it’s allowed advertisers to take their reach further, expanding to new markets and finding new opportunities. Let’s take a closer look at its uses, benefits, challenges, and speculate what its future might be, too. ## What Is Third-Party Data? Third-party data refers to information collected by a company that doesn’t have a direct relationship with the person whose data is gathered. This data is then sold or shared, mainly for advertising purposes. Unlike information you collect directly, third-party data comes from an outside source, including various places online and offline, like websites, apps, or public records. The company collecting the data is separate from the one that interacted with the user in the first place. Historically, third-party data was collected using small bits of code, called third-party cookies, placed on web browsers. These cookies helped build a picture of online activities and interests as a person visited different websites. The cookies then crumbled and were combined into groups by data brokers or special data platforms, which advertisers could then buy to target ads to expand their reach. ### What Are The Different Types of Third-Party Data Available? Third-party data comes in a bunch of different forms, each offering different insights for targeting ads. One of the most common types is demographic data, which includes facts like age, gender, income, and education. It may sound basic, but this is a huge boost to helping advertisers understand their audience's profile. Another is behavioral data, which tracks online actions like websites visited or products viewed. This provides those much-needed clues about users’ interests, like categorizing someone as an exercise enthusiast if they visit lots of fitness sites, for example. Lastly, intent data suggests that a person is taking those first steps to buy something soon. This comes from specific actions, like repeated searches for a product. If someone searches for "best electric cars" — especially more than once — it shows clear intent to buy. As an advertiser, that’s the group you want to be reaching. ### How Is Third-Party Data Typically Used in Digital Advertising? Third-party data has been key for reaching more and more people using increasingly precise targeting. It helps advertisers find and connect with potential customers who fit their ideal buyer, but haven't directly engaged with their brand (yet) by giving access to profiles of people similar to their existing customers. This expands campaigns beyond the usual immediate customers, branching out and finding new prospects, e.g., a pet supply company finding pet owners who haven't discovered their store. The possibilities for new growth can be limitless. Advertisers use third-party data to make audience targeting significantly better. It gives them the power to divide the online audience into specific groups based on demographics, interests, or purchase readiness, adding a level of organization and segmentation. A luxury travel company might target "frequent business travelers," and a simple shift like that from broad to specific targeting can greatly improve an ad campaign. In my experience, more specific targeting yields better ad results. It also provides insights into consumer behavior and preferences on a larger scale. By looking at combined third-party data, advertisers can understand what's popular and how people typically shop, even beyond their own customers. This wider view helps create better ad messages and suggest where to place ads, shaping campaign strategy. Even if you can’t use it for your current campaign, it’s helpful information to have down the road. But, don’t lose sight of the goal: Using third-party data is meant to improve the relevance and effectiveness of ads. When an ad is shown to someone who is genuinely interested, it’s much more likely to be noticed, clicked, and lead to a desired action. I've found that relevant ads feel like helpful suggestions, not interruptions, and that alone is a huge selling point for users inundated with ads. ## Benefits of Using Third-Party Data Third-party data offers strong benefits for advertisers, mainly for reaching lots more people and targeting them precisely. ### How Can Third-Party Data Help Advertisers Reach a Wider Audience? Third-party data helps advertisers reach significantly more potential customers beyond their known base. By using information collected by outside companies across multiple platforms, advertisers can find people similar to their ideal customers, even if they've never visited their site. This allows for a deeper market reach and helps you discover new customer groups. It’s an effective way to grow your business, find new audiences, and discover new insights and selling points. ### How Can It Enhance Audience Targeting and Segmentation? Instead of showing ads to everyone, advertisers can buy specific groups based on detailed criteria like "people looking to buy a new car." Detailed grouping makes ads highly relevant, and with increased targeting, leads to more efficient ad spending and a higher chance of connecting with interested people. This level of detail truly helps campaigns succeed. ## Concerns and Challenges Associated With Third-Party Data If I’m making third-party data sound great so far, that’s because it is…for increased targeting. But, this article wouldn’t be complete if we didn’t talk about its drawbacks as well. Along with all the advantages, third-party data brings serious worries and challenges, especially regarding privacy and data reliability. These issues are especially noticeable nowadays, and are changing the landscape of digital advertising fast. ### What Are the Key Privacy Concerns Surrounding the Collection and Use of Third-Party Data? This is the big one we need to talk about. The main privacy concern is the lack of clear permission and openness. When outside companies gather data, people often don't know what's collected, how it's used, or who sees it, which can make them feel uneasy and like they've lost control of their personal information. The idea that without permission, an unfamiliar company knows their age, income, and browsing history, not to mention whatever else they discover, feels deeply intrusive. This worry has led to public outcry and the passing of some major new laws across the world. ### How Do Regulations Like GDPR and CCPA Impact the Use of Third-Party Data? You’ve probably heard about, or seen a popup about, one or both of these when visiting a website. Rules like GDPR (General Data Protection Regulation) in Europe and CCPA (California Consumer Privacy Act) in the United States have fundamentally changed how third-party data can be collected and used by giving people more control over their data. GDPR requires clear consent, and CCPA gives consumers rights to access, delete, or opt out. They’re not flawless, but they help, and we absolutely needed some regulations in place. From the advertiser’s perspective, though, these rules have made using third-party data much more complex and risky without strong compliance, and has pushed them to be more careful overall. On top of that, data accuracy and quality are real concerns, too. Since this data is pulled from many different places without consistent checks, its reliability can be questionable, and it isn’t always correct. Profiles might be old or interests wrongly guessed, for example, and using incorrect data wastes ad money and creates a bad user experience as a whole. It's hard to ensure data is clean when you don't collect it directly. The lack of transparency in data sourcing is a big problem. Advertisers often buy audience groups without knowing exactly where or how the data was collected, and that makes it hard to check its quality, legality, or ethics, creating risks. Not knowing the full story is a real challenge if you’re trying to practice responsible advertising. Finally, relying too heavily on third-party data carries risks. It can mean targeting the same groups as competitors, raising costs. Also, as laws get stricter and technology changes (like the removal of cookies), a strategy built mainly on third-party data becomes unstable. ## The Future of Third-Party Data in a Privacy-Focused World Third-party data is a field that’s changing fast due to privacy concerns and internet technology shifts and growth. So, what does its future look like? Hard to say for sure, but one thing I’m pretty confident of is that it’ll have to be about adopting new ideas and moving away from traditional tracking. ### How Are Changes in Browser Privacy Settings Affecting Third-Party Data Usage? The biggest change is the ongoing removal of third-party cookies by major web browsers (looking in your direction, Google Chrome). These cookies were the main technology for third-party data collection for years, allowing tracking across websites, and getting rid of them means the old way of building user profiles across multiple websites for advertising will largely stop. It’s a major shift, forcing advertisers to rethink how they’ll need to find and reach audiences. ### What Are the Emerging Trends and Alternatives to Third-Party Data? As the industry moves to a cookieless online world, new ideas are popping up and gaining traction, all focused on privacy and user permission. One key shift is a bigger focus on first-party data. This is the information your company collects directly from your own customers and website visitors, the big difference being that there’s usually clear permission. It includes measures you need to know, like purchases or website visits, but because you collect it, you own and control it, and it's seen as the most valuable, accurate, and privacy-friendly data. First-party data might be your most reliable asset moving forward. Zero-party data is even more direct, i.e., information customers willingly share, like preferences or personal details. This comes from quizzes or surveys, making it accurate and trustworthy since it's straight from the source. Contextual advertising is also returning strongly. Instead of tracking users, ads are placed on web pages with content that’s directly related to the ad. For example, a hiking boot ad on an outdoor adventure page. This method doesn't track individuals, so it's naturally good for privacy. It’s an older idea, but it’s coming back with smarter technology. ### How Are Advertisers Adapting to a Cookieless Future? Advertisers are adapting by changing their data and their overall thinking. This means focusing on collecting, managing, and using their first-party data, and building customer data platforms (CDPs) to combine information and create a full audience picture. They also try to create engaging experiences to encourage users to share zero-party data. ### What Role Will Data Clean Rooms Play in the Future of Data Collaboration? Data clean rooms are becoming important for safe data sharing. These are secure places where companies can combine their direct customer data for analysis and to find matching audiences, without showing actual private user data. It allows for valuable insights and audience use, while protecting privacy, promising safe data sharing in the future. ## Best Practices and Considerations for Using Third-Party Data The shift away from traditional third-party cookies is shaping up to be pretty clear, but third-party data in other forms may still play a role, so understanding how to use it remains important. ### What Are the Best Practices for Responsibly Using Third-Party Data? Even as the data world changes, the basic rules for responsible data use are vital, and show your audience that you care about their online privacy. Always be clear and get permission to ensure data was collected with consent, and that its use matches user expectations. Pay attention to data quality and accuracy, since bad data wastes your time and effort, too. Also, check your data sources, understand collection, and test how well third-party audience groups perform. Finally, mix it with your own first-party data — strong strategies will always combine outside insights with what you know directly about your customers. ### How Can Advertisers Ensure Compliance With Privacy Regulations When Using Third-Party Data? Making sure you follow privacy rules like GDPR and CCPA is extremely important now, and that starts with thoroughly checking your data providers. You’ll need to ask about their data sources, how they get user consent, and how they comply with privacy laws themselves. Also, set up strong data management rules within your own company, including regular checks and employee training on privacy. ### What Due Diligence Should Be Performed When Selecting Third-Party Data Providers? When picking a third-party data provider, do a thorough check. Ask about their data sources and consent methods. Understand their data cleaning processes and how often the data is updated. Also, ask about their privacy policies and how they meet rules like GDPR and CCPA, and if they offer clear ways to opt out. Lastly, consider their reputation and ask for client references. Trust is key when relying on outside companies for sensitive information. ### What Are Some Ethical Considerations for Using Third-party Data? Beyond legal compliance, ethical concerns are increasingly vital. One concern is potential for unfairness or bias: If data reflects societal biases, targeting could lead to unfair ads. Another is user control and choice: Ask yourself if using this data aligns with user expectations, and if there’s an easy opt-out. Lastly, consider the impact on user trust: If there are unclear or intrusive practices, that can reduce credibility in your brand and the online landscape. Building and keeping trust is incredibly valuable in a world where it seems like that’s being eroded everywhere. ## Key Takeaways Third-party data has been central to digital advertising, providing broad reach and scale for audience targeting. It helped advertisers find and group users from outside sources based on their demographics and behaviors. However, the industry is changing due to privacy rules and technological shifts, especially the removal of third-party cookies. This evolving landscape means a focus on privacy-first approaches. Advertisers are investing in first-party data strategies, collecting direct, consented information from their customers. Zero-party data, willingly shared by consumers, and smarter contextual advertising are key alternatives. Data clean rooms are also emerging as secure ways for companies to share data while protecting privacy. While traditional third-party data's role is changing, understanding its past, benefits, and challenges is important for the future of digital advertising. The focus is now on reliable, privacy-aware data foundations for effective audience connection. ## Frequently Asked Questions (FAQs) ### Is third-party data still a viable option for digital advertising? The role of third-party data in digital advertising is definitely changing a lot, but I wouldn’t say that it’s disappearing entirely. The traditional way of using third-party data depended heavily on cookies that tracked users across different websites, and that’s being phased out. Google Chrome's move away from these cookies is a big part of this shift, and means that tracking individual users anonymously across many websites is going to become much harder — even impossible, in some cases. The main idea of getting useful information about audiences from outside sources is the primary goal, though, and that’s not gone, not by a long shot. What we’re seeing now are new, more privacy-friendly ways to share and combine data, like data clean rooms, which let advertisers and publishers securely combine and look at their own direct customer data (first-party data) without actually sharing private user information. Methods like this create combined insights that can still help with ad targeting. Also, companies that provide third-party data are changing their approaches, finding new ways to identify users with their permission, using information about the context of web pages, and creating smarter data models. So, while buying broad, general third-party audience groups might decrease, the ability to get valuable insights from outside sources (through more privacy-friendly methods) will likely continue to be a part of smart advertising plans. ### What is the difference between first-party, second-party, and third-party data? Before you get started experimenting with any version of these, it’s best to understand them at a high level. Each type is defined by where it comes from and its connection to your business. First-party data is the information your company collects directly from its own customers and website visitors. This would include data from your website, mobile app, customer records, email lists, or even in-store interactions. When you think of first-party, consider what products they've bought or pages they visited on your site, since you collect this data yourself and you own it, control how it's used, and typically get clear permission from the user. It's seen as the most valuable, accurate, and privacy-friendly data you can have, and it represents a direct connection with your audience, too. Second-party data is essentially another company's first-party data that they've agreed to share or sell directly to you. It's a direct, agreed-upon exchange between two companies. For example, an airline might share its anonymous passenger data with a hotel chain. Much like first-party, this data is generally high quality because it comes from a reliable source, and the sharing agreement outlines how it was collected and how it can be used. It’s a go-to trusted way to gain insights about audiences beyond your own direct interactions. Third-party data is what we’ve covered most here, and as discussed, it’s data collected by a company (a data broker) that doesn't have a direct relationship with the people whose data is being collected. This type of data is gathered from many different places and then sold or shared with multiple advertisers, but has its positives and negatives. On the upside, it offers broad reach and a wide range of audience groups, but it often comes with questions about its accuracy, where it came from, and whether it follows privacy rules. This type of data has traditionally been the broadest and easiest to access for expanding reach, but its future is changing a lot due to new privacy standards and limits from web browsers. ### How can I assess the quality of third-party data? This one can be tricky since you don't control how the data was originally gathered. Even so, there are some key steps and signs you can look for. First off, thoroughly research the data provider's sources and methods. Do they clearly state where their data comes from and how it's collected and updated? Providers who are open are generally more reliable, while a lack of clarity here is a big warning sign. Second, consider how recent the data is and how often it's updated. Is it refreshed regularly, or is it old and stale? Outdated information can quickly become useless and lead to wasted advertising dollars. Third, ask for examples of performance or case studies from other advertisers who used the data. While results differ, a provider sharing success stories might offer more reliable data. Finally, and maybe most importantly, test the data yourself. Start with a small budget to run campaigns targeting these groups. Watch your key performance goals. If the data doesn't perform well, it's likely not worth the investment. Remember that your own results are the best way to judge data quality. ### What are data brokers and what role do they play? Data brokers are companies that collect, combine, process, and then sell or share information about people and groups. Their main business involves getting huge amounts of information from many different places, such as public records, commercial websites, and mobile apps, and then organizing it into various groups for advertising and marketing purposes. They’re kind of like the middlemen of the process, connecting companies that have data with companies that want to buy it. Their role has historically been to help advertisers reach a wide scale — by buying data from a broker, an advertiser could access audience groups they couldn't collect directly. This was helpful for new businesses or those trying to expand into new markets without having a large customer base. Data brokers create the "third-party data" products that advertisers use, offering ready-made audience groups like "people planning to buy a new car," and ultimately saving advertisers time and effort. The role of data brokers going forward is facing a lot of review and potential changes though, since their practices have been criticized for not being clear about where their data comes from and whether they have user permissions. With more privacy rules like GDPR and CCPA, and the removal of third-party cookies, data brokers are being forced to change how they collect data to meet stricter privacy standards. Their future role will likely shift towards more privacy-friendly methods of sharing, or focusing on insights from consented, combined data sets, rather than tracking individual users across multiple websites. --- ### Pixels: Tracking Your Audience Across the Web (and Email) URL: https://www.taboola.com/marketing-hub/pixel/ Last Modified: 2025-06-22 07:51:48 There are all sorts of ways we are tracked online, from cookies to web beacons to fingerprinting and more. Marketers in particular are adept at tracking users' online behavior, and as nefarious as online tracking can seem, often it really is done with the best interest of the user in mind, with marketers striving to only serve content that will resonate with their audience — and ultimately, of course, make sales and other conversions. One of the most effective online tracking tools is also one of the least well-known among the average internet user: the pixel. Short for “picture element,” a pixel in image form is the smallest possible amount of light that can be shown on a screen, though the same term is also used for embedded bits of code that don’t display at all. Pixels embedded on websites or in opened emails can track all sorts of things about a user’s online activity. While that can be very useful for advertisers, it can also raise security and privacy concerns. ## Understanding Tracking Pixels ### What Are Common Names for Pixels? In digital marketing, pixels are commonly referred to by several names, all essentially describing the same small snippets of code used for tracking user activity and collecting data. They go by the term tracking pixel most often, but also by web beacon, marketing pixel, conversion pixel, pixel tag, and spy pixel. Whatever the name, a pixel's primary function is to track user behavior and gather data for analysis and targeted advertising. ### How Does a Pixel Collect Data About User Behavior? When a user visits a webpage that uses tracking pixels (or opens an email that does), the pixel's image request automatically triggers data collection about the person’s activity. This data, including IP address, browser type, device information, and actions on the page, is sent back to the marketer’s server, allowing them to track user behavior and personalize their experience. ## How Pixels Work ### How Is a Pixel Typically Implemented On a Website or In an Email? When a user visits a website or opens an email containing the pixel, their browser downloads the image, sending a request to the server where the pixel is hosted. This action triggers a data transmission back to the server, providing information about the user's interaction. On a website, tracking pixels are implemented by inserting a small piece of code, usually in the header or body of a webpage's HTML code. This code typically contains an external link to the server where the pixel image is stored. ### What Information Can a Pixel Track? Pixels can gather a wide range of information about a user's online activity, including their IP address, the device type they’re on, the browser they are using, and even the specific pages they visit, items they purchase, or other conversions, like a form submission. They can also track when an email is opened or a link is clicked within an email. ### How Is the Collected Data Transmitted to Advertising or Analytics Platforms? When a visitor lands on a webpage, a pixel’s code is executed by their browser. The code then gathers information about the visitor's actions and interactions on the site. This gathered data, such as page views, clicks, session duration, and so on, is then transmitted to the platform's server for processing and analysis. Pixels are closely related to the better-known tracking tool, the cookie. Pixels act as triggers, notifying a server or analytics platform when a user interacts with a website or email. When a pixel "fires," it instructs the user's browser to set or retrieve a cookie. This cookie then stores information about the user's behavior or preferences. This combination allows for more in-depth tracking and personalization. For example, a pixel can track a user's visit to a specific product page, and a cookie can remember that visit to later serve them a targeted ad based on that behavior for that product on another page, or even on another platform. ## Types of Pixels and Their Uses All pixels are used to collect data about a person’s online activity, but different types of pixels have specific uses. ### What Are Advertising Platform Pixels and What Are They Used For? Advertising platform pixels, such as a Facebook pixel or a Google Ads conversion tracking pixel, are used to track user behavior, interactions, and conversions, providing valuable data for advertisers and marketers. This data helps in enhancing targeting, measuring campaign performance, and building custom audiences. See how Taboola can maximize your affiliate marketing campaigns. ### What Are Analytics Platform Pixels and What Data Do They Track? An analytics platform pixel (like a Google Analytics tag) tracks a variety of metrics, including website visits, email opens, ad impressions, conversions, and user behavior. This data helps businesses understand user engagement, measure the effectiveness of digital initiatives, and personalize marketing strategies. ### What Are Email Tracking Pixels and What Information Do They Provide? Email tracking pixels, also commonly known as web beacons or, more ominously, spy pixels, are tiny, often invisible images embedded in emails. They track email opens and can provide detailed information about how recipients interact with the email, such as time of opening, location, device type, and even whether they clicked on links. ### What Are Third-Party Data Pixels? Third-party data pixels are snippets of code, often one-by-one transparent images, that are embedded within websites or emails to track user activity and send the collected data to a server controlled by a third party. These pixels gather information about user behavior, such as website visits, clicks, and interactions, which can be used for analytics and targeted advertising. ## Benefits of Using Pixels in Digital Advertising Knowledge is power, and when it comes to marketing, knowledge about customer behavior means targeted ads, better segmented audiences, and more conversions. ### How Do Pixels Enable Conversion Tracking? Conversion tracking pixels enable marketers to track user actions and conversions after they click on an ad or simply visit a website. These small pieces of code trigger a notification to a server when a user performs a desired action, like making a purchase or filling out a form. This allows marketers to measure the effectiveness of their ads (or the lack thereof) and helps them to better understand how users are behaving on their websites. Pixels can be vital for measuring the effectiveness of ad campaigns by tracking user actions and interactions within and across websites. They help advertisers understand which campaigns are driving conversions, leading to site visits, and more, and they allow for optimized targeting and the creation of audiences for future campaigns. ### How Are Pixels Used for Remarketing and Retargeting Audiences? By triggering the placement of cookies in a user's browser, pixels allow advertisers to recognize users even as they visit other sites or use other platforms. Marketers can therefore show users relevant ads as they browse, thanks to what they learned when the pixel was active. This allows for retargeting campaigns that re-engage users who previously interacted with a website, but didn't complete a desired action. ### How Can Pixels Contribute to Building Lookalike Audiences? Pixels contribute to building lookalike audiences by providing the source data used by advertising platforms to find new users sharing similar characteristics to existing ones. Basically, if one user spends time on X, Y, and Z websites and then goes off to A, B, and C and converts, another user who also seems to like X, Y, and Z may respond to ads for A, B, and C. ### How Do Pixels Facilitate Cross-Device Tracking? Tracking pixels facilitate cross-device tracking through a combination of techniques, but they do so primarily by collecting and analyzing data associated with a user across different devices. The data a pixel sends back carries information about the user and their device, such as an IP address, browser type, operating system, and unique identifiers like user IDs, if the user is logged in. With this information, marketers can tell when the same ID pops up in other places. If a user logs into a service (such as a social media platform or an email provider) on multiple devices, the tracking pixel can link these devices to the same user based on their login information. ## Privacy Considerations and Pixel Usage Considering one of the names for pixels involves the word “spy,” it’s no great shock that these digital tools raise some security, privacy, and ethics concerns. ### What Are the Privacy Implications of Using Tracking Pixels? Tracking pixels raise significant privacy concerns due to their ability to collect personal data without explicit consent, and the potential for misuse. They can collect browsing history, location, and device information, and can be used for cross-site tracking, profiling, and even exploitation by malicious actors. Regulations like GDPR (General Data Protection Regulation) and CCPA (California Consumer Protection Act) have addressed these concerns, requiring explicit consent for tracking and offering users the right to opt out. Collection of this data can be troubling to some people on its own, but as it’s often shared or even sold to other parties, it can put users in a compromised position. ### How Do Regulations Like GDPR and CCPA Affect the Use of Pixels? The GDPR requires explicit, informed, and freely given consent before parties collect and process personal data via pixels. This means website visitors must actively agree to cookie and pixel tracking before any data is collected. Pre-checked boxes or implied consent are not sufficient. You have seen the implications of this act many times when you’ve clicked “Allow” or “OK” at the bottom of a freshly loaded site. The CCPA, on the other hand, does not require express prior consent, but it mandates that businesses implementing data collection provide consumers with a way to opt out of the "sale" or "sharing" of their personal information collected through pixels and cookies. This includes implementing a "Do Not Sell or Share My Personal Information" link, for example. ### What Are Best Practices for Obtaining User Consent for Pixel Tracking? Transparency is key! Marketers should use plain language to inform users about the use of pixels (and cookies), as well as about the type of data to be collected and how it may be used. The parties creating the ads must maintain a comprehensive and easily accessible privacy policy that clearly outlines your data collection and tracking practice, and you would be well-advised to use a consent banner or pop-up to inform users about cookie and tracking technologies at play. For people who wish to avoid being tracked, there are plenty of steps to take. Browser privacy features, e.g., have significantly impacted the functionality of tracking pixels by limiting their ability to collect and transmit user data. ## Key Takeaways You can’t see a pixel, but it can see you, and it will send the data it sees back to the marketers who placed it on a website or within an email. Pixels collect data such as a user’s IP address, the type of computer or smart device a person is using, their browser type, and about the actions they take on a site. This data is then stored by a digital cookie, which can be used to track the person across different sites and platforms. These tiny, invisible images provide data that helps businesses understand how users interact with online content and improve their marketing strategies. ## Frequently Asked Questions (FAQs) ### Are pixels visible to website visitors? Technically, yes, with a magnifying glass or microscope, but effectively, no. Pixels are tiny, nearly invisible images embedded in websites and emails that are used to track user behavior and activities. They are usually one-by-one in size and are often transparent, making them virtually undetectable to the human eye. ### Do all pixels track the same information? While pixels all function similarly, different pixels are designed to collect specific data and track different events. For example, a Facebook pixel might track website visits and conversions, while a Google Analytics pixel tracks website engagement and user behavior. ### How can I check if a pixel is installed correctly on a website? To see if your pixels are installed properly on a site, you can use tools like the Meta Pixel Helper Chrome extension, the Google Tag Manager, and other tools. ### How do tag management systems help manage pixels? Tag management systems like Google Tag Manager simplify the process of managing tracking pixels on websites and applications. These systems act as a central hub for all your tracking scripts, allowing you to add, modify, and remove pixels without directly editing your website's code. --- ### Automotive Marketing Trends 2026: Global Shifts Marketers Should Know Of URL: https://www.taboola.com/marketing-hub/automotive-marketing-trends/ Last Modified: 2026-03-08 13:51:23 Most shoppers aren’t entering car dealerships blind — today’s consumers are likely to have already done their research, seeing what’s available and what they’re likely to pay for various models and features. It’s safe to say that by the time most shoppers actually enter the dealership, they’re probably close to becoming a buyer. This research and preparation phase has become much easier as digital and physical automotive marketing tactics overlap. Automotive marketers can do a lot of work to influence car buyers from the start of their browsing journey through to driving a new car home. The automotive market is predicted to see modestly steady growth over the next decade, and marketers will need to work harder to attract and engage audiences as competition grows. ### What’s changed in our 2026 update: - New trends added. - All entries include updated and current information, advice, and stats. ## Trend 1: The Shift Toward Digital-First Car Buying Experiences Digital-first car buying experiences may have started with digital marketing, but they’ve grown exponentially with the availability of AR (augmented reality) and VR (virtual reality) options and sites like Carvana and CarGurus. Marketers should consider these areas as they’re building superior digital experiences — with a surplus of channels and sites, the primary goal for performance marketers has to be customer experience. ### A Good Experience Automotive shoppers have lots of options to buy a car today, and the experience has to be easy and seamless — no pressure and no bait-and-switch on costs, and digitally, multiple ways to explore and research vehicles. Marketers should consider all the tools at their disposal to create great automotive buying experiences, such as virtual showrooms, online configuration tools, and test-drive schedulers. Showing off products in high-def, 3D views, and connecting the customer to the brand, are important in 2025. Gathering and using data safely will also be key to building a positive, successful digital experience. ### All the Details Consumers can find a ton of detail when they’re starting their journey towards car shopping and even buying vehicles online. Providing specific information is essential from the marketing perspective, and may include: - Price comparisons. - Reviews and testimonials. - Looking for actual cars for sale. - Comparing models of an auto brand. - Looking up current car value. - Finding a local dealer. - Searching for fuel efficiency and safety rating data. Digital advertisers should keep these topics in mind as they’re generating leads and online sales for dealerships, and consider all their possible tools, including optimized websites, lead capture options, social and search ads, live chat, and retargeting. Specific automotive techniques like virtual showrooms or configuration tools can offer a way to engage a prospect, capture their contact info, and move them toward purchase. Here’s what else to know: - About 39% of car dealers now allow consumers to do every step of the auto buying process online. - 95% of car buyers search online before buying. - 14% of online car buyers surveyed didn’t test drive before buying. - 78% of American consumers who bought a car online found it a highly satisfying experience. ## Trend 2: Consumers Have Affordability on Their Minds In 2026, automotive marketers will have to get even more creative in reaching prospects and move quickly to reach audiences. According to Equifax, vehicle prices are still elevated, interest rates are high, and consumers are concerned about affordability. Because of those pressures, the average vehicle life has gotten longer, to over 12 years in the U.S. For marketers, it’s important to focus on intent signals from automotive browsers or shoppers to understand when a prospect is actually ready to buy. Consider messaging around affordability and accuracy so prospects don’t encounter sticker shock at the dealership or online point of purchase. Here’s what else to know: - The new vehicle market is projected to deliver 15.8 million units in 2026, down 2.4% from last year. - 68% of consumers are worried they will overpay for their next vehicle. - 72% of consumers expect tariffs to make vehicles less affordable. ## Trend 3: Personalization and Data-Driven Automotive Marketing Personalization and data-driven marketing can make a big impact on the success of automotive campaigns. Because car-buying still has such an essential in-person component, digital marketers should consider how to personalize the buyer’s journey to move them toward a dealership visit. ### Data-Driven Marketing Automotive marketers have to make sure their technology is up to speed so there’s a single source of truth. In a multi-channel marketplace like automotive, it’s important to capture and use up-to-date prospects and repeat customer data to avoid repetitive tactics and spamming users. Plus, it’s essential to move at speed to reach prospects when they’re ready to buy. Marketers should gather first-party data as much as possible through web, app, or dealership interactions and keep the CRM updated continually with any new details gathered. For automotive marketing, metrics should include industry-specific numbers, such as dealership visits. As with collecting consumer data in any industry, follow regulatory compliance by ensuring that users understand why you’re gathering data and give them options to opt out. Automotive CRM systems can also capture vehicle data, so make sure the same regulations are followed. ### Personalization With high-quality data and the possibilities of AI, automotive marketers can tailor messaging to prospects. These messages, timed correctly, can help bring a digital prospect into the dealership and close the sale. Personalization tactics might include: - Targeted ads. - Emails with tailored offers or promotions. - App notifications or texts. - Videos showing off vehicle features. - Web and landing page personalization. For automotive marketers in particular, demographic data, location information, lifestyle details, online behavior, car buying history, and purchase intent are all useful. These pieces of information can help inform whether a new prospect is looking for a sports car or a minivan, and whether new technology or storage space is a higher priority. Here’s what else to know about personalization and data-driven marketing: - 6 out of 10 car buyers are open to considering multiple vehicle options when they first start to shop. - Third-party sites are the most used for car shopping, by more than 80% of shoppers. - Buyers often make a decision on buying a car within 48 to 72 hours of initial dealer contact. ## Trend 4: The Growing Importance of Hybrid and Electric Vehicle (EV) Marketing Details are generally an essential part of automotive marketing, and even more so in the hybrid and electric vehicle segments, which have seen customers trying to get to grips with not only a huge technological change, but a series of ever-changing regulations and incentives. Those shopping for EVs and gas-electric hybrid vehicles need all the facts about mileage, charging, and more to have confidence in choosing the right make and model. There are also signs that electric vehicle sales may be slowing down, with hybrids continuing to grow. Marketers can promote the technology aspects of EVs as well as tailor messaging to sustainability and values-based buying. ### Hybrid/EV-Specific Information These vehicles are still new, so many potential buyers need education around them. Marketers should create content that demystifies EVs as much as possible, such as showing users how and where to charge the car, the tech and charging infrastructure roadmap, how the battery works and what it requires, how far a charge lasts, the cost of the car (and potential cost savings on gas), and other specifics. This should continue through and after the sale, such as providing training, customer support, maintenance reminders, and more. ### Unique Messaging The typical messaging for gas-powered vehicles doesn’t necessarily align with EV shoppers. Digital marketers should take into consideration the distinct benefits that EVs bring, namely sustainability. They should also focus on brand building, as many EV makers are new to the market and not household names. For established or legacy automotive brands, their electric vehicle options also might not be well known. The overall value prop for EVs is more holistic than with traditional vehicles, and buyers are often willing to pay more for the vehicle in exchange for values alignment. Still, marketers should focus on simple, benefits-led messaging, highlighting the broader value — e.g., fuel cost savings over time — and any unique features to the particular EV auto brand. Hybrid vehicles offer drivers a mix of both gas and electric, and their prices may be more appealing when affordability is top of mind. Here’s what else to know about hybrid and EV trends: - At the end of 2024, hybrids made up about 60% of all electrified vehicle sales in the U.S. - 47% of EV shoppers want to pay less than $40,000, but in 2024 there were just four models sold at that price. - The average EV transaction price was $61,702 in 2023, about $15,000 more than traditional vehicles. - Hybrid vehicle registrations were almost 14% in the first quarter of 2025, while electric vehicle registrations were at 8% in the U.S. ## Trend 5: Leveraging Video and Immersive Experiences in Automotive Marketing An automotive buyer is probably doing a lot of research online ahead of actually visiting a dealership. There’s a lot of opportunity for marketers to provide new, unique, and engaging online experiences to these shoppers, particularly as cars themselves become more sophisticated technologically. ### Video Running digital marketing campaigns for automotive in 2026 requires high-quality, likely short-form video that can connect with prospects and speed up the buying cycle. Marketers can consider how to incorporate interesting, dynamic perspectives and vehicle demos that show off compelling features and innovations. Testing various channels like YouTube, Instagram, and TikTok can help marketers find which type of content resonates with customers, and deliver different levels of information and experience at different points of the customer journey. Motion ads — featuring eye-catching graphics that move for up to 15 seconds at a time — can also be effective. Renault Australia, for example, used a mix of image and motion ads to lower their CPA by 51%, and their CPC by 16%. ### Interactive Experiences Automotive marketing has proved to be a great testing ground for virtual reality and augmented reality technology use. With prospects doing so much online research before visiting a dealership or purchasing otherwise, VR and AR can fill in a lot of the gaps of the driving experience. VR can help customers explore showrooms and simulate the driving experience in a variety of conditions. AR allows a prospective customer to envision various finishes and design options with a virtual car, as an example. Here’s what else to know: - More than 70% of car shoppers used at least two devices during their research process, according to Statista. - More than 60% of automotive shoppers visited a dealership or dealer website after watching a video about a particular vehicle. - The automotive AR and VR market is projected to reach $5.6 billion this year, a CAGR of 26% from 2021. ## Trend 6: The Strategic Use of Social Media and Influencer Marketing in Automotive Both social media and influencer marketing can drive digital marketing success. Choosing which specific platforms and influencers will depend on the automotive brand and its audiences, and creative teams should test and learn to see what engages prospects and drives them towards a conversion. Social media engagement is a good indicator of interest for marketers, while influencer marketing can be measured in engagement, click-through rate, and referrals. Automotive marketers should also pay attention to car-specific review platforms like Edmunds and Kelley Blue Book, along with Google Reviews and Yelp for dealership reviews. ### Social Media Automotive digital marketing will probably use a mix of Facebook, Instagram, and TikTok, fluctuating depending on which performs best for a particular brand and prospects. Specific targeting is very useful across social media and other digital platforms for automotive advertising, and offers lots of room for creativity and constant testing to avoid user fatigue. Targeting options usually include demographic information, such as age, type of location (urban, suburban, or rural), income, online behaviors, values and motivations around car buying, and more. Carrying out A/B testing according to demographic information can help marketers understand which format types to use with which audience segment, plus other details like copy and imagery. More tailored sites like Pinterest or LinkedIn might also be a good fit, depending on the audience and marketing goals. In addition to posting fresh content on social media platforms, a digital marketing team will have to monitor those sites to find opportunities to engage with users further, such as in comments, replies, or reshares. ### Influencer Marketing Brands with devoted fans can use influencer marketing to their advantage by choosing passionate, authentic influencers who are deeply familiar with the brand. EV brands might focus a large part of their marketing strategy on influencer marketing, particularly if there are notable public figures or celebrities passionate about EV tech and innovation. In general, influencers can be useful for automotive brands to generate excitement, offer real-life, detailed reviews, and create trust. Here’s what else to know: - More than 38% of car buyers turned to social media and influencers before purchasing. - 1.37 million influencer posts mentioned the auto industry in just one quarter of 2023, with Instagram claiming 90% of those posts. - 46% of consumers consider car influencers to be a trustworthy source of information. - Influencer-driven campaigns can increase sales leads for automotive by up to 15%. ## Key Takeaways Automotive marketing in 2026 offers a lot of opportunity, despite economic challenges and the need for speed. Digital marketers have to put the customer experience first to make sure it’s both smooth and protects the brand, while offering relevant details, looking for intent signals, generating leads and moving the buyer journey along. ## Frequently Asked Questions (FAQs) ### What are the most effective digital advertising channels for automotive in 2026? Digital marketers and advertisers have a lot of options for automotive in 2026. A compelling mix of display, vertical, and native advertising will continue to yield results, and social media platforms like Facebook and Instagram will remain essential, as will YouTube. AI-powered analytics can help marketers use time and resources wisely by finding and targeting high-intent users and meeting prospects where they are in their journey toward buying a car. ### How can automotive brands build loyalty in the digital age? Automotive brands have more channels and options than ever before to build customer loyalty through personalization and expanded channels like social media. Building loyalty can be incredibly effective and lucrative in the automotive industry, where buyers often gravitate toward a particular brand as a reflection of their own values, and recommend the brand to others. Loyalty programs using data-driven targeting can start as soon as the car purchase is complete, making sure to continue connecting with the buyer through dealership service offers and other social and digital campaign nurture tactics. While the buying cycle may be complete with a first purchase or lease, marketers can essentially restart the consideration part of the funnel for the audience segment that has already purchased from the brand. ### What are some successful examples of automotive digital marketing campaigns? Both established and upstart automotive marketers can be successful online, going beyond traditional TV ads and billboards. Chevrolet worked to combat declining brand health and reach younger audiences in 2015 with its Best Day Ever campaign, which used a unifying social media hashtag — #bestdayever — along with celebrity performances, surprise acts of kindness, and other events across video, live streams, social platforms, and more. Chevy saw great results: 75% of the campaign’s engagement was with people younger than 35, and 80% of campaign activity was on mobile devices. ### How is technology (e.g., AI, connected cars) impacting automotive marketing? New and emerging technology is having an impact on automotive marketing as digital options abound to target and convert prospective buyers. AI, for example, enables more data-driven and hyper-personalized marketing tactics, so marketers can create better, more seamless buyer journeys. ### What are the key metrics for measuring success in automotive digital marketing? Automotive marketing includes some industry-specific metrics, such as cost per qualified website visit, cost per online car configurator, or whether a prospect has visited a dealership and how often. The baseline key metrics for measuring automotive digital marketing success are generally the same as in other areas of digital marketing. They include cost per lead (CPL), customer acquisition cost (CAC), and conversion rates, along with softer metrics like social engagement and video views. Automotive marketers will also track brand metrics, such as sentiment and brand recall. --- ### Software-as-a-Service (SaaS): How It Is Transforming Digital Marketing URL: https://www.taboola.com/marketing-hub/software-as-a-service/ Last Modified: 2025-07-29 09:51:36 When it comes to digital marketing and advertising, having the right tools for the job will give your business a significant advantage. But, software isn’t always affordable or accessible, especially when you’re starting a business or need a solution for a rapidly expanding advertising team. Software-as-a-service not only offers the tools and capabilities that can improve the effectiveness of your marketing, but it does so while being more friendly to your budget and personnel. ## Understanding Software-as-a-Service Software-as-a-service, or SaaS, is a cloud-based delivery model that gives you access to software without having to install and manage software on your computer. SaaS is highly versatile, convenient, and rapidly growing in popularity. Chances are you’re already using SaaS in many ways, even if you’re not aware of it: Netflix, Slack, Dropbox, Zoom and Realize are all examples of highly popular SaaS products. ### How Does the SaaS Model Differ From Traditional Software Licensing? With traditional software licensing, when you buy a product, you need to download and install it on your computer. With the SaaS model, you buy a subscription to access software online. There’s no need to download it to your computer, so you can usually easily access the software from multiple computers. ### What Are the Key Characteristics of SaaS? The SaaS model is subscription-based, meaning you’ll need to periodically pay to retain access to the software. Subscriptions are often available monthly, though you can often save money by purchasing annual or longer subscriptions. Tiered subscription plans are also common, where you can pay more for a higher-tier plan to access more features or storage capabilities. SaaS is also cloud-based. There is no physical download to manage, and you’ll access the software using an internet browser. Since you don’t have to download and install the software, you can often easily and quickly start using SaaS. ### What Are the Benefits of SaaS for Businesses and Users? SaaS offers many benefits for businesses and users. Since it’s subscription-based, you may be able to save money by only subscribing during the times when you need the software, rather than having to make a large, expensive purchase to perpetually own software that you won’t use all that often. The subscription model is advantageous to businesses because the software is scalable. Say you have a small marketing team at the moment: You can buy a SaaS subscription for one or two seats then, as your business grows over the years, you can purchase additional seat subscriptions as you need them. SaaS also tends to be easy to use. Features like automatic updates mean you don’t have to worry about maintaining or updating the software yourself. ## SaaS in the Context of Digital Advertising: Benefit For Performance Marketers The right tools can make all the difference for digital advertisers, streamlining processes, enhancing the quality of their work, and providing more accurate and powerful analytics and reporting. SaaS platforms are highly appealing because they offer many benefits to digital advertisers, solving some common challenges like the expense and scalability limitations that often come with traditional software. ### What Types of SaaS Tools Are Commonly Used by Digital Advertisers? Many digital advertisers use SaaS tools to create and manage digital ads. These tools can help advertisers research and target their audiences, create graphics and copy, optimize campaigns, and analyze campaign results. Some common SaaS digital advertising tools include Canva, Adobe Creative Cloud, Facebook Ads Manager, Google Ads, and Realize. ### How Does SaaS Provide Cost Savings Compared to Traditional Software? SaaS platforms are much more affordable than traditional software. Since you can buy SaaS subscriptions for periods as short as a month, you can save significantly compared to the cost of purchasing traditional software outright. Many SaaS platforms even offer free plan tiers that you can use to get started and try out the platform, while upgrading to more expensive tiers will give you access to additional features. The pay-as-you-go model offers excellent flexibility and cost savings. You can add on and remove subscriptions for team members as your team size fluctuates, so you’re never overpaying for software that you don’t need. ### How Does SaaS Advertising Tools Provide Efficient Campaign Management? SaaS platforms are powerful campaign-management tools that can streamline your processes. These platforms function as a central place to store and analyze campaign data, improving efficiency and organization. You can use them to store lead data, for example, then segment the data, A/B test campaign components and messaging, and leverage these analytics to optimize your campaigns. ### How Do Automatic Updates and Maintenance Benefit Advertisers? The automation features available from SaaS tools can also enhance campaign management and save time. These tools are capable of automating repetitive tasks including lead generation, creating landing pages and forms, sending out email sequences to nurture leads, and onboarding new users. By automating time-consuming tasks, your team can focus on the tasks that require a personal touch. Since the platform developer is responsible for the updates and maintenance, your IT department can focus on other needs. When you’re using a platform that is well-maintained, you’ll have the assurance of knowing that its security features and functionality are always monitored and updated, meaning the platform is likely to experience minimal downtime and to offer strong data protection. ### How Does It Offer Scalability to Handle Growing Advertising Needs? Since SaaS platforms are subscription-based, you can quickly scale them as your advertising team grows, and getting new team members started on the platforms is quick and easy. These platforms are also scalable in terms of their features and functionality: Many offer the ability to upgrade your subscription if you want more storage, and some offer advanced features available with higher-tier plans. A smaller business might not need those features initially, but the option to upgrade a subscription makes SaaS platforms highly scalable, so they can be long-lasting solutions for digital advertisers. ### How Does It Facilitate Collaboration and Accessibility for Advertising Teams? Since your team members can access SaaS on their own computers without having to worry about downloading and installing software, SaaS platforms are highly accessible. The ability to purchase subscriptions, rather than buying software outright, makes these platforms accessible even to smaller advertising teams working for startups. SaaS platforms are cloud-based, and most allow for real-time collaboration between your advertising teams. Features like the ability to edit documents in the cloud, make comments, and chat with team members in real time further promote collaboration. And, since the platforms can host documents in the cloud, your team members will always be working on the most recent document. ## Key Considerations for SaaS Adoption in Advertising Many SaaS platforms are well suited for adoption in advertising, but it’s always important to choose a platform that’s right for your team’s specific needs. SaaS platforms share several benefits, but by selectively choosing a platform with the capabilities that are most valuable to your team, you’ll enjoy the full benefits of SaaS in advertising. ### What Factors Should Advertisers Consider When Choosing a SaaS Platform? - Features: Start by making a list of the features and capabilities that are most important to your team. Consider factors like design and content creation tools, collaboration and commenting features, integrations with the platforms you’re already using, and available analytics and reporting features. - Ease of use: Consider the platform’s ease of use, too. The member of your team with the least technology experience should be able to easily learn to use the platform. Find out the type of customer support available, and think about whether the support is available during the hours when your team will be working. - Scalability: Make sure that the platform can scale as your business grows, and consider whether subscriptions for more users will reach a point where they’re cost-prohibitive. - Security: Data security is paramount when choosing a SaaS platform. Look for a platform with comprehensive security measures, and verify that the security measures are frequently updated and adhere to regulatory standards. ### What Are the Implications of Data Security and Privacy When Using SaaS? When you use SaaS, your data is stored in the cloud. This means you’ll have less control over your data than you would if it were stored on your company’s servers, so the risk of data breaches and cyber threats is heightened. When you collect data, you must meet applicable regulatory standards, like GDPR, the European standard protecting personal data, and CCPA, which protects business and individual privacy in California. Any SaaS that you use for advertising needs to also adhere to these regulations when you’re going to be uploading or collecting lead data. To prioritize data security and privacy, look for SaaS platforms that openly discuss their security and privacy measures. Seek out platforms that actively monitor and log their security operations, and that continuously work to improve and update their security measures. ## Challenges and Trends in SaaS for Advertising The use of SaaS in advertising is rapidly expanding and evolving, and its proliferation brings exciting potential to the industry. As SaaS becomes a go-to for advertisers, several challenges and trends are emerging. ### What Are Some Potential Challenges Associated With Using Multiple SaaS Platforms? While SaaS platforms offer many benefits to advertising teams, using multiple platforms can also pose several challenges. Platforms don’t always integrate fully with each other, and some features may be redundant across multiple platforms, so your team might not get the best value out of them. You’ll also need to consider the practicality of using several different platforms. Your team will need to learn to use the different platforms, and switching back and forth can cause confusion. Keeping multiple platforms open all day long can get tedious for team members, and you’ll lose some of the streamlined, time-saving advantages that you would enjoy when using a single platform. ### How Is the SaaS Landscape for Advertising Evolving? SaaS has already had a tremendous impact on the advertising industry, and we’ll only see its use and influence increase over the coming years. Today, SaaS platforms are viewed as essential for advertising teams, and their increased versatility makes them practical choices for small startups as well as for large corporations. SaaS platforms have impacted everything from data analytics to precise audience targeting, and the continuous development of new technology means these tools are more powerful now than ever. The platforms are keeping pace with new trends in advertising, including artificial intelligence (AI) integration, and we’ll likely see them continue to shape and support advertising efforts in the coming decade. ### What Are Some Emerging Trends in SaaS for Marketing and Advertising? As smaller teams and startups with limited budgets and resources have discovered the value of SaaS platforms, the platforms are evolving to meet their needs. For example, platforms like Jotform are increasingly offering no-code or low-code tools that allow smaller teams without coding experts on staff to create their own forms and websites. The use of AI is also a major trend that should continue to evolve as the technology advances. SaaS platforms feature AI tools that can provide detailed analytics, generate content, streamline workflows, engage with leads, automate follow-up processes, and more. ## Key Takeaways SaaS not only offers many benefits to digital advertising teams, but it’s transforming the way the teams work. These platforms can streamline workflows while helping teams to enhance the quality and accuracy of their work. Most importantly, these platforms are affordable and accessible to any digital advertising team, empowering teams to maximize their advertising capabilities and effectiveness. ## Frequently Asked Questions (FAQs) ### What are some popular SaaS platforms for digital advertising? There are many popular SaaS platforms for digital advertising, including Buffer, Canva, Adobe, Hootsuite, and Realize. ### How do I ensure the security of my data when using SaaS? To ensure your data is secure, look for a SaaS platform that clearly explains its privacy and security features. Make sure the platform’s security is regularly updated, too, and that it complies with any applicable data privacy regulations. ### What should I look for in a SaaS agreement? Thoroughly read any agreement before signing up for a SaaS platform. Look for information about the pricing, renewal, and termination clauses. Carefully review details about how the platform protects your data, including whether it is compliant with GDPR or CCPA regulations. Verify that you retain ownership of any data and intellectual material that you upload. ### How does the pricing of SaaS typically work? SaaS pricing is typically subscription-based. You’ll usually have an option to subscribe monthly, but can often save if you choose an annual or longer subscription. Some SaaS subscriptions also offer a brief free trial. --- ### Customer Retention Rate: What It Is, How To Maximize It URL: https://www.taboola.com/marketing-hub/retention-rate/ Last Modified: 2026-06-22 08:37:51 Effectively attracting customers to your business is essential in driving sales, but customer retention is equally important. By providing your customers with an excellent experience and top-quality products and services, you can increase retention, meaning your business will receive greater value from those customers over time. To drive profits and maximize your business’ success, it’s essential to understand customer retention, why it matters, and how to maximize it. ## Understanding Customer Retention Rate Your customer retention rate is the percentage of customers who continue paying for your business’ products or services over a specific period of time. For example, if you started the year with 1,000 subscribers and had 900 subscribers by the end of the year, your customer retention rate would be 90%. Customer retention rate is a key metric for many businesses, often used by those with recurring customers, such as those that sell subscription products or services, or businesses that repeatedly sell products to the same customers. A cleaning business offering weekly commercial cleaning services would find value in understanding the customer retention rate, and the same is true of a business like a big box retailer, where customers might shop weekly or monthly. Customer retention rates are particularly important to digital advertisers, since retaining existing customers is more cost-effective than continuously marketing to and attracting new customers. With high retention rates, digital marketers can use their advertising budgets wisely and get more value out of the customers they attract. ### How Does Retention Rate Relate to Customer Loyalty and Lifetime Value? Customer retention rate is closely linked to customer loyalty and lifetime value. A high customer retention rate suggests that your business also has a high customer loyalty rate, which refers to the number of customers who continue to support your business. It also indicates that many of the same customers are likely repeatedly buying from your business. Your customer lifetime value refers to the total value a customer will bring to your business, based on the average purchase value, the number of purchases per year, and the average customer lifespan. If you have a high customer retention rate, then more of your customers are buying from your business for a longer period of time. That drives up your customer lifetime value, meaning you’re making more money off of each customer during their relationship with your business. ### What Is the Formula for Calculating Customer Retention Rate? You can calculate your customer retention rate using a simple formula: ((CE - CN) / CS) x 100 - To start, determine the time period that you’re measuring, such as a month or a year. - Determine CE (Customers, Existing), which is the number of customers your business had at the end of that time period. - Determine CN (Customers, New), which is the number of new customers you acquired during the period. - Determine CS (Customers, Start), which is the number of customers your business had when the period began. - Plug those numbers into the formula and you can easily calculate your customer retention rate. Let’s say, for example, that you had 120 customers at the start of a six-month period, and 100 at the end, and that you acquired 25 new customers in the same period: 100 - 25 = 75 75 ÷ 120 = 0.625 0.625 x 100 = 62.5 So you have a customer retention rate of 62.5%. ### What Is Generally Considered a Good Customer Retention Rate? While there isn't a single "good" retention rate across all industries, sources(1, 2) suggest that an average customer retention rate is around 75-75.5%. However, a good retention rate can typically range from 35% to 84%, with the ideal figure depending on the industry. ### Good Retention Rates by Industry in 2025 - Automotive (Automotive & Transportation): A good retention rate for the automotive and transportation industry is around 76-83%. Factors like regular maintenance, service plans, and warranties contribute to higher retention. - Finance (Financial Services & Banking): Financial Services: Generally, a good retention rate for financial services is around 74-78%. - Banking: The banking sector tends to have a high retention rate, around 75%, due to long-term customer relationships and the effort involved in switching banks. - Health Services (Healthcare): A good retention rate in the healthcare industry is typically around 77%. - Home Services: While specific "home services" retention rates are not as broadly categorized, related industries offer insights: HVAC Services: Around 66%. - Construction & Engineering: Around 79-80%. - Generally, factors like online booking and excellent customer service are crucial for retention in home services. - Technology: IT Services: A strong retention rate in IT services is around 81-83%, often due to the continuous support needed. - Software (including B2B SaaS): The average for software is around 77%. For SaaS specifically, an average is closer to 68-74%, but a "good" rate is generally 70% or higher, with top performers reaching above 85%. - Travel (Hospitality, Travel, & Restaurants): This sector tends to have some of the lowest retention rates, around 55%. This is often attributed to price sensitivity and commoditized products. - Retail / Shopping / E-commerce: Retail (general): A typical retention rate for retail is around 63-67%. - eComm: The e-commerce industry generally faces significant challenges in customer retention, with average rates around 28-38%. A rate below 25% indicates a problem, while rates above 40% are considered excellent in e-commerce. ## Factors Influencing Retention Rate Many factors, from customer service to personalized engagement, influence your retention rate. Understanding and focusing on these factors can position your business for optimal customer retention. ### What Role Does Customer Satisfaction Play in Retention? Customer satisfaction is closely linked to retention. When your customers are satisfied, they’re more likely to make repeat purchases and remain loyal to your business, increasing retention. Satisfied customers may also refer their friends and family to your business, and word-of-mouth marketing can be valuable in attracting new customers who are well-aligned with your business. Your customer service and support can closely impact customer satisfaction, meaning they also relate to customer retention. Providing top-quality, highly responsive customer service can build trust in your business and ensure that customers have positive experiences. Prioritizing customer service and support, such as ensuring prompt responses and giving customers a way to easily reach a human, can ultimately impact your retention rate. ### How Does the Quality of Products or Services Impact Retention? When your business offers quality products or services, customers are more likely to want to buy from you again, which can boost retention. You can also demonstrate your commitment to offering quality services and products by asking for customer feedback about what you can do better. When you gather and then implement feedback to make improvements, you’re showing your dedication to customer satisfaction while also offering even better products and services. These strategies can also help build a strong brand and community. If your customers feel like valued members of your community, they may feel more loyal to your business, prompting them to make repeat purchases. ### How Can Personalization and Engagement Strategies Influence Retention? Personalization and engagement strategies can have a tremendous influence on customer retention rates. In today’s advertising world, where consumers are overwhelmed with messaging, personalized engagements stand out and are more likely to make a positive impression. Personalization goes beyond just incorporating a customer’s first name into emails and messaging, though: Segmenting your messaging so your customers receive highly relevant content, offering personalized loyalty programs tailored to customers’ individual preferences, and responding to inquiries with an email or message crafted specifically for the customer all contribute to a customer’s connection with your brand, which can positively impact your retention rates. ## Measuring and Analyzing Retention Rate As noted above, the formula to calculate your customer retention rate is ((CE - CN) / CS) x 100, but the process of measuring and analyzing your retention rate is more involved. It’s worth developing a process to track, measure, and analyze your retention rate, though, since this can provide you with valuable information about your business’ performance and how well you’re connecting with customers. ### How Do You Track and Measure Customer Retention Rate Effectively? To effectively track and measure your customer retention rate, measure your rates across specific and comparable time periods. For example, you might want to measure your rate across each quarter or year. Keep these periods consistent to accurately compare your retention rates. You also need detailed, accurate data to effectively measure customer retention. If you don’t have precise numbers, your calculations won’t be accurate. Tools and platforms that can accurately track your customers across time are essential in measuring your customer retention rate. ### What Tools and Analytics Platforms Can Help Monitor Retention? Customer analytics platforms can help you track essential data and monitor customer retention. These tools include customer success platforms, subscription and renewal management tools, product analytics tools, and customer communication platforms. When choosing the right tools and analytics platforms for your business, look for tools that can capture the exact type of data you need. Make a list of your key requirements and consider whether the tool can grow with your business. ### How Can You Segment Your Customer Base to Analyze Retention Patterns? By segmenting your customer base, you can better understand their preferences and behaviors contributing to retention. You can also examine the factors contributing to customer churn, i.e., the rate at which customers stop using your products or services over a certain period of time. Common reasons for customer churn include poor customer service, competition, a lack of engagement with your current customers, and even pricing problems that prompt customers to look to your competitors. When you segment your base and are able to identify these issues, you can take steps to fix them. To segment your base, you can sort customers based on factors like engagement levels, lifetime value, and behavioral patterns. Look for trends in your retention data among each segment, and determine which segments are more loyal to your business. You can also look for segments that are likely to churn so you can implement strategies to prevent it, like sending the segment special offers or targeted marketing campaigns. ## Strategies to Improve Retention Rate Once you know your retention rate, continuously work to improve it. Improving your customer retention can help boost sales and maximize the value you get from each customer. ### What Are Some Effective Strategies for Onboarding New Customers and Improving Initial Retention? To improve initial customer retention, personalize your onboarding experience. For example, if you’ve gathered data about new customers during your marketing and have segmented your leads, you can provide them with specific onboarding paths based on their background, product or service interests, and their previous interactions with your company. Don’t overlook the importance of a truly personal touch, too: A call from your customer service team checking in on a customer’s experience with your product or service after their purchase demonstrates your dedication to their satisfaction, and will make your company stand out. You can also use that call to gather customer feedback. By acting on that feedback and ensuring a positive customer experience, you can improve retention. ### How Can You Proactively Engage With Customers to Prevent Churn? By segmenting your customers, you can identify those who are at risk of churn, such as those who haven’t made a second purchase within a certain period of time, or those who left negative reviews of your business. Consider having your customer service team reach out to these customers personally. Loyalty programs and rewards can also proactively boost customer retention. For these programs to be effective, they need to offer real value to your customers, such as a steep discount or the option to apply rewards to the specific product or service each customer wants. Offering these programs early on in the customer’s journey can incentivize them to make future purchases and help prevent churn. ## The Relationship Between Acquisition and Retention While acquiring new customers is key to growing and expanding a business, customer retention is closely related and is also a top priority. Some customer churn is inevitable, so acquisition is necessary to maintain and grow your customer base. Maximizing your customer retention, meanwhile, can increase the lifetime value of your customers, so your customer acquisition efforts lead to greater profits. ### Why Is It Often More Cost-Effective To Retain Existing Customers Than To Acquire New Ones? Acquiring new customers takes time and can require significant financial investments, but retaining existing customers can be simpler and requires less resources. According to the Harvard Business Review, in fact, acquiring a new customer can cost 5-25x more than the cost of retaining an existing customer. ### How Does a Focus on Retention Impact Your Overall Marketing ROI? The same Harvard Business Review report points out that increasing your customer retention rates by just 5% can increase profits by 25% to 95%, highlighting the significant impact that retention can have on your overall marketing ROI. After investing in acquiring customers, working to prevent customer churn can increase the value your business receives from each customer, boosting your ROI and your profits. ### How Can Acquisition and Retention Strategies Work Together Effectively? A successful business needs to focus on building effective acquisition and retention strategies, since the strategies are closely paired. If you’re building a brand-new business, then you’ll need to initially focus more heavily on acquisition, but as you establish a customer base, you might shift more toward retention efforts. Think of your retention strategies as helping to preserve the value of the customers that your acquisition strategies have generated. There’s no universal approach that works for every business, and you may find that the balance shifts depending on the phases your business goes through. With time, though, you’ll find the ideal balance between acquisition and retention strategies to help your business thrive. ## Key Takeaways Your customer retention rate provides you with valuable information about how your business is performing and the experience that customers are having. Working to improve your customer retention can pay off with higher profits and greater customer lifetime values, maximizing the value you see from your acquisition efforts. Strategies like personalizing the customer journey, providing quality customer service, and offering a great product or service can all boost retention and help build your profits. ## Frequently Asked Questions (FAQs) ### What is the difference between retention rate and churn rate? Your customer retention rate refers to the percentage of customers who continue buying your products or services across a designated period of time. The churn rate refers to the percentage of customers your business loses over a period of time. ### How often should you calculate your retention rate? Ideally, you should calculate your retention rate monthly or quarterly. By frequently calculating your retention rate, you can promptly spot any fluctuations in the rate and can then identify potential causes. ### Is a high retention rate always a sign of success? A high retention rate suggests that customers are pleased with and loyal to your business, but that’s not the only indicator of success. If your business isn’t effectively attracting new customers, it will be difficult to grow it and increase profits, even if your customer retention rate is high. ### How can you use retention rate to forecast future revenue? If you know your customer lifetime value, multiply it by the number of retained customers to estimate the revenue you will receive from those customers. --- ### User Experience: UX and Its Impact Explained URL: https://www.taboola.com/marketing-hub/user-experience/ Last Modified: 2026-06-22 08:33:45 If you’ve ever tried to wade through a website that was overwhelmed with dense text, opened an app that you just couldn’t figure out how to use, or failed to complete a checkout because the website wasn’t optimized for your smartphone, you’ve experienced the effects of a poor user experience. User experience has a tremendous impact on everything from customer loyalty to your conversions, and it’s a key consideration in your marketing and sales strategy. Understanding how to improve the user experience can help your business provide prospects with experiences that are memorable for the right reasons, increasing the chances that they’ll become loyal customers. ## Understanding User Experience User experience, or UX, refers to the quality of a person’s interaction with a product, service, or resource. If the user’s interaction is a positive one, and they’re able to easily use the resource or product and are satisfied, then the user experience is positive, or good. A bad user experience can negatively impact a customer’s impression of a business, product, or service, and it could cost your business a sale or a lead. ### What Are the Key Principles of Good UX? Several key principles contribute to good UX: - User focus: A product or website needs to be designed with the end user in mind. Research that person’s needs and goals, and then test your product to identify and resolve issues that frustrate those users. - Usability: A product needs to be easily usable, meaning a user should be able to accurately and effectively engage and complete actions with the product with minimal difficulty. - Visual hierarchy: When designing products like a website, visual hierarchy refers to the design elements that lead a user through the page. Features like color, contrast, and scale can help a user easily navigate and interact with your website. - Consistency: Consistency is important to UX. Focus on creating design elements that feel familiar to users for a more welcoming, comfortable experience. Users who are already familiar with elements of your product will be able to learn and use it more quickly. - Accessibility: A product needs to be accessible to as many users as possible. It should be designed to accommodate the needs of individuals with disabilities, as well as for those without. - Context: Context refers to how your product will be used and any factors that might affect its use. For example, when designing a website, you’ll need to consider the different devices that users might choose to access the website. In this case, ensuring the website is optimized for mobile devices is paramount to the end-user experience. ### How Does UX Impact User Satisfaction and Business Outcomes? When a business creates a good UX, it can improve user satisfaction. Being able to easily and successfully navigate a website, access the information needed, and seamlessly make a purchase leaves a user with a positive impression of the business, increasing the chances that they will return again to make another purchase. In contrast, poor UX can negatively impact user satisfaction. If a user can’t find the information they need on a website or encounters numerous glitches while trying to check out, their satisfaction drops. They may not complete the purchase at all, and if they do complete the purchase, they may not return to make another purchase because of their frustration with their initial experience. ## Key Elements of User Experience Design Many factors contribute to successful user experience design. It’s important to consider how everything from information architecture to visual design contribute to the overall user experience. These elements all need to work together for an optimal UX. ### What Is Information Architecture and Its Role in UX? Information architecture is the process of organizing information in a logical way so that it’s easy to read and digest. Information architecture involves the content organization, or structure, as well as the process of labeling content with clear classifications. For example, when designing a website, information needs to be presented clearly so that readers can find the information they’re looking for without having to read through the entire site. Content that’s presented illogically can quickly frustrate and confuse website visitors, prompting them to leave the website without finding what they were looking for. ### What Is Interaction Design (IxD)? Interaction design is the process that makes websites and apps into products that users can interact with. Rather than giving a user an app full of text that they can read through, interaction design creates a dialogue between the user and the app. For example, a menu would allow the user to navigate through different sections of the app, and an artificial intelligence (AI) chat function could help the user find the information they want. Interaction design is closely linked to UX. If the interaction design of a product like a website or app is poor, the product’s usability is limited and can be frustrating for a user. A product with quality interaction design provides a more engaging experience, which is easier for users to navigate. ### What Is Visual Design and Its Impact on UX? Visual design, which encompasses a product’s color, layout, whitespace, font, images, and more, is about more than a product’s visual appeal: Done strategically, the visual design of a product like a website can enhance the user experience, since it can improve a website’s readability and usability. This, in turn, can impact a user’s emotional experience while using the site, while helping guide the user for an interaction that’s logical and easy. ### What Is Usability Testing and Why Is It Crucial? Usability testing is the process of evaluating how usable a product is. By observing users as they navigate a product like a website, this testing process can help you identify issues that need to be fixed, and it can tell you how well the site meets users’ needs and expectations. Testing your website before you fully launch it can help ensure you’re delivering an optimal UX experience, so you don’t risk unknowingly losing potential customers to a poor experience. ### How Does Content Strategy Contribute to a Positive UX? Your content strategy, encompassing planning, creating, and managing your content, can directly affect the UX of a product. By strategically planning and developing your content, you can ensure that you’re providing content that meets your users’ needs. For example, if you’re marketing insurance products, developing comprehensive, well-organized, and easy-to-read content around relevant insurance topics can help your users get answers to their questions so they’re better prepared to make a purchase. Quality content helps users to feel supported and understood, contributing to a positive UX. ## UX in Digital Advertising UX is essential in digital advertising, and it can contribute to the success or failure of advertising campaigns and businesses. In digital advertising, UX refers to a user’s interaction with digital products like websites, apps, digital ads, social media, and more. ### How Does UX Apply to Landing pages and Ad Experiences? Ad designers need to consider UX during the entire ad design process. An ad with poor UX may be confusing or difficult to read, or its messaging might not apply to the target audience. All of those factors mean that viewers are unlikely to click on or engage with the ad. But, done well, an ad that’s logical, clear, engaging, and designed with the target audience in mind can increase viewer interactions. Well-designed ads are more likely to bring users to a landing page, but UX is equally important to the landing page’s performance. A landing page needs to clearly prompt users to take a desired action, like subscribing to a newsletter, and it needs to make the case for why they should take that action. The landing page must be visually appealing, logically organized, easy to read, and easy to use. If it’s lacking these qualities, site visitors will have difficulty or won’t sign up for the newsletter, and you’ll lose out on leads. All in all, the landing page UX experience can significantly impact the success of your marketing. ### How Can Poor UX on a Landing Page Negatively Impact Ad Performance? Poor UX on a landing page can cost you leads that your ad generated. If the landing page is confusing or if there are usability issues with a submission or signup form, you won’t be able to capture lead contact information. Your ads might be generating click-throughs, but your landing page bounce rate may be unusually high, signaling lost lead opportunities. ### What Are Some UX Best Practices for Advertising Creatives? It’s important for advertising creatives to focus on simple, easy-to-read ad designs. Consider how well messaging shows up against backgrounds and carefully choose fonts and colors to ensure the ads are easy to read. Present information in a clear, logical way. Choose images that reflect the messaging and that make sense when paired with the content. Keep business branding principles in mind, and ensure that messaging and branding is consistent across all ads and platforms to avoid confusion. Ads need to be easy to follow, and they should lead to a landing page on a website or app that’s consistent with the ad’s messaging. The landing page needs to present a clear and appealing value proposition, and the call-to-action should be easy to identify. Don’t forget to test out any form submission processes to make sure that it’s easy for leads to sign up for an offer. ## Measuring and Improving User Experience Since UX is so integral to everything from lead generation to conversions and sales, it’s important to continuously measure and improve UX. By working to improve it, you can quickly identify any issues that arise and maximize the value you get from the leads you’ve generated. ### How Can You Measure the Effectiveness of Your UX? Data about how your users interact with products like your website or app can help you determine how to improve that experience. Usability testing, in which you watch users interact with your website, can help you learn about what may be confusing, which tasks are difficult to perform, and what areas in your site need to be improved. Surveys can be another valuable source of information. For example, after a customer makes a purchase, you could follow up with a brief survey asking about their experience. Don’t forget to offer an incentive for completing the survey, such as a discount off a future purchase. Ask questions about how easy it was to find the product, whether the customer could find the information they needed, how simple the checkout process was, and what they would change about the website. Your website analytics also provide a snapshot of your site’s UX. For example, if certain pages have an unusually high bounce rate, look carefully at those pages to see if there might be a usability issue. If you have a high cart abandonment rate, you might need to troubleshoot your checkout process to identify potential problems that are causing shoppers to abandon carts instead of completing the checkout process. ### What Are Some Key Metrics to Track Related to UX? Tracking certain metrics can alert you to potential UX issues, so you can then work to fix them: - Bounce rate: A high bounce rate — the percentage of users who leave your site after viewing only one page — may indicate a UX issue with a webpage. The page may be confusing, difficult to navigate, or users might arrive on the page incorrectly because of a linking error. - Conversion rate: A low conversion rate — the percentage of users who take a desired action, like signing up for a newsletter — can signal issues with a landing page. The landing page may be confusing, the offer unclear, or the signup process difficult. - Average time on page: A low average time on page — the amount of time a user spends on one webpage before they navigate to another page — can signal issues with your webpage usability. Users will often quickly navigate away from webpages that are difficult to read, confusing, or that are slow to load. - Shopping cart abandonment rate: A high cart abandonment rate, or an abandonment rate that suddenly increases, can indicate that your checkout process is confusing or difficult. ### What Is the Iterative Process of UX Design and Optimization? The iterative process of UX design encompasses a cycle of refining your product’s UX. The process begins with gathering feedback, such as from customer surveys or usability testing, to identify areas to improve. From there, you will need to create solutions, test their performance, and refine and evaluate them further. Then, the cycle begins again. ## The Relationship Between UX and Other Disciplines ### How Does UX Relate to UI (User Interface) Design? User interface design is a component of the overall user experience. UI design encompasses a product screen or website’s look, feel, and interactive features. A good UX design is dependent on a good UI design. ### How Does UX Intersect with SEO? UX can affect your SEO performance. For example, ensuring your website is mobile-friendly is key to a good UX, but it may also help improve your Google SEO ranking. The same is true of your website page load speed; a fast-loading website offers a better UX, and Google tends to rank pages that load faster higher. So, ensuring a good website UX may also boost your overall SEO results. ### How Does UX Impact Conversion Rate Optimization? UX directly impacts your conversion rate. If your website is highly usable, is well-organized to help users find the information they want, has a well optimized call-to-action, and the site’s appearance makes it easy to navigate, users are more likely to convert. A poor UX caused by poorly organized copy, a site that is difficult to view and understand, and a confusing call-to-action can negatively impact conversions and lead to a lower conversion rate. ### How Does UX Contribute to Overall Customer Experience? UX is tightly linked to the overall customer experience. In fact, the customer experience often begins when they navigate a website or an app. Think of UX as a component of the customer experience, which can also encompass the customer’s overall impression of your business, including the support they’ve received before and after a purchase. ## Key Takeaways User experience impacts customer loyalty and satisfaction, and it directly affects key metrics like conversion rates and bounce rates. Additionally, user experience is closely linked to other marketing factors, like your SEO and the overall customer experience. Since user experience is so paramount to marketing and sales, it’s important to continuously assess, test, and improve your products, like your website and app, to create the best user experience possible. ## Frequently Asked Questions (FAQs) ### What is the difference between UX and UI? While UX encompasses an individual’s entire interaction with a product like a website, UI is a more focused subset of UX. UI focuses only on the look, feel, and interactive features of a product screen or a website. ### How can I improve the UX of my website? To improve your website UX, you’ll need to identify the areas that aren’t performing well. Consider surveying your visitors, reviewing your site’s analytics, and implementing usability testing to determine what you need to improve. From there, you can design, test, and implement solutions. ### What are some common UX mistakes to avoid? Avoid common UX mistakes like overloading users with content that’s difficult to scan. It’s easy to make apps and websites overly complicated, which can make them hard to understand or navigate. Additionally, some companies make the mistake of ignoring user feedback or research, and the result is a poor UX. ### How much should I invest in UX research and design? There is no one magic formula to determine how much to invest in UX research and design, since every business is different. When deciding what to invest, consider your budget, and also consider how your UX will affect other components of your marketing and sales efforts, like your SEO ranking and your conversion rates. ### What are some key UX principles to always keep in mind? Always look for ways to make your platforms simple, clean, and inviting. Keep your audience in mind as you design your platforms and craft content, and spend plenty of time testing for usability. --- ### Marketing Technology Trends for Advertisers to Know 2026 URL: https://www.taboola.com/marketing-hub/marketing-technology-trends/ Last Modified: 2026-03-08 13:47:40 In 2026, the practice of marketing — automated, personalized, and fast-moving — can only happen successfully with the right technology in place. The basic tenets of good marketing don’t change much from year to year, but strategies and tactics are both shaped by and influence the technology choices that marketers make. Software options abound and can easily overwhelm a marketing team, even as teams are being asked to reach prospects and demonstrate their contribution to sales more than ever before. ### What’s changed in our 2026 update: - New trends added. - All entries include updated and current information, advice, and stats. - Data trends based on Realize data added. ## The MarTech Landscape in 2026 The global marketing technology market size is estimated to be worth a little under $6 billion now, and is projected to reach $2,380 billion by 2033 — a huge compound annual growth rate (CAGR) of 20%. AI and ML are driving this, as well as demand for marketing automation, the need for personalization, a shift toward omnichannel strategies, and the desire for real-time customer insights, among others. While retail and e-commerce are leading the charge, growth is strong among all major categories. There are a few broad trends in marketing technology in 2026 to shape planning and purchasing for success: ### Data-driven The tools now exist for marketers to collect, refine, analyze, and incorporate tons of data into their marketing tactics. There’s no reason not to make better decisions using all that data, instead of relying on gut instinct or outdated ideas or methods. ### Customer-first Putting the customer first is a tried-and-true tenet of marketing, but technology advancements make it possible in 2026 to truly reach and engage each customer in a cohesive way, using personalization techniques and tools like agentic AI, first-party data, and predictive analytics. ### Unified tools After a few decades of software development, many marketing teams are struggling to get the most out of a mess of disparate legacy technologies. For many, this is the year they’ll unify and simplify the martech stack. What else to know: - The global marketing technology market size is expected to reach $2,380 billion by 20333, a CAGR of 20% from 2025. - The social media tools segment led the market in 2025, making up more than 23% of the global martech revenue. ## 5 Marketing Technology Trends for A Better Strategy in 2026 If marketing technology in 2026 seems overwhelming, it’s not only because your team is strapped for resources, the budget is tight, or you’re being asked to tackle too many challenges at once: It’s also the sheer volume of the marketplace. The most recent Marketing Technology Landscape report counted 15,384 solutions, up 9% from the previous year. (For some perspective, in 2011, there were just 150 solutions available.) So, where should a marketer start in identifying and then implementing trending technologies? These following trends can bring a ton of value, and each business will approach adoption their own way based on goals, industry, and budget. - ### Hyper-Personalization at Scale The convergence of brand importance, AI, big data, and machine learning have led to hyper-personalization as key for modern businesses. Consumers are moving quickly, taking non-linear user journeys across multiple channels per day. It’s essential for marketers to deeply understand their audiences by collecting the right data, segmenting audiences accordingly, and reaching them on their preferred channels with the right messaging and offers. Using AI-driven technology helps brands to hyper-personalize down to the customer, predicting customer behaviors based on real-time and past data. Marketers can create product recommendations on a landing page, send carefully timed emails during the buying process, and much more. Precise targeting leads to better customer engagement, more effective marketing campaigns, and higher ROI, and understanding customer behavior trends can help marketers make changes quickly and increase conversions. But, solid technology is necessary to do all of this at scale. Here’s what else to know: - 81% of consumers ignore irrelevant marketing messages, but 96% are likely to purchase when brands send personalized messages. - 92% of businesses are using AI-driven personalization to drive growth. - 56% of consumers say they’ll become repeat buyers after a personalized experience. ### 2. Maturing AI It’s ubiquitous in technology conversations in 2026, and for marketing technology in particular, AI offers a ton of opportunity. These include direct prospect or customer interaction, such as in chatbots or customer service agents, as well as workflow process automation. Marketing teams are also starting to hone how they use AI for creative work to scale. AI is generally used for marketing in at least one of these ways: - Generative: Tools like ChatGPT can help marketers brainstorm copy, as well as create imagery and videos. Using AI can help marketers and advertisers scale quickly, testing variations and ensuring human-friendly creative outputs. - Data work: LLMs offer marketers a way to query datasets in a conversational way, allowing them to incorporate more data into strategy work. Data can be presented simply, showing anything from average ad spend to campaign performance by channel. - Targeting: AI enables hyper-personalization tactics which, at scale, would likely not be possible for most marketing teams in terms of time and resources. - Automation: Marketing automation tools often incorporate AI to make it possible to remove the burden of manual tasks from teams, such as automating monthly processes or setting up workflows for repetitive tasks. - Customer service: AI chatbots have become commonplace as a way for marketing teams to connect with customers at any time without adding to or using tight resources. - Predictive: With data analytics technology and ML models in place, marketers can forecast results based on detailed past data. Predictive AI technology can offer a huge boost to marketing teams, especially in times of uncertainty, to make better decisions about their strategy. Here’s what else to know: - 60% of marketers say AI and ML will have the biggest impact on marketing strategies in the next five years. - Marketing teams using AI report 44% higher productivity. 58% of marketers now use AI for content ideation and optimization. ### 3. First-Party Data and Privacy-Centric Marketing First-party data has become more essential, and more valuable, in understanding prospects and customers. First-party data comes directly from customers, so it’s more accurate, and collecting first-party data helps businesses stay compliant while building strong, trusting customer relationships. It’s key to privacy-centric marketing, which customers demand and regulatory bodies insist upon. There’s a lot of opportunity to build brand and customer loyalty when gathering this data, as well as using it for better personalization alongside other technology like predictive analytics. A single source of enterprise truth for data — like a customer data platform (CDP) — is the foundation for using customer data wisely. Successful marketers have to make sure that data is continually fresh and integrated, with a company-wide strategy for using data to reach and engage customers. That way, customers will only get the right messages at the right time in their buying journey, and marketers won’t waste time on manual data work. Poor data quality can lead to frustrated customers and unclear campaign results, and it’s also bad for AI success. Here’s what else to know: - 86% of companies recognize the importance of first-party data. - 64% of U.S. customers say they’d provide their email address for a $20 discount, and 31% would share their full name. - 60% of gen AI projects were abandoned due to poor data quality or unclear business value. ### 4. The Convergence of MarTech and AdTech As part of building the data foundation, technology (and the teams using it) has to be integrated and synced across the business. There are many ways to annoy and lose customers with disparate technology siloes — sending multiple unrelated emails in one day, the purchase and return departments unable to communicate, and much more. Many of these common issues stem from separate martech and adtech stacks, whether it’s from lack of internal communications, legacy software tools not working together, or both. Martech generally refers to the technology and tools that marketing uses, such as a CRM, email marketing, social media marketing, etc. Adtech refers to the tech used for advertising and buying media, including any digital campaigns and optimizing spending. Bringing these platforms together will be essential for AI success, data quality, and customer experience, as well as other point solutions that may exist. The customer experience should be the gold standard for teams building unified stacks, and combining team goals to get to that excellent experience can then help the wider business meet its goals. And, perhaps most importantly, a unified platform saves a lot of wasted budget and resources internally. Privacy concerns are top of mind in both adtech and martech in 2026, with 73% of consumers saying they’re more concerned about their data privacy now compared to a few years ago. Here’s what else to know: - The global ad tech market is projected to reach $1,580 billion by 2030, with a growth rate of 14%. - Nearly 29% of U.S. ad agencies said they used six to seven ad tech and martech tools as part of their tech stack, and 17% used more than 10. - Disconnected or misaligned ad tech and martech tools can lead to a 10-13% loss in resources. ### 5. Voice Search and More Smart speakers like Amazon’s Alexa and iPhone’s conversational assistant Siri are commonplace among consumers. With these tools established, alongside a robust podcast industry, marketers should consider voice search and commerce as a tactic in their strategy. This may include optimizing content for voice search, using long-tail keywords and natural language. Voice commerce includes the field of audio marketing, which may include podcast sponsorships or streaming service ads, depending on the product, audience, and industry. As with any of 2026’s marketing trends, marketing teams should use all the data at their disposal, target appropriately, and conduct testing to see what works on this channel. More broadly, businesses that are willing to experiment and disrupt within their category have seen big gains in the past few years. Technology can play a huge role in disruption and reaching consumers in new, interesting, and authentic ways. Here’s what else to know: - 32% of consumers use voice assistants weekly. - 74% of voice assistant users have completed at least part of a buying process through conversational AI. - 71% of consumers would rather use voice search than manually typing queries. - Brands willing to disrupt themselves or their category have created $6.6 trillion in value over the past 20 years. ## Key Data Trends From Taboola’s Perspective in 2026 Taboola’s Realize platform brings together AI technology, modern marketing techniques, and data-driven strategy. We’ve seen a few key trends emerge lately: Human-created and AI-created ads work best in tandem. AI-generated creative work is a lot more common, but humans are much more aware of it, too. A recent study found that AI-generated ads either perform as well as or outperform human-generated ads, with Realize’s AI technology incorporating creative best practices accordingly. Modern ad success requires data-driven strategies. We’ve found that Realize users who focus on the outcomes up front can make big gains with prospects, especially on the open web. For example, choosing to maximize conversions vs. maximize value before launching a campaign brings a lot more clarity, and specific, action-oriented data to analyze and act upon. Iterating quickly for efficient scale is essential for modern advertisers. Advertisers have to work at massive scale. It’s not enough for advertisers and performance marketers to create and deploy ad campaigns at a pre-generative AI scale. At Taboola, the numbers show how fast and how massive businesses need to run to attract new users: The Realize platform trains 400 AI models, processes about 2 PB of data, and makes 181 billion predictions each day. ## Key Takeaways Marketing technology reflects the bigger tech landscape in what’s changing the game, with AI, user focus and privacy, and rich data all in the spotlight. Whether teams are building their tech stack from scratch or phasing out legacy tools, customer data should be the focus — capturing it safely, using it responsibly, and making data-driven decisions to succeed. ## Frequently Asked Questions (FAQs) ### Where is MarTech headed in the next five years? The next five years offer lots of room for growth and experimentation for marketing technology. Agentic AI, ensuring customer privacy while capturing data, strong data foundations, and hyper-personalization will all continue to mature as marketers access more sophisticated technology and use continuous testing to understand what customers respond to. They’ll need to strike a balance to continue to deeply understand customers while using tools like AI to scale quickly. ### What are the trends in voice search optimization for marketers? Voice search optimization is on the rise, with big implications for traditional SEO practices. Voice search is more conversational than text-based searches, so marketers can use long-tail keywords and make sure they’re optimizing for natural language and the short questions users may be asking their phone or smart speaker. As this segment becomes more commonly used, marketers need to define and track the metrics that fit voice search usage. ### How is generative AI being used in marketing? Marketers are using generative AI to quickly create, test, and iterate on messaging, imagery, and videos, particularly as technology like Realize includes AI that takes into account human preferences. Gen AI offers opportunities for personalization at scale and streamlined workflows, such as removing manual work for email sequences, as well as conversational data analysis to quickly understand campaign performance. --- ### Leads: A Comprehensive Guide for Digital Advertisers URL: https://www.taboola.com/marketing-hub/lead/ Last Modified: 2025-08-21 14:06:48 Leads are the foundation of your business growth. If you want to expand your customer base and increase your sales and profits, you’ll need a steady supply of new leads. Understanding how to generate quality leads for your business is just the first step of the process; from there, you’ll qualify, nurture, and convert leads. Businesses that do so successfully can grow significantly and are well positioned for success. ## What Is a Lead? A lead is an individual or organization who has expressed some sort of interest in your business. The lead has the potential to become a customer, and they may have demonstrated their interest in your business by visiting your website, signing up for a free trial, or submitting contact information in exchange for a piece of gated content, like an e-book or guide. By generating and nurturing leads, your business can potentially convert them into paying customers, growing your customer base and increasing your revenue. ## How to Generate Leads for Your Business Lead generation, which is the process of engaging with and drawing in potential customers, is the foundation to your lead nurture and conversion processes. Effective lead generation can supply your marketing and sales teams with a steady stream of leads, helping to grow your business. There are many ways to generate leads, so I’ve highlighted some of the most popular options that work for most businesses. Effective lead generation depends on providing your audience with value to foster a meaningful connection and build their interest in your business, and you’ll see that providing value is a common theme in these strategies. ### Invest in Search Engine Optimization (SEO) When you optimize your website for search engines, you can better drive potential customers to your site. By incorporating keywords that potential customers are likely to search for in your website content, you can position your business and products as being a solution. Creating blog or website content around those keywords and answering common questions potential customers are likely to have can help solve their problems while introducing them to your business. ### Share Knowledge Through an Article or Blog Related to the above strategy, write articles or blogs that offer real value to your potential customers. Use your specialized knowledge to solve a common problem, answer questions, or provide important “insider information” or insights that your potential customers would find interesting. Focus on creating quality pieces of content that readers are likely to engage with and share, rather than on simply posting lots of blogs. Consider writing an article or blog for another business or another publication, too. By publishing content for another business that shares a similar target audience, you can expand your reach and generate new leads. ### Host a Webinar Webinars can be an excellent way to generate quality leads. Similar to blogs and articles, webinars allow you to share your knowledge and expertise with leads, but you can often do so in more depth with a webinar. You can also bring in other industry experts on panel webinars to enhance the value your audience receives. Require webinar attendees to submit their contact information to register, and you’ll build a pipeline of leads. ### Engage With Leads on Social Media Social media is also an effective way to generate leads. When you establish an active social media presence for your business, you can reach leads in several ways: By posting engaging and informative content, leads may interact with your posts, and you can respond to comments and start to build relationships. Social media ads can also reach a highly targeted audience, driving leads to your site, encouraging them to sign up for webinars, and more. ### Generate Referrals From Current Customers Your current customers are a valuable source of leads, too. Your customers likely have friends or connections who would also be interested in your business and your products or services. Establish a referral program to incentivize your customers to refer others to your business. In exchange, offer your customers discounts or rewards for their referrals. ## How to Qualify Leads Leads aren’t all the same quality, meaning some leads are more likely to convert than others. Qualifying leads refers to the process of determining which leads most closely align with your ideal customer profile and are therefore most likely to convert into customers. Lead qualification is important because it allows you to focus your nurturing efforts and resources on the highest quality leads. In doing so, you can increase your conversion rate and use your time and financial resources wisely. As you generate and collect leads, you’ll need to qualify them through several steps. To do so effectively, make sure that you’re collecting and organizing leads in a CRM. ### Create an Ideal Customer Profile You’ll need to build a profile of your ideal customer, which will help you identify which leads are strongest. Your ideal customer profile can also inform and better focus your marketing efforts, so it’s worth investing time in thoroughly developing the profile. To create an ideal customer profile, outline the features of your most valuable customers, including their demographics, pain points, goals, industry, and more. If you’re running a B2B company, then details like the customer’s business revenue and position are also important. Record all of this information; you’ll reference it during the lead qualification process. ### Determine the Criteria You’ll Use to Score Leads Work with your analytics and sales team to identify the criteria that are most important in valuable leads. Your goal is to create a checklist outlining these criteria, so you can go down the list and determine the lead’s value. The criteria you use will be unique to your business and your customers, but consider looking at features like: - Geographic location. - Engagement with your social media pages. - The number of times they’ve visited your website. - Whether they’ve downloaded any of your content. - Whether they’ve been referred by a current customer. - Any conversations they’ve had with your sales team. - Whether they’ve requested additional information or a demo of a product or service. ### Research Leads Depending on your business structure and industry, researching leads may be an important part of the qualification process. It’s more common for B2B businesses to research leads, including looking for details such as their business structure, revenue, and their purchasing or decision-making authority within their business. The information you learn can better inform your lead qualification. ### Contact Leads and Evaluate When your sales team contacts the leads, those initial conversations can provide valuable information about each lead’s quality. Many B2B businesses hold initial phone conversations, giving your team a chance to ask questions about a lead’s pain points, budget, experience with your competition, and where in the buying process they are. All of that information can inform your evaluation of the lead quality, and should be recorded in your CRM. ### Qualify Your Leads Once you’ve gathered information on your leads, evaluate them using the scoring criteria you identified. You’ll also want to incorporate any notes from your team’s initial conversation with the leads. Leads that meet more of your criteria are your higher-quality leads and should be prioritized and moved along in your nurture funnel to receive personalized interactions. Lower-quality leads are of less priority, and these should receive more automated, less personalized interactions. ## How to Nurture Leads By nurturing leads, you can provide them with more information about your business, build important relationships with them, and eventually convert them into customers. The lead nurture process looks different for every business, but you can build your nurture process around these best practices: ### Plan Your Nurture Funnel Plan out what your lead nurture process will consist of, including the communication channels you’ll use, the offers you’ll make, the type and timing of your messaging, and your lead nurture campaign goal. ### Personalize Your Interactions Make your interactions with engaged, quality leads as personal as possible. Take detailed notes about your interactions with them and use the information you learn to customize your future interactions. Personalizing your content and messaging can not only make it more meaningful and memorable for leads, but also demonstrates that you’re listening to what they’re sharing during conversations and showing you’re able to meet their specific needs. ### Use Automation While your quality leads require personalized interactions whenever possible, you can still nurture lower-quality leads by using automation. Schedule emails containing pre-written messaging and content to go out to these leads at designated intervals. While not as powerful as a personalized nurture campaign, automated funnels can still potentially nurture leads and may generate some conversions. ### Use A/B Testing Continuously use A/B testing to improve your lead nurture efforts. A/B test each part of your funnel, like your offer language or the emails sharing your new blog post. Don’t be afraid to try out new techniques, too, and see if they can improve your nurture funnel. ## How to Convert Leads Your lead nurture funnel helps you learn more about your leads while also building your leads’ familiarity and connection with your business. These steps can help you with your ultimate goal at the end of the funnel, which is to convert leads. ### Try Different Conversion Tactics Lead conversion is a bit of an art, so try out different conversion techniques. Encourage your sales team to suggest new tactics based on their conversations with leads and what they think the leads want or need most. A new tactic could revitalize your lead conversion process. ### Make Appealing Offers An irresistible offer can help convert leads. Offer a discount, a free add-on, a heavily discounted package, or another incentive to sweeten the deal. Consider having a second, even more enticing offer that you can provide if your initial offer isn’t quite enticing enough to convert a lead. ## SQL vs. MQL vs. Prospect The terms prospect, Marketing Qualified Lead (MQL), and Sales Qualified Lead (SQL) all describe various phases of a lead’s progression through the sales process. Understanding how these terms are linked can help you tailor your approach to each lead to increase your chances of conversion. Prospect MQL SQL Stage of the lead qualification process Early Early/mid Mid Qualification A prospect is a lead that your team has qualified; you’ll know who they are, their role, and their business. An MQL is a lead who has engaged with your content, such as signing up for a newsletter or downloading a piece of gated content. A SQL is an MQL who has been qualified by your team, based on more than just their actions on your website. Additional qualifications might be based on information like their pain points and professional role. Which businesses use this term “Prospect” is commonly used by sales-oriented businesses. Marketing-oriented businesses commonly use this term. Marketing-oriented businesses commonly use this term. The Marketing/Sales Conflict: Should We Stop Generating Leads That Aren’t Sales-Qualified? Sales-qualified leads are highly desirable leads, since your team has identified features that indicate they’re ideal for your business. Still, that doesn’t mean that you shouldn’t generate leads that aren’t sales-qualified leads. For example, a marketing qualified lead, which has engaged with your content, has demonstrated interest in your business. With proper nurturing, that lead could turn into a sales-qualified lead, and then ultimately convert into a sale. Develop a nurture strategy that prioritizes your sales-qualified leads, but don’t ignore your other leads, which could become more qualified. ## Key Takeaways Leads are key to growing your business, expanding your customer base, and increasing profits. Refining your lead generation, qualification, and nurturing processes can help you to drive sales and convert leads into new customers. Testing and continuously improving these processes will help maximize your business and revenue growth. ## Frequently Asked Questions (FAQs) ### What qualifies as a good lead in marketing? A good lead is a lead that closely aligns with your ideal customer profile. For example, a lead that shares key demographics, geography, income, pain points, and interests with your current top customers would be considered a good lead that is likely to convert into a customer. ### What is scoring? Lead scoring is a process that helps businesses to evaluate the quality of their leads and determine which ones to prioritize in the lead nurture process. In scoring leads, businesses create formulas that allow them to assign numerical values to each lead. Then, businesses can rank the leads from the highest score to the lowest score, with the highest-scoring leads being the most likely to convert. ### How do I create an effective lead nurturing email sequence? To create an effective lead nurturing email sequence, keep things simple: Focus on one topic or idea per email, and address your leads’ pain points or interests. Keep the emails short and format them so they’re inviting and easy to read. Personalize them as much as possible, and use A/B testing to improve your email performance. ### What tools can help me track and analyze lead behavior? There’s no shortage of tools that can help you track and analyze lead behavior, including CRMs, social media analytics, Google Analytics, and various marketing automation and analytics platforms. To get the most value out of the tools you use, look for tools that offer plenty of customization options, which will allow you to tailor their use to track and analyze the data that matters most to your business. ### How do small businesses generate quality leads? When it comes to generating quality leads, small businesses can stand out from the competition by leveraging their brand stories and the personal connections they’re able to form with their audience. Referrals from current customers are a valuable source of leads, but small businesses can also leverage content marketing, performance-based display advertising, social media, and SEO to generate quality leads. --- ### Pay Per Lead (PPL): The Advertiser's Guide URL: https://www.taboola.com/marketing-hub/pay-per-lead/ Last Modified: 2025-06-09 12:28:44 In digital advertising, it’s not always enough to get clicks or impressions — most businesses want sales. Enter pay per lead (PPL). This performance-based advertising model enables companies to pay only for qualified leads. In this guide, I’ll cover everything you need to know about PPL — what it is, how it works, and how it differs from pay per click (PPC). ## What Is Pay per Lead (PPL)? Pay per lead (PPL) refers to an advertising model where advertisers pay only when a customer lead is generated. That’s because the goal of a PPL campaign isn’t just to drive traffic, but to acquire potential customers. Different actions can be considered leads, like filling out an online form, signing up for a free trial, or requesting a quote. ## What Is Pay-per-Lead Marketing? Pay-per-lead marketing is closely related to PPL. However, while the latter is simply a pricing model, PPL marketing refers to the strategy behind PPL that advertisers use to drive leads through performance-based marketing campaigns. With pay-per-lead marketing, you need to set clearly defined criteria for what constitutes a lead. You can do this by using contact forms, downloads, or sign-up events, such as webinars or appointment bookings, which are easy to track. The key is to be efficient — you want every dollar you spend to be tied to a measurable outcome. This way, you can calculate your return on investment (ROI) and customer acquisition costs. ## Why Is PPL Important? PPL can reduce wasted ad spend because you’re only spending money on leads — users who have taken a concrete step towards becoming a customer. In other words, you only pay when someone has shown genuine interest in your product or service. This can also make it easier to scale your campaigns, because there is less guesswork about your users, so you can be more confident in the outcomes. ## Pay-per-Lead Channels PPL campaigns can be run on many different channels. The challenge is knowing which platform is the right one for your product or service, and your target audience. Here are some common PPL channels: ### Search Engine Advertising Most search engines, like Google or Bing, operate on a pay-per-click (PPC) model. However, you can run PPL campaigns on these platforms by tracking what happens after the user clicks. To do this, you need to determine what will qualify as a lead — an online form submission, or a request for a quote, for example — and then use Google Ads’ or Bing Ads’ conversion-tracking tools to optimize for leads. ### Affiliate Marketing PPL is very popular with affiliate marketers, who make money by promoting a business' product or service and driving sign-ups. For example, an affiliate for a software company would earn a commission if a user clicked on an affiliate link and signed up for a free trial of a software product. Even though the client hasn’t bought anything yet, a lead has been generated, increasing the likelihood of purchase. ### Social Media Advertising Most social media platforms allow you to structure your campaigns to generate leads and optimize for cost per lead (CPL), even though they operate on a PPC or cost-per-impression (CPM) model. For example, in this case study, PPL Labs generated over 200 leads for Disc Centers of America using Facebook Ads: ### Email Marketing While email marketing is typically designed to build relationships with potential leads over time, email campaigns can also drive users directly to lead generation forms or sign-up pages, especially when a lead magnet is used, such as an e-book or webinar. ## Pay-per-Lead Models Here are some common ways in which PPL campaigns are structured: - Pay per qualified lead: This model aims to identify leads that are most likely to become customers. Leads must meet clearly defined requirements to qualify and for payment to be generated. - Pay per appointment model: This approach aims to generate appointment bookings, whether for a consultation or a product demonstration. You can agree to pay for a booked appointment or a completed appointment. - Pay per download: If your goal is to drive conversions of a lead magnet, consider paying per download. It’s easy to measure the performance of downloads, but be warned, some users may just be looking for a freebie with no intention of making a purchase. - Pay per attendee: If you’re holding a webinar, you can pay based on the number of attendees. That said, webinar guests may be lower-quality leads and not as ready to buy as someone prepared to book a 1:1 appointment or sign up for a free trial. ## Advantages of Pay-per-Lead Marketing PPL marketing can be an effective way to drive business growth when run correctly. Here are some benefits to PPL campaigns: ### Cost Efficient Since you only pay when you receive a lead, your budget can stretch further. This method is ideal for smaller businesses or businesses with a limited marketing budget. ### Easier to Track ROI Because PPL campaigns have clear attribution models (rules for measuring outcomes), it’s much easier to track ROI. This also makes it easier to test different creatives and channels to determine which ones are most effective. ### Generates Warm Leads PPL campaigns are usually focused on mid- or bottom-funnel actions. Remember, you’re not just looking to drive traffic: The leads you generate (and pay for) are more likely to convert into paying customers than brand-awareness campaigns. ## Considerations with Pay per Lead Marketing PPL marketing has its advantages, but it also has some drawbacks. Here are some potential challenges to consider: ### Lead Quality Can Vary Not all leads are created equal. If you don’t execute your campaign targeting effectively, or your leads come from low-quality sources unlikely to convert into paying customers, you could generate unqualified leads and waste time and money. ### Risk of Fraud If you use a third-party vendor for your PPL campaign, you could be subject to fake or incentivized leads that don’t result in sales. For example, let’s say you’re running a PPL campaign for a roofing company, and you decide to hire a third-party lead-generation vendor. Your goal is to find homeowners who are looking for roofing services, and you offer the vendor $75 for every qualified lead. The vendor, in an effort to drive a high volume of leads, involves several affiliate partners. One of the affiliate partners runs an aggressive sweepstakes offer, where users can enter to win a $500 Home Depot gift card. The sign-up may gather all of the contact information required to fulfill the lead, but the user has no intention of hiring a roofer — they just want the gift card. ### Campaigns Can be Difficult to Manage A high-performing PPL campaign requires ongoing optimization, including testing various creatives and platforms, as well as close collaboration between the marketing and sales teams. This adds complexity, and if things are not aligned properly, you could lose valuable leads. ## Choosing the Right Pay-per-Lead Program Before committing to a specific PPL program, make sure it includes the following: ### Well-Defined Lead Criteria Everyone involved needs to understand what qualifies as a lead and how those leads will be delivered and tracked. This will reduce the number of low-quality leads and irrelevant submissions. It also sets the foundation for effective tracking and optimization, and protects ROI. ### Proven Track Record Only work with vendors or affiliates who have a track record of success in similar industries and customer segments. You can ask for references, read online reviews, check with industry experts, and read case studies. For example, Taboola publishes case studies of successful advertising campaigns on its website. ### Scalability Look for programs that allow you to scale your campaign as your results improve, without sacrificing the quality of your leads. Not all vendors can handle large volumes of leads, so you want to find large networks with multiple traffic sources that offer automation and lead-filtering tools, to ensure high-quality leads. ## How to Optimize PPL Costs ### Improve Your Targeting Don’t just send your ads to everybody: Use audience segmentation and behavioral data to target the right audience for your ads. While this can take time, laser-sharp targeting will result in more relevant leads and lower ad spend. ### Test Continuously Always be A/B testing your ad creatives, offers, and landing pages. Even small changes to images, text, or call-to-action (CTA) buttons can result in higher conversion rates. Take advantage of performance advertising platforms that offer AI-powered A/B testing, which allows you to continuously test and optimize in real time. ### Automate The more you can automate the various steps in your PPL campaign, the more quickly leads can move through the funnel. For example, identify tools that can automate lead scoring (assigning a rating to each lead based on the likelihood of the user becoming a customer), customer relationship management (CRM) integration, and follow-ups. ## PPL vs. PPC Pay-per-lead (PPL) and pay-per-click (PPC) models may sound similar, but each method is used to accomplish different marketing goals. For starters, PPL is outcome-focused: You only pay when you receive leads, which are more likely to result in paying customers. The purpose of PPC campaigns is to drive traffic. You pay for every click, even if the user doesn’t convert. PPC campaigns typically sit at the top or in the middle of the sales funnel. PPL campaigns operate in the middle to the bottom of the funnel. ## Key Takeaways Pay-per-lead (PPL) marketing allows advertisers to generate high-quality leads while keeping ad costs low. Because you’re only paying for qualified leads, it’s easier to track return on assets (ROA) and scale campaigns based on actual results. But, successful PPL campaigns rely on clearly defined lead criteria, choosing the right partners, and actively managing campaigns to avoid low-quality or fraudulent leads. If done correctly, PPL can be an excellent way to grow your business. ## Frequently asked questions (FAQs) ### What is the pay-per-lead generation model? A pay-per-lead generation model is a type of advertising in which businesses only pay when a qualified lead is generated, instead of paying for traffic, clicks, or impressions. The goal of PPL advertising goes beyond brand awareness to finding potential customers. ### Is pay per lead legit? Yes, PPL is a widely used digital marketing model. However, like any model, it can only be effective when used properly, which is why it’s crucial to choose a partner and establish clearly defined lead criteria. Otherwise, you could be subject to fraud or poor results. ### PPL vs. PPC vs. CPA: What’s the difference? With pay per lead (PPL), you pay when, e.g., a user submits a form or expresses interest in your product or service. With pay per click (PPC), you pay as soon as someone clicks your ad (you don’t pay for impressions only). With cost per acquisition (CPA), you pay when someone completes a specific action, such as making a purchase or subscribing to a service. In the sales funnel, PPL sits between PPC and CPA. ### What is an example of a pay-per-lead campaign? A window company runs a PPL campaign where they only pay when someone fills out a form to request a free quote for new windows. They partner with a lead-generation platform that drives traffic through Google Ads and Facebook Ads. Users must click on the ad and complete the form, which includes their name, home address, and contact information. Every time it happens, the platform qualifies it as a lead and charges the window company $50. ### How do you calculate pay per lead? A straightforward way to calculate PPL is to divide the total marketing campaign spend by the number of leads generated. For example, if you spend $1,000 on a campaign and generate 100 leads, your PPL is $10. --- ### Ad Inventory: How to Make the Most of It? URL: https://www.taboola.com/marketing-hub/ad-inventory/ Last Modified: 2026-06-22 08:51:26 The term “advertising inventory” can be a little confusing, as it refers to a concept of space or individual slots, rather than tangible items. While it previously meant the pages offered in print publications, over time, it has come to almost exclusively refer to available slots in digital advertising. Digital advertising inventory is the amount of internet advertising space a publisher is making available for purchase by advertisers. These opportunities exist in multiple formats such as desktop websites, mobile apps and websites, videos and video ads, and audio ads. Here, I’ll explain what this all means for advertisers and publishers, and how to make the most of ad inventory when it comes to optimizing campaign performance. ## Defining Inventory in Digital Advertising ### Who Owns and Manages Digital Advertising Inventory? Ad publishers own and manage digital advertising inventory, and it is theirs to sell to any chosen buyers. In the context of digital advertising, the ad inventory is essentially a publisher’s property, or their personal shop. Think of it as how a grocery store owner gets to choose which pantry items get featured on their shelves, by making business deals with various product suppliers and selecting what goes on top, what goes on the bottom and middle shelves, and what might get displayed right near the checkout line. ### Why Is Understanding Inventory Crucial for Both Advertisers and Publishers? Publishers need to know how much ad inventory they have and how much it’s worth, so they can sell as much of it as possible to the highest bidder. Marketers need to know how much quality ad inventory they can afford, and where their advertising dollars will be most effective, as this will determine the avenues they have for sharing their campaign messaging and reaching consumers in the smartest ways possible. ## Types of Digital Advertising Inventory ### What Is Display Advertising Inventory? Display advertising inventory refers to the space a publisher can sell an advertiser for promoting display ads such as banners, interactive ads, animations, and other visual formats. ### What Is Video Advertising Inventory (Pre-Roll, Mid-Roll, Post-Roll)? Video advertising can exist as pre-roll, mid-roll, and post-roll. Pre-roll ads appear before the video content a viewer has set up to watch. A mid-roll ad is shown during the chosen video, with the video action paused during the ad. Both pre-roll and mid-roll ads have decent chances of at least being partially watched, as the viewer is unlikely to get up and leave while watching a video they chose. A post-roll ad is shown after the selected video has run its course. Post-roll ads can be a risk, as the user doesn’t have a particular reason to stay if they’re finished watching a video. However, they can sometimes work, especially if the viewer is sticking around for the next video in the queue. ### What Is Native Advertising Inventory? Native advertising inventory refers to the ads that blend organically with a platform’s content, such as sponsored content or in-feed advertisements on social platforms. These kinds of ads can be effective as they don’t shout “ad” to the viewer and can feel like a natural part of the user’s experience on the site or app. ### What Is Audio Advertising Inventory? Audio advertising inventory is the amount of ad space available during audio presentations such as digital radio, streaming music, or podcasts. ### What Is Connected TV (CTV) Advertising Inventory? CTV advertising inventory is the available space for advertisers to show their ads on connected TV, which is a device showing content that is delivered via the internet. This includes smart TVs that stream from platforms like Netflix, as well as Roku and gaming consoles. ### What Is Out-Of-Home (OOH) Digital Advertising Inventory? Out-of-home (OOH) digital advertising inventory is the space available on screens for advertisers to get their messaging to consumers in public. These screens might include digital billboards or interactive displays in public spaces such as transit hubs and shopping malls. ## How Inventory Is Bought and Sold ### What Is Programmatic Advertising and How Does It Involve Inventory? Programmatic advertising allows advertisers to buy inventory using software that speeds up and automates the bidding and purchasing experience. The software uses algorithms based on data and targeting goals, with the aim of buying inventory that will help advertisers reach the right viewers at the right time for the best prices. ### What Are Ad Exchanges and How Do They Facilitate Inventory Transactions? An ad exchange is a real-time marketplace where publishers and advertisers sell and buy ad space. The exchanges facilitate inventory transactions by serving as the digital forum where these transactions take place. ### What Are Supply-Side Platforms (SSPs) and Their Role in Inventory Management? Publishers use supply-side platforms to manage their inventory and sell to the demand side. The SSP helps publishers connect with potential buyers such as advertisers looking for space to put their ads. As part of programmatic advertising, an SSP facilitates real-time bidding and automated auctions, managing inventory as it gets sold or made available. Without the technology used in programmatic ad-buying, marketers and publishers would engage in direct buys, i.e., the deals would be done by human beings making their own informed choices based on their observations and experiences. ### What Are Demand-Side Platforms (DSPs) and How Do They Access Inventory? A demand-side platform helps advertisers purchase ad space automatically in real-time from publishers on the supply side. It uses data to aid advertisers in selecting ad space based on audience targeting and algorithms. It is the other side of the programmatic advertising ecosystem, in relation to an SSP. ## Factors Influencing the Value of Inventory ### How Does Website Traffic and Audience Size Affect Inventory Value? Website traffic and audience size can have a large impact on inventory value. For example, a fledgling website with barely any traffic and a small audience can't demand the same price as a website with a wide and dedicated fan base and heavy viewership. When marketers buy ad inventory, they’re investing in how much traffic and viewership they expect for that particular moment on that particular day. ### How Does Ad Placement and Viewability Impact Inventory Pricing? It stands to reason that ad placement and viewability will affect inventory pricing. By making a display ad larger and easier to see, or placing an audio ad during a hit podcast, publishers are giving marketers’ advertisements more of a chance of reaching target audiences. By contrast, marketers seeking a smaller or less prominently featured ad can secure more budget-friendly (if perhaps less effective) inventory purchases. ### How Does Seasonality and Demand Affect Inventory Prices? During holidays and back-to-school seasons, there is a greater demand for ad space due to these being prime time for shopping. Advertisers are eager to pitch their sales and promotions, meaning many people are vying for space, which leads to higher pricing due to heavier demand but limited inventory on the publisher side. ### How Does the Quality and Relevance of the Content Surrounding the Ad Space Influence Value? The quality and relevance of the content surrounding the ad space are key factors in whether an advertiser will get good value from purchasing said ad space. If you buy a space on a site that nobody looks at or trusts, or your ad appears in a video that has no relation to the demographic you’re trying to reach, you haven’t gotten a good value and you won’t see much outcome from your expenditure. But, if you use your data and targeting strategies wisely when making a purchase, such as on the right website or the right podcast, you can help ensure a significant return on ad spend (ROAS) with a comparably small customer acquisition cost (CAC). ## Managing and Optimizing Inventory ### How Do Publishers Manage Their Ad Inventory to Maximize Revenue? Publishers can use a variety of software programs to help them manage, track, and control their ad inventory. They can also use software to help analyze performance, automate optimization, schedule ad placements, and otherwise handle tasks large and small. Along these lines, advertisers also want to maximize revenue by ensuring they are accessing high-quality and relevant inventory. They can do this by checking if a chosen platform can reach a brand’s target audience in a way that will maximize certain KPIs or other goals, and by comparing their success with one publisher versus another. They can also use AI-powered A/B testing to make direct comparisons in real time, and use data analytics to track what worked and what didn’t. ### What Is Inventory Forecasting and Why Is It Important? Ad inventory forecasting means estimating the amount of ad space that will be available in the future. It helps advertisers and publishers get a sense of what kind of pricing and bidding strategies they should line up, especially when it comes to holidays and other busy seasons. The data extracted from ad servers helps advertisers see what might be coming up, which can aid them in planning for upcoming busy or fallow times. ## Key Takeaways Ad inventory is the amount of space publishers have on hand to sell to advertisers. Advertisers then buy the space for a specific time, and showcase their ads on the publisher’s platform. Bidding can be handled programmatically or via human decision-makers. It’s important for advertisers to buy from publishers that can help them reach the right audience at the right time and in the right way, in order to maximize their ad spend. ## Frequently Asked Questions (FAQs) ### What is the difference between guaranteed and non-guaranteed inventory? Guaranteed inventory is reserved for a specific buyer at a specific price, whereas non-guaranteed inventory is available to all buyers who are willing to compete for it. ### What is remnant inventory? Remnant inventory is inventory that a publisher has not yet been able to sell to an advertiser, so they offer it at a discounted price. ### How does header bidding impact ad inventory? Header bidding, also known as pre-bidding or advance bidding, allows multiple DSPs to compete for high-demand inventory on multiple exchanges before an auction occurs. This generally leads to higher pricing for the ad inventory, meaning higher costs for advertisers and higher revenue for publishers. ### What are private marketplaces (PMPs) and how do they relate to inventory? A private marketplace is an auction where only advertisers selected by the publisher can bid on the advertising inventory. It can benefit both sides, as advertisers get to bid on premium placement and publishers get to select who buys that placement (and also charge a hefty price for it). --- ### Open Internet: Key Aspects and Importance for Advertisers URL: https://www.taboola.com/marketing-hub/open-internet/ Last Modified: 2025-07-21 12:22:01 Most of the time when we’re talking about the “web,” we are referring to the open web or open internet — the publicly accessible internet, free from control by any single entity. In other words, the internet we all take for granted. Thus far, the open web operates under the principle of net neutrality, i.e. that internet service providers don’t have the right to do site arbitrage, in a manner of speaking. In theory, your internet cable or fiber optic service is required to give you equal access to all sites (excluding illegal material). For example, your internet service provider (ISP) cannot decide to charge Netflix an extra penalty on threat of throttling their traffic, even though Netflix consumes much more data than the average company, even by streaming standards. Similarly, ISPs cannot, for example, charge customers an extra $10 each time they purchase something from Amazon. The open net is a matter of some controversy, however. Detractors argue that some companies, such as streamers or GenAI tools that generate graphics and videos, consume more than their fair share of bandwidth and thus should pay extra to the ISP, just as a company using lots of electricity should pay for that electricity by wattage like everyone else. South Korea, e.g., attempted to impose extra penalties on Twitch, causing the livestreaming service to withdraw from that country. Not every nation enforces net neutrality, either: As of this writing, for example, Australia and China do not enforce it. In the U.S., the enforcement is a state-level decision, not a federal one. What, then, does not constitute the open web? That would be the so-called “walled garden,” which controls what the user sees, and — most importantly to marketers — which ads are shown, when they are shown, and how frequently they are shown. Examples of this would be Google and Meta, which have powerful search capabilities, enormous amounts of data, and billions of users, yet remain closed systems. The openness or non-openness of a web-based system is based on whether it restricts/filters data for users; the huge number of visitors has no bearing on the definition of open web. Marketers should understand how to leverage both the open web and the walled garden. The latest AI-driven marketing tools are revolutionizing how brands can maximize their impact in each. Here’s a breakdown of how to use both the open web and walled garden to your advantage. ## What Is the Open Web in Advertising, and Why Is It Important for Marketers Especially? The open web is a gold mine for advertisers seeking flexibility, transparency, and direct connections with audiences. Unlike walled gardens, where platforms like Google or Meta keep their data close to the vest, the open web offers a clearer view of campaign performance, more control over ad placements, and a chance to build robust first-party data systems. With privacy laws tightening — particularly in the European Union — and third-party cookies being phased out, the open web’s transparency and ethical targeting options make it a critical playground for brands aiming to stay ahead. ## Three Key Aspects and Benefits of the Open Web ### 1. Diverse Publishers, Niche Audiences The open web spans untold numbers of independent sites useful to marketers, from tech blogs to cooking magazines. This diversity lets brands target specific audiences in contextually relevant environments, like placing a coconut water ad on a health blog. ### 2. Transparency and Control Advertisers get detailed metrics on viewability, clicks, and conversions, allowing real-time tweaks to optimize budgets and boost return on investment (ROI). ### 3. First-Party Data Power As third party data gets phased out, platforms utilizing first party data will become more and more vital. Realize, e.g., lets marketers harness over 17 years’ worth of behavioral data, creating tailored campaigns without relying on opaque platform algorithms. ## How Can Marketers Leverage the Open Internet in Their Strategy? Artificial intelligence (AI) is transforming how marketers operate on the open web, making campaigns smarter, faster, and more effective. Here’s how to leverage these tools on the open web: ### Programmatic Advertising with AI Programmatic platforms like The Trade Desk or Google’s Display & Video 360 use AI to automate ad buying across the open web. Real-time bidding lets you snag ad space on thousands of sites instantly, targeting users based on behavior, location, or interests. The latest AI-powered marketing tools also allow you to set granular targeting parameters — like age, interests, or device type — and let the algorithm optimize bids for maximum ROI. For example, a cosmetics brand could target users reading about “best lipstick for olive skin” on beauty blogs, adjusting bids in real-time to prioritize high-engagement sites. ### Contextual Targeting in a Cookie-Less World With third-party cookies phasing out, AI-driven contextual targeting is a game-changer. Tools like Realize or Oracle’s Contextual Intelligence analyze webpage content in real time, matching ads to relevant topics without invasive tracking. ### Native Ads via Content Discovery Platforms AI tools have allowed marketers to blend ads seamlessly into editorial content, recommending products or stories based on user interests. These native ads feel less intrusive, boosting engagement. ### AI-Powered Creative Optimization AI-powered tools analyze ad creatives, suggesting tweaks to images, headlines, or calls-to-action (CTAs) based on performance data across open web campaigns. You can even upload multiple ad variations to an AI tool and let it identify top performers, then scale the winner across premium news sites. ## Open Web vs. Walled Gardens The following grid provides a detailed comparison, outlining the key distinctions between the Open Web and Walled Gardens across various critical aspects of the digital landscape. Feature Open Web Walled Gardens Content Access Free, open access Requires login or app Data Control & Ownership Fragmented Centralized Targeting Capabilities More data sources and less precise Highly precise targeting Transparency Higher transparency "Black box", less insights Audience Reach Broad and diverse Large, but confined, audience Competitive Landscape Highly competitive and fragmented Monopolistic environment Privacy Concerns More control over data sharing Platform dictates data usage policies Cost More cost-effective solutions Higher costs Innovation Diverse Platform-limited Brand Safety Requires careful management More controlled environment User Experience Fragmented and inconsistent, with more variety and choice Seamless, curated, consistent user experience Walled gardens offer scale and ease, but limit control and data access. The open web provides transparency, brand safety, and flexibility, but requires more effort to manage. - Open web strengths: Granular targeting, transparent metrics, and first-party data ownership. Ideal for brands building long-term, privacy-compliant strategies. - Walled garden sStrengths: Massive reach and polished ad formats. Perfect for quick wins or broad awareness campaigns. On average, you should allocate 60-70% of your budget to walled gardens for reach, and 30-40% to the open web for precision and data-building. Use AI tools to monitor performance and rebalance monthly, based on ROI. The distinct characteristics of the Open Web and Walled Gardens necessitate tailored strategies for advertisers. Understanding these differences empowers more informed decision-making for optimal campaign performance and sustainable market presence. - Balance is Key: A holistic approach leveraging the unique strengths of both ecosystems is crucial for maximizing reach and precision. - Data Strategy Evolution: Adapting to evolving data privacy regulations and leveraging first-party data effectively will be paramount in both environments. - Audience-Centric Planning: Understanding where target audiences spend their time and their mindset in each environment is vital for effective engagement. - Continuous Adaptation: The digital landscape is dynamic; ongoing monitoring and adaptation of strategies to shifts in market dynamics are essential. ## Key Takeaways The open web is a transparent, flexible space for reaching niche audiences with precision. By utilizing the correct tools — especially those powered by AI, which grant greater scale and speed — it’s possible to reach an audience beyond that offered by traditional walled gardens. Advertisers should make sure to integrate into their strategy performance tools which supports a wide range of ad formats, including native, display, vertical, and carousel ads, empowering advertisers to achieve their objectives with creative flexibility. ## Frequently Asked Questions (FAQs) ### Open vs closed internet: What’s the difference? The open web is decentralized, with thousands of independent sites offering transparency and control. Walled gardens, like Meta or TikTok, are controlled platforms with proprietary data and restricted visibility. ### What is an open internet network? An open internet network is the publicly accessible internet, free from control by any single entity. In other words, the internet we all take for granted. ### What is open display advertising? Open display advertising is the practice of placing banner or rich media ads across open websites via programmatic exchanges. --- ### Keywords: Importance, Types, Strategies URL: https://www.taboola.com/marketing-hub/keyword/ Last Modified: 2026-06-22 08:37:33 The internet is a seemingly endless trove of resources, but in the saturated world of online marketing, it’s easy for your website and ads to get lost. Keywords are crucial signposts that point consumers, as well as Google and other search engines, to your content. Keywords are terms and phrases that large segments of your target audience are searching for in their online queries. When these keywords are strategically incorporated into content, it communicates to search engines that your website answers your audience’s questions. As a result, your website will appear higher on the search engine results pages (SERP), making it easier for people to find your products and services, leading to more traffic and conversions. ## Understanding Keywords ### What Is a Keyword in Digital Marketing? In digital marketing, keywords are terms and phrases that represent your brand, and values that are important to your ideal customer. At the same time, search engines use keywords to identify and sort relevant content. By identifying what keywords your audience uses and incorporating them into content emphasizing experience, expertise, authoritativeness, and trustworthiness (E-E-A-T), companies can appear higher up in Google searches and attract new customers. ### Why Are Keywords Important for Search Engine Optimization? Keywords are important for search engine optimization (SEO) because they help Google determine what content people should see first. When people search for answers to questions or solutions to problems, they often click on results that appear high on the first page of their search results. When your website incorporates the same keywords your audience is searching for, while following E-E-A-T principles, the content has the potential to rank higher on Google and other search engines. Keywords can have a crucial impact on both organic and pay-per-click (PPC) traffic because they directly impact who sees your content. These search terms and phrases are essential for ad campaigns because they offer more potential for conversions, while reducing overall ad spend. Keywords fall into two categories: short-tail and long-tail. Short-tail keywords consist of one or two general search terms that people are searching at a high volume, but also words that other businesses use, making search engine ranking more competitive. Long-tail keywords are phrases of three or more words with lower search volume but less competition, potentially leading to longer-term traffic and sales. Determining the right short-tail and long-tail keywords depends on your industry and niche, but SEO tools like Google Keyword Planner can help guide research. ## Types of Keywords ### Informational, Navigational, and Transactional Keywords Several types of keywords relate to search intent: #### Informational Keywords These are used by individuals who are seeking information on a specific topic, or looking to learn or understand a new concept. Informational keywords are essential because they help build trust and expertise with your target audience. Although informational keywords sometimes suggest lower purchase intent, they present an opportunity to educate potential customers about your brand. For instance, a healthcare company might optimize for informational keyword searches like, “how to prevent heart disease.” #### Navigational Keywords These are search terms and phrases used by people looking for a website or topic on that website, such as “Facebook business page.” #### Transactional Keywords These indicate that a consumer is ready to buy a product or service, and have more potential for conversions. As a result, transactional keywords tend to include search terms that imply higher purchase intent, like “buy” or “download.” ### Branded vs. Non-Branded Keywords Branded keywords refer to any navigational keyword search that includes your brand name. Conversely, non-branded keywords are broader search terms that may be used by people unfamiliar with your business. While branded keywords help target customers who’ve already been to your website, non-branded keywords attract new leads by reaching a wider audience. ### Negative Keywords and Their Impact Negative keywords are terms and phrases that companies intentionally leave out of their SEO strategy. This ensures that your content will not appear in irrelevant searches for people who are less likely to click through content and convert to customers. For instance, luxury brands may exclude words like “discount” or “cheap.” ### Geo-Targeted and Seasonal Keywords #### Geo-targeted Keywords These are important for local businesses that want to target their content to customers in their area. By setting up a Google Business Profile, companies can gain insight into what keywords their local audience uses and tailor content accordingly. #### Seasonal Keywords These help companies optimize content for specific times of the year. A query for “Halloween stores in Phoenix” is an example of a geo-targeted keyword that utilizes seasonal search terms. ## Keyword Research Strategies ### How to Conduct Keyword Research Keyword research starts with understanding what your ideal audience wants and needs, and how this influences what they might be searching for online. After brainstorming words and phrases potential customers may use that relate to your products and services, you can use keyword data and SEO tools to get an idea of how many people are using these search terms and how competitive they are. ### Best Keyword Research Tools for Advertisers Free research tools like Google Keyword Planner and subscription platforms like Semrush and Ahrefs help marketing professionals identify and analyze keywords related to their brand and hone in on which terms have the most opportunity for ranking. Google Keyword Planner shows search volume for keywords and categorizes the level of competition for keywords as low, medium, or high. Semrush and Ahrefs are subscription services that combine search volume and competition with other data to determine keyword difficulty (a percentage between 1 and 100 that estimates how difficult it is to rank for specific search terms and phrases — the lower the keyword difficulty, the easier it is to rank high on search engine results pages). Google Trends can also help identify what keywords may spike in relevance during the news cycle or certain times of year. More comprehensive tools like Semrush track keyword data over time, identify trends with keywords, and recommend related keywords, phrases, and synonyms. Ahrefs also leverages search data to estimate potential traffic related to keyword trends and provides a graph showing how keywords' rankings change over time. ## Using Keywords in Campaigns ### Incorporating Keywords Into Ad Copy Once you’ve researched and compiled a list of relevant keywords, the next step is strategically incorporating them into ad copy. An effective optimization strategy involves using keywords in headlines, title tags, meta descriptions, the first paragraph, and around three more times throughout the content. That said, it’s also important to avoid over-optimizing content with excessive keyword use, which can backfire with SEO. Tools like Semrush can help flag keyword stuffing and suggest semantic keywords or synonyms for better ranking prospects. ### How Keywords Influence Quality Score Keywords have a significant impact on the quality score, a rating on a scale of 1 to 10 that gives marketing professionals an idea of how their Google ads compare to competitors. Keywords signal relevance to Google and other search engines, potentially helping your content to rank higher. This can then boost click-through rate (CTR), which increases an ad's quality score. Likewise, using branded and long-tail keywords that indicate an intention to purchase can also improve your conversion rate, positively impacting your quality score. Overall, a high quality score is crucial for successful campaigns because it makes it more affordable to bid on keywords, leading to higher ad positioning with a lower cost per click (CPC). ### Landing Page Optimization with Keywords Landing pages are crucial for keyword optimization because they have a specific goal, like signing up for a newsletter or purchasing a product. To appear high up in your audience’s search engine results page, you have to incorporate relevant keywords, ideally with lower competition and keyword difficulty. When customers click on a landing page, it suggests that they are ready to take action, so it’s important to create informative, detailed content while utilizing high-intent keywords like “buy now” or “subscribe.” Landing pages can also use keywords to link to internal content that can further educate consumers on your brand and increase CTR and conversions. ## Key Takeaways Keywords are terms and phrases that communicate to search engines how relevant your content is for your target audience’s online queries. Websites optimized with relevant keywords while adhering to E-E-A-T principles can rank higher on Google and other search engine results pages. Keywords with higher search volume and lower keyword difficulty have the best potential for SERP ranking. Remember that different keywords serve different purposes: While branded and long-tail keywords are good for targeting customers with high purchase intent, informational keywords can educate new customers. ## Frequently Asked Questions (FAQs) ### How do I find the right keywords for my business? To find the right keywords for your business, brainstorm about who your ideal customer is and what problems your products and services can help solve. For example, a wellness brand may want to engage with customers who are looking to improve their eating, exercise, and sleep habits, so keywords like “how to lose weight” or “best ways to improve sleep” would be long-tail keywords to research further. Free tools like Google Keyword Planner can give you a basic idea of how many people are using certain keywords and how competitive it is to rank for them. Subscription platforms like Semrush and Ahrefs can offer more detailed information about keyword difficulty and competitor analysis. ### What is keyword stuffing and why should it be avoided? Keyword stuffing is a manipulative SEO strategy that involves the excessive use of keywords to trick search engines into thinking that content is more relevant for queries than it is. The problem is that when irrelevant pages are ranked higher, it hurts the overall user experience and erodes trust in your brand. Since Google has made several core updates to its algorithm to address this issue in recent years, websites can get penalized for keyword stuffing with lower rankings. Resources like Semrush and Ahrefs measure content's keyword density and can help marketing professionals avoid over-optimization. ### How often should I update my keyword strategy? Companies should update their keyword strategy quarterly. However, any time Google announces a significant change to its algorithm, or if there are any major shifts in your industry, it’s important to reevaluate your SEO approach. Declines in traffic can also be a sign that you should reassess your keyword strategy. --- ### Meta Descriptions: What Are They? Why Are They Important? URL: https://www.taboola.com/marketing-hub/meta-description/ Last Modified: 2026-06-22 08:33:04 Meta descriptions are the unsung heroes of search engine results pages (SERPs). You can create the perfect blog post or landing page with the best research and optimize it to the nines with relevant keywords, but if your meta description is missing or isn’t optimized, users are unlikely to click through to your page, putting your content efforts out to pasture. Think of meta descriptions as your content’s elevator pitch, giving readers a taste of what’s to come if they click into that blue underlined link from Google. While these short descriptions don’t directly influence where your content ranks on SERP, they can make or break click-through rates (CTRs). ## Understanding Meta Descriptions Let’s talk about the basics: A meta description is an HTML element that provides a brief summary of a web page’s content. This snippet usually appears beneath the page title in search results, giving users a preview of what’s to come if they click through to your site. But, here’s where things get interesting: Search engines won’t always display your carefully crafted meta description, showing preference for content elsewhere on your page if the algorithm senses it better matches the user’s search intent. It’s Google’s Uno reverse card, if you will. ### Where Does the Meta Description Appear in Search Engine Results Pages (SERPs)? Meta descriptions occupy prime real estate in search results, just below the clickable page title and URL. On desktop, you’ll see the first one or two lines of text, while mobile results truncate it further due to screen size. The description often shows bolded keywords that match the search query, making relevant terms stand out visually. Sometimes, you’ll also see meta descriptions used via social sharing, though many platforms now use Open Graph tags instead. Additionally, meta descriptions can appear in Google News results, local business listings, and other specialized SERP features. ### What Is the Purpose and Importance of a Meta Description for SEO and Advertising? Meta descriptions are quality indicators for both users and search engines of content value and relevance. Meta descriptions aren’t a direct ranking factor on SERP, but they still play an indirect role in your content’s SEO success — Google uses these descriptions for the snippet in search results more than a third of the time, for instance. The primary purpose of a meta description is simple: Convince someone to click on your result instead of the other options on the page. This matters because Google considers user engagement signals like CTR when identifying the relevance and quality of your web page for specific queries. Higher CTRs can signal to search engines that your content is valuable and relevant. In turn, you could get more Google love through improved rankings over time. ## Key Elements of Effective Meta Descriptions Like other SEO elements, writing a compelling meta description is part art, part science. The best descriptions balance keyword optimization with clever copywriting, all within a strict character limit. Here are some key elements of effective meta descriptions: - Short length: Meta descriptions should be short, between 150 to 160 characters. - Compelling copy: Think about what motivates your target audience to click, and write your description accordingly. For instance, do they want quick answers, detailed tutorials, product reviews, or entertainment? Connect the dots in the meta description. - Action-oriented language: Avoid generic platitudes and get specific. So, instead of, “Learn more about X,” try, “Use 5 proven strategies that increase X by 200%.” See the difference? ### Should Meta Descriptions Include Relevant Keywords? How? Yes, they should, but there’s a fine line between a strategic keyword mention and trying to stuff the description to the brim with keywords. Relevant keywords in your meta description help search engines understand your content’s relevance and create visual anchors for users scanning the SERP. But, keyword stuffing can work against you and hurt user experience. Instead, incorporate your primary keyword and maybe one or two additional terms in a natural, unforced way that provides genuine value. Focus on search intent here: For example, if someone searches for “best budget laptops,” your description should include terms like “affordable,” “value,” or “budget-friendly” rather than regurgitating “budget laptops” several times. ### What Is the Importance of a Clear Call to Action in a Meta Description? A strong call-to-action (CTA) can give you a leg up on click-through rates by offering users a clear next step. Effective CTAs in meta descriptions are subtle and benefit-focused rather than having pushy sales language, which can be a turn-off. Use phrases like, “Discover how,” “Learn the secrets,” “Get your free guide,” or “Find out why.” These phrases build curiosity while suggesting concrete value that users will receive if they keep clicking. To accomplish this, match your CTA to search intent. Informational queries respond well to educational CTAs like “Learn” or “Discover,” while commercial queries might get a boost from action items like “Compare prices” or “Start your free trial now.” ## Impact on Click-Through Rate (CTR) Click-through rate is where meta descriptions work their magic. As discussed already, while they might not directly influence where your page ranks, they do play a key role in how many people actually go to your page once it ranks. ### How Does a Well-Written Meta Description Influence Organic Click-Through Rates? Strong meta descriptions set accurate expectations and highlight unique value propositions. When users see exactly what they’re looking for in your description, they’re more likely to click through — and to stay put once they arrive on your page. Research consistently shows that meta descriptions with specific numbers, emotional triggers, and clear benefits outperform generic copy. For instance, “5 Expert Tips That Doubled Our Conversion Rate in 30 Days” will likely beat “Tips for Better Conversion Rates.” The more specific and action-oriented, the better. ### What Are Some Common Mistakes That Can Lead to Low CTR from Meta Descriptions? Even if your content ranks well, certain missteps can hurt your CTRs. Duplicate meta descriptions across multiple pages is a frequent error, and identical or similar descriptions on every page of a site aren’t helpful when individual pages appear on SERP. Other CTR-killing mistakes include vague or generic language, descriptions that are too long and get truncated as a result, and descriptions that don’t match page content. Nothing frustrates users more than clicking on a promising description only to find irrelevant or lackluster content. Technical errors can also tank CTRs, such as missing meta descriptions. This forces search engines to create their own snippets, which seldom capture your content’s key selling points. Keyword-stuffed meta descriptions also read awkwardly and can signal low-quality content to users and search engines. ### How Can You A/B Test Different Meta Descriptions to Optimize for CTR? A/B testing meta descriptions requires patience and consistent tracking, but the payoff can be huge. Start by identifying pages with decent rankings but less-than-optimal CTRs; these represent your biggest optimization opportunities. Create two versions of your meta description: one control (your current copy) and one variant with different messaging, keywords, or CTA language. Tools like Google Search Console can help you track performance shifts, though you’ll need to give changes a few weeks to show statistically significant data. Stick to testing high-impact pages first, such as your homepage, primary product pages, or top-performing blog posts. Small wins on these pages can help drive significant traffic bumps across your entire site. ## Meta Descriptions in Paid Search Ads The same principles you use to write effective meta descriptions for organic search extend into paid advertising, where compelling ad copy directly impacts your ad ROI. ### How Are Meta Descriptions Used in Search Engine Advertising? In paid search, meta descriptions transform into ad descriptions — the text that appears below your headline in Google Ads. While the character limits and best practices aren’t wildly different, the stakes are definitely higher for paid search, since you're paying for each click. Paid search descriptions require clear value propositions, relevant keywords, and compelling CTAs to win eyeballs and clicks. However, you can afford to be more direct with commercial language in paid search metas, since users expect promotional content in paid ads. ### Do Ad Extensions Replace the Need for Compelling Meta Descriptions? Ad extensions enhance your ads with additional information like phone numbers, location details, or site links, but they don’t replace strong ad copy. Think of these extensions as bonus features that complement strong descriptions. The main ad description still does the heavy lifting of convincing users to click — extensions simply provide additional paths to conversion and make your ad appear more prominently on the page. On average, Google users click on a paid ad at a rate of 6.42%, according to WordStream’s Google Ads Benchmarks 2024 report. The combination of strong descriptions and relevant extensions is the perfect marriage for driving the best performance. ## Best Practices and Optimization To master your meta descriptions, you have to pay close attention to search behaviors, what your competition is doing, and how your own content strategy might shift over time. ### How Often Should You Review and Update Your Meta Descriptions? Meta descriptions aren’t a one-and-done task; they need constant care and attention. Here are some scheduled review cadences to consider: #### Quarterly Audit Schedule - Conduct comprehensive meta description reviews every three months. - Prioritize updates based on performance data, not arbitrary timelines. - Focus most attention on pages with strong rankings but low CTR. - Leave well-performing descriptions alone unless you make significant content updates. #### Content-Driven Updates - Major blog post revisions with new data or info. - Seasonal campaign launches or promotional periods. - Shifts in search trends or user behaviors. - Algorithm updates that impact SERP display formats. #### Monthly Performance Monitoring - Track Google Search Console data for CTR dips. - Identify pages where rankings improve but clicks don’t follow. - Flag pages with high impressions but low CTRs. - Monitor competitor SERP changes that might impact your visibility. ### What Tools Can Help You Analyze and Optimize Your Meta Descriptions? Numerous SEO tools can help you analyze and fine-tune your meta descriptions, with more AI-powered platforms popping up to help you optimize copy, too. The best option for you depends on your company’s budget, size, and content needs. #### Premium Tools: - Semrush. - Ahrefs. - Yoast SEO. - Screaming Frog. - DeepCrawl. - BrightEdge SEO. #### Free Tools/Resources - Google Search Console. - SERP Snippet Preview. - Character Counting Tools. - Google’s Mobile-Friendly Test. ### How To Handle Missing or Poorly Written Meta Descriptions at Scale When you spot bad meta descriptions or pages that are missing them, roll up your sleeves and get to work addressing the issue. First, you’ll want to prioritize based on three tiers: - Tier 1: Homepage, primary product pages, and top organic traffic pages. - Tier 2: Category pages, popular blog posts, and conversion-focused landing pages. - Tier 3: Archive pages, older content, and low-traffic supporting pages. #### E-commerce Solutions - Implement programmatic description generation for product pages. - Include key items like brand name, price range, and primary benefits. - Create dynamic templates that pull from product databases automatically. - Ensure consistency without sacrificing uniqueness across product variations. #### Template-Based Approaches - Blog Posts: " Tips That + ." - Product Pages: " + + + ." - Service Pages: " in + + ." - About Pages: " + + + ." ## Key Takeaways When done thoughtfully and well, meta descriptions can drive organic traffic to your site even though they don’t directly impact ranking factors. Well-optimized meta descriptions can increase CTRs by 30% or more, translating to big traffic wins even without a bump in rankings. Keep descriptions short and sweet (150 to 160 characters, max), include relevant keywords naturally, and focus on compelling value propositions that match search intent. Regularly audit and test your copy to find opportunities for improvement and maintain solid CTRs over time. ## Frequently Asked Questions (FAQs) ### Does Google always use the meta description I provide? Google uses them for the descriptive page snippet in the search results 37.22% of the time, according to Ahrefs. Search engines often create their own snippets when they think other page content better matches the user’s specific query. Create valuable, relevant descriptions instead of trying to force Google to use your exact text word for word. ### Are meta keywords still relevant? Meta keywords tags aren’t a thing anymore. Google, Bing, and other search engines ignore meta keywords entirely due to historical spam abuse. Instead, focus your keyword optimization efforts on title tags, meta descriptions, and actual page content. ### How important are emojis in meta descriptions? Emojis can help your search results stand out visually and, in some instances, boost CTRs, depending on the industry. However, the general rule of thumb is to use them sparingly and only if they truly add value or match your brand personality. When in doubt, don’t use emojis. ### Should every page on my website have a unique meta description? Yes, unique meta descriptions are necessary on each page to identify its purpose and target different keywords. Having identical or too-similar descriptions on every page of a site can confuse users and search engines. ### What are some examples of effective meta descriptions? Strong meta descriptions combine specific benefits, relevant keywords, and compelling, action-oriented CTAs. For instance, “Discover 7 proven email marketing strategies that increase open rates by 40%. Get actionable tips and free templates.” This description includes numbers, specific benefits, results, and a value-added offer, all within the character limit. --- ### Referral Marketing vs. Affiliate Marketing: Which Is Best for Your Business URL: https://www.taboola.com/marketing-hub/referral-vs-affiliate-marketing/ Last Modified: 2025-06-05 10:07:34 Referral marketing and affiliate marketing are both highly effective ways to help businesses drive sales and expand their audiences. While these marketing strategies work similarly, there are some key differences in how they’re structured, their target audiences, and the motivating factors that affect conversions. Below, I’ll compare referral marketing versus affiliate marketing to help you better understand which types of businesses, products, and goals these marketing strategies are best for. ## 12 Core Differences to Know Between Referral and Affiliate Marketing Referral Program Affiliate Marketing 1. Who does the referring? Content creators, influencers, or marketers who have an audience aligning with the business', and who usually have a connection to the business, such as having been past customers. Content creators who may not have connections with the business. 2. Relationship with potential customer Referral program participants often promote within their existing personal network, such as friends or family, as the personal connection helps establish trust. Affiliates often have no direct connection to the potential customer, and they target anyone reading the content. 3. Types of rewards Cash rewards, coupons, gift cards, account credits, or other non-cash rewards. Cash, as a percentage of sales generated or a flat fee. 4. Incentive structure Incentives are usually paid as a one-off. Incentives are a percentage of sales and are usually paid out monthly. 5. Promotion Participants may promote products or services via email, social media, and text. Marketers promote products or services via websites, blog posts, social media, ads, and videos. 6. Tactics Participants share a referral code within their networks. Affiliate marketers share an affiliate code through content on their established platform. 7. Trust and credibility Referral programs are usually seen as more trustworthy, because the person doing the referring often has a personal connection with the business. Affiliate marketing is sometimes seen as being less trustworthy, because the marketers don’t necessarily have personal connections to the business. 8. Audience reach The audience is limited to the participant’s network of friends or family. Audiences can be larger, depending on the established marketing platform and its reach. 9. Goals Expand the business' audience and reach customers who are likely to refer others. Expand the business' audience and drive sales volume. 10. Costs Costs are minimal, often including gift cards or discounts. Costs are based on the number or value of goods and services sold, and are a flat fee or percentage. 11. Tracking Referral link with embedded referral code. Affiliate link with an affiliate code. 12. ROI Dependent on product profits and incentive structure. Dependent on product profits and commission structure. ## Differences Between Referral and Affiliate Marketing Explained Through referral marketing and affiliate marketing, businesses can sell and promote their products or services to new audiences. Both of these strategies are effective ways to expand your marketing efforts, and since the individuals doing the promoting are outside of the business, they require minimal business staff time. Plus, since you only pay out incentives after sales are made, both of these marketing techniques are budget-friendly. While referral marketing and affiliate marketing can both help expand audiences, the strategies work differently and are often best suited for certain industries or goals. Many businesses successfully use both marketing strategies simultaneously. ### Referral Referral marketing, which is sometimes called ambassador marketing or word-of-mouth marketing, is based heavily on trust. In referral marketing, a business partners with a participant, who is often a current or previous customer. The participant refers the business' services or products to their existing network of friends and family. Since the participant or marketer is reaching out to current connections and has firsthand experience with the business, there’s an element of trust that can help build confidence and drive sales. Friends and family can ask the participant about their personal experience with the business, and since they’re already warm leads, referral marketing can generate higher conversions. Referral marketing incentives may include cash, but it’s more common for businesses to offer incentives like store credit, coupons and discounts, and free products. Program participants won’t necessarily receive incentives for every item or subscription sold — instead, the business may set an incentive structure, such as awarding a $20 gift card for every 10 products sold. For example, a referral participant who has bought hair care products from a certain business would receive a referral link to track their sales. The participant might email friends and family about a special sale on those hair products, including the code for their network to make a purchase. When anyone from the network buys the hair care products using the code, the sale is logged, and once the sales reach the designated milestone, the participant will receive an incentive. Referral marketing can be an excellent way to generate high-quality leads with a high conversion rate. Harvard Business Review found that customers who joined businesses through referrals made more purchases than customers who were acquired via other methods. Additionally, referred customers themselves end up referring 30% to 57% more new customers through referrals than other, non-referred customers. All in all, then, referrals can be a very valuable source of business. Referrals often work well in industries like B2B SaaS, financial services, and online insurance. In these industries, new leads often need to have a good degree of trust in the products and services they’re buying, and that trust is provided through their personal connection with the referral. Have a look at some of the best referral programs by industry in 2025. ### Affiliate Affiliate marketing can also help a business expand their audience and drive sales. Affiliate marketers have built a platform with a large audience, and they can promote a business' products on that site or through video, social media, and email. The marketer doesn’t necessarily have an established connection with the business or the products, though some marketers perform product testing and write product reviews to help build audience trust and drive sales. Compared to referral marketing, affiliate marketing tends to have a greater reach. While referral marketers reach out to their current personal networks, affiliate marketers can develop a much larger audience. They can also leverage the power of organic search, social media marketing, and more to help their content reach a larger audience, potentially increasing sales. Due to its success, affiliate marketing is rapidly growing. According to Business Research Insights, the global affiliate marketing industry reached $17.33 billion in 2024, and is predicted to reach $63.87 billion by 2033. Affiliate marketing can be a source of valuable data. By choosing a trusted affiliate network, your business can access a dashboard including detailed sales reports and statistics to help you better understand your audience and the affiliate program’s performance. Affiliate marketing incentives are cash-based, but businesses may structure them as a flat fee or a percentage of a product’s sale price. An affiliate marketer who has developed a platform, like a website specializing in smartphone advice, can sign up for an affiliate marketing program. The marketer will receive an affiliate marketing link to use for each recommended product, and they can embed that link into their website content. When consumers buy products using the links, each sale is tracked, and the marketer will receive the appropriate incentive. Affiliate programs often work well when selling individual products, as well as subscriptions. It’s excellent for the commerce industry, and it’s also effective for B2C subscriptions and online education courses. Since consumers may research high-value purchases, like when buying electronics or an appliance, affiliate marketing content that includes recommendations as well as detailed research and information on the products can help drive sales. ### Which Is Better? Both referral and affiliate marketing have advantages and disadvantages, and neither option is definitively better than the other. If you want to get started with referral or affiliate marketing, then consider which is right for your existing business and goals. Referral and affiliate marketing sometimes work better in certain industries: Affiliate marketing can work well for subscription-based products, while referral marketing can be a good fit when you’re promoting a purchase that relies heavily on trust, like when promoting a service provider such as a financial advisor or a performance advertising platform. Referral marketing can also work well for niche businesses, particularly if you have a network with a shared interest in that niche. ## Key Takeaways Referral marketing and affiliate marketing are both highly effective strategies that can increase your business' reach and help build your audience while driving sales. Businesses may use these techniques independently or together, depending on their goals. If trust is a major element in driving sales or you’re selling a niche product, you may find more success through referral marketing. Alternatively, if you’re running an e-commerce business or want to reach a broader audience, affiliate marketing may be the better choice. Both of these techniques can be a low-cost way to help build your audience, without requiring a significant time investment on your team’s part. ## Frequently Asked Questions (FAQs) ### Does referral marketing typically have a smaller but more engaged audience reach, compared to the potentially wider reach of affiliate marketing? Yes — since referral marketing is based on a marketer’s established network of connections, like friends and family, its audience reach tends to be smaller than the audience an established affiliate marketing platform could reach. Since a referral audience has an established connection with the marketer, though, that audience tends to naturally have more trust in the marketer. Additionally, since the marketer has usually purchased from the business or used the product they’re marketing, that experience and their connection with their audience can help build further trust and drive sales. ### How does the tracking of referrals differ from the tracking of affiliate sales? Businesses track referrals and affiliate sales using referral or affiliate links that are generated to be unique for each participant or marketer. The links include a code that signals the number of times each link has been used, as well as how many purchases or subscriptions the code was used to create. While the codes and links are very similar, referral and affiliate sale tracking differs in how the sales are monitored and counted. In affiliate marketing, marketers receive a commission for every sale generated, so every individual sale is counted. In referral marketing, participants may receive an incentive for a certain number of sales. For example, they might need to make 10 sales to qualify for an incentive. All of the sales are counted, but the business tracks how many incentive milestones are reached and then awards incentives accordingly. ### What types of rewards are typically offered in referral programs compared to affiliate programs? Referral program rewards can include cash, as well as other incentives like store credit, free products, discounts, or gift cards. Affiliate programs focus more on cash incentives, whether the cash is a percentage of the sales or a flat fee. --- ### 10 Worthwhile Referral Programs by Industry: SMBs, Marketers, Content Creators Opportunities URL: https://www.taboola.com/marketing-hub/best-referral-programs/ Last Modified: 2026-03-16 13:25:25 As businesses continue looking to referral programs to connect with new customers and grow revenue, it seems like everyone’s trying to find the best options. Since the “best” ultimately means the one that makes the most sense for your particular brand, it can be a tough nut to crack. With that in mind, below, I’ve broken down the best referral programs for each industry by straightforward features like value and accessibility. What’s changed in our 2026 update: - All entries include updated and current information, prices, and advice. - Pros and cons added to every entry. - All FAQs updated with new and current information. ## Best Referral Programs by Industry in 2026 ### 1. Digital Advertising: Realize This leading performance advertising platform is designed to help brands scale beyond the walled gardens of search and social, which continue to show diminishing returns. Realize’s tools incorporate advanced artificial intelligence (AI) to help marketers target customers on the open web by intent, rather than just identity. This makes it ideal for businesses interested in reaching customers at the decision stage through video placements, carousel ads, real-time A/B testing and optimization, and much more. Offer: Earn $1,000 cash when someone you refer spends $5,000 within 45 days. How it works: Simply join the program and start referring. You’ll be paid in cash, not ad credit, and you don’t even need to be a customer to participate in the referral program. Anyone can join: Whether you’re a marketer, publisher, or consultant, the simple structure makes it an easy program to learn. How to enroll: To join, you’ll simply fill out the signup form and submit your first lead. The team will review your application. Once you’re approved, the Realize team will take over onboarding your referred lead. Once the advertiser you refer reaches the spend threshold, you’ll be able to send an invoice and get paid. Pros: - Cash payouts. - Open to non-customers. - High referral reward. - Hands-off onboarding. Cons: - Spend threshold required. - Not currently accepting leads outside of the United States, Europe, and South America. ### 2. Email Marketing: MailChimp With more than half of the market share, it’s safe to say MailChimp is the most popular email marketing solution on the market today. In addition to its email tools, MailChimp also offers landing pages, a CRM, and tools for building social media posts and ads. Offer: Earn $30 in bill credits for each friend you successfully refer to the MailChimp platform. How it works: The MailChimp referral program works through a badge that you put in your emails and signup forms. That badge includes a link: When someone clicks on that link with cookies enabled and signs up for a paid plan within 60 days, you earn a commission. How to enroll: You’ll first need to enable the referral badge in your account settings. You can then add the badge to your website and emails. Credits will be applied automatically when someone clicks over and signs up for a paid plan. Pros: - Easy setup. - No approval process. - Automatic tracking. Cons: - Credit-only rewards. - Low payout value. ### 3. SaaS: Notion Notion is a productivity solution that combines note-taking, task management, and database tools. With more than 100 million users, Notion has become a go-to tool for freelancers and teams, who love its customizability and ease of use. While the initial payout for each referral is low, Notion offers an ongoing percentage that can keep paying dividends for the first year. Offer: Earn $50 per referral plus 20% of year-one revenue for each person you successfully refer. To qualify, a referrer must upgrade to a Plus or Business plan within 180 days of clicking your affiliate link. How it works: When you create a link through PartnerStack, you’ll then share that link through your own channels. You don’t have to sign up for anything, simply create a link and share. Commissions are paid to the owner of the last link the person clicked before signing up for the platform. How to enroll: To get started, sign into PartnerStack and create an affiliate link. Start sharing on social media, your website, through email, and anywhere else you interact with customers. You can immediately monitor your status in the PartnerStack dashboard. Pros: - Recurring commissions. - Long attribution window. - Strong brand adoption. Cons: - Paid plan required. - Competitive space. ### 4. Cloud Accounting: FreshBooks FreshBooks is an easy-to-use accounting platform geared toward small businesses and freelancers. Its tools make it easy for users to invoice, track expenses, and prepare for taxes. Since its founding in 2003, FreshBooks has attracted more than 30 million users. Offer: Earn a $100 account credit for each new user you bring to the platform, once they remain a paying subscriber for at least 60 days. How it works: FreshBooks offers a referral link that you can use to invite others to join. When someone signs up using that link, you’ll be noted as the referrer. As long as the person remains a paying member for 60 days, you’ll receive an account credit. How to enroll: While logged in to FreshBooks, choose your profile picture or initials in the top-right corner, then select Refer a Friend. From there, you can either copy the link, enter the email address of the person you want to refer, or click a social share icon. Pros: - Simple referral link. - Clear qualification rules. - SMB-friendly audience. Cons: - Account credit only. - Limited earning potential. ### 5. Banking and Fintech: SoFi SoFi is a financial technology (fintech) company that offers loans, investments, banking services, and more. The company focuses on younger demographics who like the ease and convenience of online banking. Offer: SoFi has a variety of referral options for members, with rewards ranging from $50 to $1,500. The biggest rewards come from its student loan division, but you can also earn $300 by referring someone who takes out a personal loan. How it works: You’ll need a SoFi account to participate in the referral program. One of the biggest differentiators between SoFi and other referral programs is that you’re prohibited from blasting the offer to your entire network — you’re encouraged to only share the link with people you know personally. How to enroll: In the SoFi app, log in and click on Refer a Friend. Use that referral link to get started. Pros: - High-value rewards. - Multiple referral options. - Trusted consumer brand. Cons: - Account required. - Restricted sharing. ### 6. E-Commerce: Shopify Shopify provides online stores and retail point-of-sale systems to businesses across the globe. The built-in tools make it easy for even the smallest startup to build a store and begin selling to customers. In addition to storefront builders, the site also provides payment processing services and marketing tools. Offer: The Shopify Partner Program offers commissions of up to 20% of a referred merchant’s monthly subscription. The platform also offers limited-time offers like a current (at time of writing) offer to earn $500 for every referred client in addition to the 20% commission. How it works: Shopify encourages members to participate in the platform in a variety of ways, including referring new members, building apps, and developing themes. The exact amount you can make depends on where you’re located, what subscription tier you’re on, and the types of contributions you make. How to enroll: Current Shopify customers can sign up by joining the Shopify Partner Program. Once accepted, you’ll have access to a dashboard that lets you manage referrals and submit apps or themes. Pros: - Recurring commissions. - Bonus incentives. - Partner dashboard. Cons: - Approval required. - Variable payouts. ### 7. Cloud Storage: Dropbox Dropbox has become one of the top ways to transfer files via the cloud. The simple user interface and vast product integrations have it dominating the space, even as contenders like Google Drive and OneDrive grab a share of the market. Offer: Earn 500 MB in free storage when you successfully refer someone. If you’re a paid user, you’ll earn 1 GB for each successful referral. Once the person you refer creates an account, both you and the person you refer will see the extra storage in your accounts. How it works: Refer a friend, family member, associate, or anyone else to Dropbox and you’ll both receive free storage. You’ll get the maximum reward if you’re a paid member, which costs $9.99 a month. How to enroll: You don’t have to sign up for a program to invite others to join Dropbox: Simply log in to your account via either the website or mobile app. You’ll then click Refer a Friend under Settings. You can also copy the invite link and send it to others. Pros: - Instant rewards. - No application. - Two-sided incentive. Cons: - No cash payouts. - Low reward ceiling. ### 8. VPN Service: NordVPN Privacy is a serious concern these days, and virtual private networks (VPN) provide some much-needed protection. NordVPN is one of the top contenders in the VPN space, offering accounts for as little as $3.09 a month. Offer: NordVPM has recently shifted its emphasis to an affiliate model that rewards partners with revenue share based on subscription length. How it works: When someone joins using your referral link, you’ll automatically get three free months of service. Your friend will get three free months for choosing a 1- or 2-year plan, but only one free month for signing up to pay month by month. How to enroll: Sign in to your NordVPN account and choose Refer a Friend from the options on the left. You can also find the referral link under Settings in the mobile app. Pros: - Revenue-share model. - Two-sided rewards. - High consumer demand. Cons: - Service credits only. - Limited long-term value. ### 9. Web Hosting: Bluehost Bluehost is one of the top web hosting and domain registration services, largely known for its WordPress integrations. Its setup attracts small businesses, bloggers, and e-commerce startups looking for affordable and easy-to-use hosting. Offer: Earn $65 per qualified sale. How it works: Bluehost’s referral setup is more of a traditional affiliate model, with members placing links and banners on their website to entice clicks. Once you’ve set up your links, you can use Bluehost’s built-in marketing tools to boost your success. How to enroll: Bluehost customers simply sign up to get a referral link. From there, you can browse a library of advertising images, including banners and ads. You’ll then place the image(s) on your website to start attracting referrals. Pros: - Cash commissions. - Affiliate tools included. - Easy link placement. Cons: - No recurring revenue. - High competition. ### 10. Online Education: Thinkific Thinkific is a popular platform that lets educators and creators build and sell online courses. The built-in course creation tool walks educators through building a class and launching it. Both educators and businesses use Thinkific to host courses they promote to their target audiences. Offer: Earn 30% lifetime commissions on all referrals who sign up for an annual or monthly paid plan. How it works: When someone clicks on your referral link and signs up for Thinkific, you’re tagged as the referrer. You’ll then earn a commission on that person’s monthly or annual plan payment. Thinkific caps annual commissions at $1,800. How to enroll: When you join Thinkific, you’ll receive a welcome message on PartnerStack. This message includes a referral link that you can then use to invite others. Pros: - Lifetime commissions. - Recurring payouts. - Creator-focused audience. Cons: - Annual commission cap. - Partner approval required. ## Key Takeaways Referral programs serve as one of the most affordable ways to scale customer acquisition in 2026. In digital advertising, programs that pay cash instead of ad credit stand out, but marketers may also want to look for options that are open to non-members. ## Frequently Asked Questions (FAQs) ### What are the typical conversion rates for referrals in different digital advertising niches? Conversion rates vary widely, but for performance-driven campaigns, conversion rates of 3% to 10% are typical. Rates tend to be higher in some industries than others. In SaaS, referral conversion rates are often significantly stronger than in e-commerce. It’s common to see SaaS referral conversions in the 15% to 30% range, compared with 5% to 10% in e-commerce. The difference is largely driven by buyer intent. SaaS audiences are often already familiar with a product’s value before recommending it, which makes referrals more deliberate and more likely to convert. ### What are the eligibility requirements to join different referral programs? While Realize’s cash-based referral program is open to everyone, that isn’t the norm. In most cases, you’ll first need to be an existing customer or registered partner. You may also need to apply and wait for approval, which means a program administrator will ensure your application meets all requirements, including that you’re the minimum age. For more selective referral programs, applications are typically reviewed to ensure brand alignment and compliance. This often means providing links to websites or social media accounts, outlining general promotional strategies, and explaining where and how the product or service will be promoted. ### Which referral programs are best suited for SMBs? Small and midsize businesses (SMBs) can be limited by their budgets and short history. For that reason, it’s best to target referral programs with low barriers to entry, no spend commitments, and cash-based incentives. Programs that perform best for SMBs tend to minimize complexity. A single, clear incentive — such as a discount or free month — combined with automated delivery can reduce friction while also encouraging participation. When the offer is closely aligned with user intent, these referral programs commonly see conversions as high as 12%. ### Why have referral programs become so popular, especially in the SaaS world? Referral programs have emerged as a great way to bring in revenue while also reaching new audiences. This is especially beneficial for SaaS-based businesses, which have seen rising customer acquisition costs and ad fatigue in recent years. In crowded digital environments, referrals scale trust more effectively than traditional paid advertising. Recommendations from peers feel more authentic than ads, helping brands move past skepticism and drive higher-quality conversions. Referral Program Industry Offer Realize Performance Advertising Earn $1,000 cash when someone you refer spends $5,000. MailChimp Email Marketing Earn $30 in bill credits for each friend you successfully refer to the MailChimp platform. Notion SaaS Earn $50 per referral plus 20% of year-one revenue for each person you successfully refer. FreshBooks Cloud Accounting Earn a $100 account credit for each new user you bring to the platform. SoFi Banking and Fintech Rewards ranging from $50 to $1,500 (e.g., $300 for personal loan referrals). Shopify E-Commerce Commissions of up to 20% of a referred merchant’s monthly subscription, plus limited-time offers. Dropbox Cloud Storage Earn 500 MB (free user) or 1 GB (paid user) in free storage for each successful referral. NordVPN VPN Service Revenue share based on subscription length. Bluehost Web Hosting Earn $65 per qualified sale. Thinkific Online Education Earn 30% lifetime commissions on all referrals who sign up for an annual or monthly paid plan. --- ### Pay-Per-Click: Understanding The Role of PPC URL: https://www.taboola.com/marketing-hub/pay-per-click/ Last Modified: 2025-06-30 08:58:32 Pay-per-click (PPC) advertising is one of the most effective ways to achieve your campaign goals, whether you’re trying to drive traffic to a website, generate leads, or increase sales. To maximize your ad spend, you need to understand how to create and manage an effective PPC campaign. In this guide, I’ll explain how PPC works, explore the different PPC ad types, review major ad platforms, and share some practical tips for setting up and managing your next PPC ad campaign. ## What Is Pay-Per-Click (PPC)? Pay-per-click is a type of digital advertising where advertisers pay for each click on their ads. It’s essentially a way to purchase visits to your website or app, rather than attempting to drive traffic through organic search. ## How Does PPC Work? PPC operates on an auction-based system. Advertisers bid on specific keywords or audience segments. When a user searches for a keyword or content related to a keyword, the ad platform runs an auction to determine which ads to display to the user. Here’s how the PPC auction process might look: - Keyword targeting: Advertisers select keywords that they wish to target. - Bid submission: Once the keywords are chosen, advertisers set the maximum amount, or bid, they are willing to pay for a click. - Ad quality evaluation: The ad platform assesses the relevance and quality of each ad. - Ad placement: Ads appear based on the combination of the bid amount and quality score. The beauty of PPC advertising is that if you can create a high quality ad that matches the user’s intent, you can win a higher placement at a lower cost than other advertisers. ## Why Is PPC Important? PPC has several advantages that have made it a critical element of most digital marketing strategies. For starters, PPC ads can generate traffic almost immediately — SEO, while vastly cheaper, can take months to see results. PPC ads also target specific audience segments, which ensures ads are delivered to users looking for specific solutions. Because PPC ad platforms include in-depth tracking and analytics, you can easily measure return on investment (ROI) by evaluating your cost-per-click (CPC), conversion rates, and return on ad spend (ROAS). Finally, you can quickly scale PPC campaigns up or down, allowing you to tailor them to your business’ specific needs. ## What Are the Different Types of PPC Ads? ### Search Ads The most common type of PPC ad is the search ad. These appear on search engine results pages, such as Google SERP, when users enter relevant keywords. Search ads are driven by keyword targeting and are a great way to capture high-intent search traffic. ### Display Ads Display ads are ads that use images, videos, and text to draw viewers. They often include a call to action button and can be placed in different areas of a website, social feed, or app. Display ads also come in various sizes and dimensions, and are excellent for brand awareness and retargeting campaigns because they combine visual appeal with precise audience targeting. Depending on the display ad network, advertisers can opt for a PPC pricing model, as well as CPM (cost per mille), CPA (cost per action), CPL (cost per lead), and more. ### Native Ads Unlike display ads, which are designed to stand out from the surrounding content, native ads blend seamlessly into website or app content, which can make them hard to distinguish as ads. They often appear in-feed on social media posts, online games, or video streaming platforms. Native ads often use a PPC model, which means you’ll only pay when a user clicks on your native ad, and not for impressions. For an idea of their effectiveness, this case study explains how global sportswear giant Adidas successfully boosted its brand awareness using a native ad campaign powered by Taboola. ### Shopping Ads Shopping ads are designed to showcase store names, product images, prices, and descriptions directly on search engines. They’re highly effective for e-commerce businesses looking to drive sales. When it comes to PPC campaigns, the average CPC on Google Shopping ads varies by industry. However, data provided by Mega Digital in 2023 showed that among popular industries, retail had the lowest CPC at $0.85, whereas automotive was $3.11. The shopping ad below appeared after a search for “New Balance running shoes:” ## Common PPC Advertising Platforms You can run PPC campaigns on a wide range of ad networks, each one offering unique strengths and audiences. ### Google Ads Google Ads is the most well-known PPC platform. Your ads can appear on Google Search, Google Maps, YouTube, Gmail, and the Google Display Network. With more market share than any other ad platform, Google Ads is ideal if you need massive reach across search, YouTube, mobile apps, etc., are in a high-volume market, or are looking to scale quickly. ### Bing Ads Bing Ads, also known as Microsoft Ads, is a cost-effective alternative to Google Ads. While it lacks Google’s market share and reach, it offers a similar interface and less competition, and according to Wordstream, its average CPC rates are 33% lower than Google Ads. ### Realize If you’re looking to expand your reach beyond search and social, you’ll want to consider a performance marketing platform like Realize. The platform allows you to create and target display ad campaigns within minutes, then run them on premium publisher websites with high-intent audiences, such as USA Today, The Weather Channel, NBC, MSN, Business Insider, Bloomberg, and CBS. Realize uses AI and machine learning to target users by intent, rather than just identity, as well as automatically optimizing your PPC campaigns, helping you maximize your ROI. ### Facebook Ads (Meta Ads) Meta Ads run on Facebook, Instagram, Messenger, and the Audience Network, and is one of the most powerful PPC platforms when it comes to targeting, scalability, and A/B testing. You can target users by demographics, interests, and behaviors, and build custom and lookalike audiences. In addition, your PPC campaigns can optimize for traffic, leads, conversions, video views, engagement, app installs, and more. ### Amazon Ads Amazon Ads is ideal for e-commerce PPC advertising within the Amazon ecosystem. Amazon users are already on the site to research or buy specific products, making this a high-intent platform. You can target by keyword, product, product category, or audience segment, e.g., demographics, interests, remarketing, etc. Because purchases happen on Amazon, it's very easy to measure ad performance. ## How to Do Effective PPC Keyword Research One of the most effective ways to run a PPC campaign is by targeting high-opportunity keywords. It takes a lot of practice and testing, but the better you get at this, the more likely you are to reach the right audience. Here are some best practices to follow: ### Brainstorm Ideas Before you dive into keyword research tools or other data, put yourself in your customer’s shoes. What words or phrases would someone use if they were looking for your product or service? This is all about capturing ideas before you filter or optimize. Sites like Answer The Public can be useful for finding keywords adjacent to your main target. ### Use Keyword Tools Once you’ve built an initial keyword list, leverage keyword tools that can provide you with valuable data on search volume, keyword competition, and cost-per-click. While premium tools like Semrush or Ahrefs are very expensive, there are more reasonably priced options, such as Ubersuggest, as well as free keyword tools, like Google Keyword Planner. ### Focus on Intent Remember that not all keywords are created equal. If your goal is to sell a product or service, you’ll want to prioritize transactional or commercial keywords over informational keywords. For example, if you’re selling tennis rackets, you would rather target “where to buy a tennis racket,” (transactional), or “best beginner tennis racket for women,” (commercial). Keywords such as “how to buy a tennis racket” or “how heavy is a tennis racket” are informational keywords that show little purchase intent. ### Identify Negative Keywords Make sure you use negative keywords in your PPC campaign. This will prevent your ads from being shown for irrelevant search queries, and help you avoid wasting ad spend on low-quality traffic. Using the tennis racket example, if you sell high-quality, new tennis rackets, you might want to include negative keywords like “restring,” “stringing services,” “tennis racket repair,” or “pickleball paddle.” ## How to Manage PPC Campaigns Creating a PPC campaign is just the first step. Once your campaign is live, it’s critical that you manage it effectively. Here are some tips to maximize your ad performance: ### Set Clear Goals Your PPC campaign should have clearly defined goals and KPIs. Is your primary objective to drive website traffic, generate leads, make sales, or get return on ad spend? During the campaign, you can track different metrics, like CPC, CPL, CPA, CTR (click-through-rate) and conversion rates, to assess your ad performance against your goals. ### Optimize Your Landing Pages While it all starts with creating a high-quality ad, the great user experience needs to continue on your landing page. Make sure you’re using high-quality visuals, relevant and compelling messaging, and clear calls to action (CTAs) on your landing pages. Also, ensure your landing page loads quickly, as slow load times can lead to higher bounce rates and wasted ad spend. ### Always A/B Test Your Ads A/B testing refers to the process of comparing two different versions of an ad or landing page to see which one performs better with users, based on your campaign goal. By testing different deadlines, product descriptions, CTAs, or even landing pages, you can improve your campaign performance. ### Monitor Your Budget As you identify your top performing ads and keywords, don’t hesitate to reallocate your campaign budget accordingly. Doing so will lower your CPC and improve the performance of your PPC campaign. ## How to Track Your PPC Campaign’s Performance Effective PPC campaigns depend on your ability to understand the data. By tracking your performance, you can avoid the guesswork and make decisions based on results. Here are some ways to do that: ### Track Conversions The major ad platforms all have conversion tracking capability. Take advantage of these features to see which actions users are taking after they click on your ad. Actions can include a newsletter sign-up, webinar registration, form submission, purchase, and more. ### Track Cost Per Click (CPC) Monitoring your Cost Per Click is crucial for understanding the efficiency of your ad spending. CPC tells you how much you're paying for each click on your ad. By tracking CPC, you can identify which keywords, ad groups, or campaigns are costing you more or less per click, allowing you to adjust bids and optimize your budget for better returns. A high CPC might indicate strong competition or a need to refine your targeting. ### Use Analytics Tools such as Google Analytics can help you track user behavior when they land on your website. Consider metrics like bounce rates, time on page, time on site, and pages per session. ### Evaluate Attribution Models Attribution models determine which ads or marketing steps helped someone decide to make a purchase or take an action. This helps you understand what step in your campaign worked best. Common attribution models include last-touch, first-touch, linear, and time decay: - Last-touch attribution: Full credit for the conversion is given to the last action the user took before converting. - First-touch attribution: Full credit for the conversion is given to the first action the user took before converting. For example, if they clicked on a Google Search ad, then later clicked on a Facebook ad to make the purchase, the Google Search ad would get the credit. - Linear attribution: Equal credit is shared across all actions. For example, Google Search Ad → Facebook Ad → Email sign-up. All three touchpoints receive equal credit (33%). - Time decay attribution: The action closest to the conversion gets the most credit, but the other actions still receive some credit. The best attribution model for your PPC campaign will depend on your campaign goals, and the complexity of the customer journey. ## Key Takeaways PPC advertising offers a powerful way to reach high-intent audiences and drive measurable results while you maintain full control over your ad spend. Focus on choosing the right ad platform for your business goals, targeting high-intent keywords, optimizing your landing pages, and continuously testing your ads. Also, remember to monitor your PPC ad campaign’s performance by tracking conversions and making use of the available data. With time, and by following the right strategy, PPC can help fuel the growth of your business. ## Frequently Asked Questions (FAQs) ### PPC vs. SEM vs. SEO: What’s the difference? PPC (pay-per-click) is a form of paid digital advertising where advertisers pay a fee every time an ad is clicked. SEM (search engine marketing) includes both PPC and search engine optimization (SEO), and refers to any marketing effort that involves search engines (Google Ads is a good example). SEO involves optimizing your website for organic, unpaid traffic. ### PPC vs. CPC: What Is The Difference? PPC (Pay-Per-Click) is an advertising model where businesses pay a fee each time a user clicks on their online ad, essentially "buying" website visits rather than earning them organically. CPC (Cost-Per-Click), on the other hand, is a specific metric within the PPC model, representing the actual amount an advertiser pays for each individual click, making it a vital measure for evaluating campaign efficiency and cost. Feature PPC (Pay-Per-Click) CPC (Cost-Per-Click) Nature An advertising model or strategy A specific metric or cost Scope The overall system of paying for clicks The actual price paid for each click within that system Role How you acquire traffic to your website How much each unit of that traffic costs you Application Setting up campaigns, choosing platforms, targeting, etc. Analyzing campaign performance, optimizing bids, managing budget ### What are cost caps? Cost caps are bidding strategies that help you control your average costs by allowing you to set the amount you’re willing to pay per action or conversion. For example, let’s say you’re using cost cap bidding on Facebook Ads. If your target cost per lead is $10, you can set that as your cost cap. Facebook will then attempt to generate as many leads as possible at or below that amount. ### What is targeted reach? Targeted reach refers to the number of users your ad can reach within a specific audience segment. When using PPC platforms, you can target by location, interests, behavior, and more, increasing the chances of your ad being seen by the right people at the right time. --- ### Targeting: What It Is, How It Works URL: https://www.taboola.com/marketing-hub/targeting/ Last Modified: 2026-06-24 12:25:37 If you were out fishing for striped bass, would it be best to use a three-way swivel rig or a trawling net? For those of you who aren’t anglers, I’ll just tell you right away that it’s the former, the three-way swivel rig, which is one of several fishing rigs often recommended for landing stripers. Now, a trawling net may well catch a few striped bass, but it will also haul in cod, shrimp, mullet, hake, rocks, trash, an old boot, and so much more. Long story short, you’d be casting much too wide of a net — literally — to accomplish your goal, and you’d be wasting a lot of time and effort catching all that unwanted stuff. When it comes to digital marketing, if you’re not targeting your audiences properly, then you’re likely casting too wide of a metaphorical net, and you might not even be catching your desired audience along with all the chum. ## Understanding Targeting in Digital Advertising When it comes to digital advertising, targeting refers to the practice of identifying and reaching specific groups of consumers who are most likely to be interested in a particular product, platform, service, and so on. It's a core element of effective online and in-app marketing strategies, as it allows advertisers to deliver relevant messages to the right audiences in the right places at the right times, thereby maximizing the impact of their ad campaigns and minimizing wasted ad spend. ### What Are the Different Levels at Which You Can Target Audiences? Audience targeting in digital advertising can be applied at multiple levels within a larger campaign structure. Generally, you can target audiences at the whole campaign level with some materials, as part of a smaller ad group, and with further refined ad levels. This allows for ever more granular targeting, from broad campaign-level targeting to highly specific ad-level targeting. Effective targeting is crucial for successful marketing campaigns because it allows advertisers to focus their efforts on the most likely prospects, resulting in higher engagement, more effective messaging, and ultimately a better return on investment (ROI). By understanding your audience's needs, interests, and behaviors, you can tailor your campaigns to resonate with them, leading to increased conversions and stronger brand relationships. ## Types of Targeting Options Today’s performance advertisers have access to a sophisticated suite of targeting strategies designed to reach users based on who they are and how they act online, scale, and optimize for conversions and lower CPAs in real-time. ### What Are Demographic Targeting Options? Just as the phrasing suggests, demographic targeting options in marketing include targeting audiences based on their age, gender, location, income, education levels, and more. These factors help marketers create more relevant and personalized campaigns that resonate with their intended audience. This is similar to the concept of audience segmentation. ### What is Broad Targeting? Broad targeting is a dynamic advertising strategy that shifts the heavy lifting of audience selection from the marketer to the platform's machine learning. Instead of manually defining strict demographics or interests, you allow the algorithm to serve as your primary scout. By operating with minimal restrictions, the system enters a phase of continuous testing and iteration. It analyzes real-time user behavior to identify who is most likely to engage with your content. Essentially, you define the objective—such as sales or lead generation—and the platform’s AI scours the landscape to match your ad with the users most likely to convert. ### What Are Interest-Based Targeting Options? Interest-based targeting options in online advertising allow marketers to reach users based on their interests as established by online behavior. This is achieved by tracking user activity, including website visits, content consumption, social media interactions, and other data points. ### What Are Behavioral Targeting Options? Behavioral targeting is a combination of techniques that use people's actions — like website interactions, purchase and browsing histories, and more — to deliver highly relevant marketing messages. The more information marketers get from tracking their audience, the better they will know their behavior online. ### What Is Contextual Targeting? Contextual targeting is an advertising technique that displays ads based on the content of a web page or application. It leverages the context of the web page, including keywords, topics, and overall content, to show ads that are relevant to the specific content the user is viewing. ### What Is Remarketing or Retargeting? Remarketing, also known as retargeting, is a digital marketing technique where businesses re-engage with individuals who have previously interacted with their website or brand. It involves displaying ads to these users on other websites, social media platforms, or even within email campaigns, aiming to remind them of their previous interaction and encourage further engagement or a conversion. ### What Are Lookalike Audiences? Lookalike audiences, also known as similar audiences, are a type of audience targeting used in online advertising. They are created by using algorithms to identify individuals who share similar characteristics to a source audience you provide, such as your existing customers. As opposed to broad targeting—which relies on the platform to find an audience from scratch—lookalikes use your specific data as a roadmap. This allows you to reach a wider audience of people who are likely to be interested in your products or services, without relying on traditional demographic targeting. ### What Is Device Targeting? Device targeting involves tailoring content or actions based on the specific device being used by the audience. This can include targeting device types like smartphones, tablets, or computers, as well as factors like operating system, model, or even carrier. ## How Targeting Improves Ad Performance Targeting advertising campaigns optimizes your budget by ensuring your ads are shown to the most relevant audiences, reducing wasted ad spend and increasing efficiency. By reaching the right people with the right message, you can improve ROI and achieve higher conversion rates, making your advertising budget work harder. ### How Does Targeting Help You Reach the Right Audience? Targeting helps with reaching the right audience by enabling businesses to tailor their messaging and channels to specific groups based on their demographics, interests, behaviors, and location. It essentially increases ad relevance and engagement by ensuring ads are seen by the intended target audience, making them more likely to resonate and allowing for more efficient and effective marketing campaigns, leading to higher engagement and conversion rates. ### How Can Targeting Improve Click-Through Rates (CTR)? Targeting helps improve click-through rates (CTR) by ensuring your ads reach the right users, leading to more relevant clicks. By refining targeting criteria and using tools like audience segmentation, you can narrow your reach to individuals who are more likely to be interested in your message, thereby boosting engagement. ## Ethical Considerations in Targeting ### What Are Some Ethical Concerns Related to Certain Targeting Practices? Certain targeting practices raise ethical concerns due to potential privacy violations, discrimination, and manipulation of consumer behavior. Concerns include the use of personal data without consent, algorithmic bias leading to unfair targeting, and the exploitation of vulnerable groups. It can also be a serious problem if marketers target children or people with reduced mental faculties. ### How Do Privacy Regulations Impact Targeting Options? Privacy regulations, like the GDPR (General Data Protection Regulation) and the CCPA (California Consumer Privacy Act), impact targeting options by limiting the ability to use extensive data collection and tracking for personalized ads. This forces marketers to adapt their strategies and explore alternative methods, like contextual advertising and first-party data. ### What Is the Importance of Transparency In Ad Targeting? Transparency in ad targeting is crucial because it fosters trust, builds stronger relationships with consumers, and ensures ethical and legal compliance on behalf of the marketers. It helps consumers understand why they are seeing specific ads, allows them to make informed decisions about their online experiences, and helps them to gain confidence in the advertising platforms and the people behind them. Transparency also benefits advertisers by helping to avoid potential legal issues or consumer backlash. ## Best Practices for Effective Targeting Testing and refining targeting strategies is crucial because it helps ensure that marketing efforts are focused on the right audience, resulting in more effective campaigns, improved ROI, and a deeper understanding of customer preferences. This process allows marketers to identify what works and what doesn't, leading to more efficient resource allocation and better customer experiences. You should review and adjust your targeting settings regularly, ideally at least weekly. This allows you to stay on top of ad performance, identify potential issues early, and make necessary adjustments to your strategy. More detailed reviews, such as monthly or quarterly, can help you assess progress toward longer-term goals and make more significant adjustments. ### How Do You Align Your Targeting with Your Campaign Goals? To most effectively align targeting with campaign goals, start by clearly defining your goals and make sure you have a true understanding of your target audience. Then, choose the right channels and formats to reach them, and continuously monitor and evaluate your campaign's performance to make necessary adjustments. To avoid overly narrow or overly broad targeting, a balanced approach is crucial. Start with a broader targeting strategy to assess the potential reach and effectiveness, then refine it based on performance metrics, insights, and audience segmentation. Consider factors like demographics, interests, and purchasing behaviors when defining your target audience to retarget them effectively.. ### How to Combine Different Targeting Methods for Better Precision To combine targeting methods for better precision, marketers can leverage a multi-faceted approach, including aligning audience segments across platforms, using first-party data, and refining targeting based on performance. By layering and refining different targeting methods, you can create hyper-specific campaigns that resonate with your audience. ## Key Takeaways Targeting in advertising offers significant benefits, including increased personalization, improved ROI, and enhanced customer engagement. By delivering relevant messages to specific audiences, businesses can optimize resource allocation, boost conversion rates, and foster stronger customer relationships. Delivering relevant content to the right audience can also enhance brand recognition and improve overall brand perception, while reducing waste and maximizing the impact of advertising spending. ## Frequently Asked Questions (FAQs) ### What is the difference between audience targeting and content targeting? Audience targeting focuses on reaching people based on who they are (demographics, interests, behaviors), while content targeting focuses on reaching people based on the content they are engaging with (specific websites, keywords, apps, and so on). Think of audience targeting as fishing for specific types of fish with a unique bait, and content targeting as fishing in specific locations where those fish are likely to be found. ### How can I determine the best targeting options for my business? To determine the best ad targeting options for your business, you need to understand your target audience, their needs, and how your business can best address them. This involves analyzing your existing customers, researching the market, and conducting competitor analysis. By focusing on the most relevant and responsive segments, you can maximize your marketing efforts and achieve better results. ### Can targeting too narrowly limit my reach? Yes, targeting too narrowly can significantly limit your advertising reach and effectiveness. While precise targeting can improve ad relevance and potentially lower costs, over-refining your audience can result in a smaller pool of potential customers and reduce overall ad delivery. This can hinder your ability to achieve broader awareness, especially for products or services that appeal to a wider audience. ### What are some emerging trends in ad targeting? The most notable emerging trends in ad targeting include a strong focus on AI and machine learning for personalization and automation, the rise of short-form video ads, and the increasing importance of data privacy and ethical marketing practices. --- ### UTM Codes: Urchin Tracking Modules Explained URL: https://www.taboola.com/marketing-hub/utm-code/ Last Modified: 2026-04-12 07:01:16 The term Urchin Tracking Module may not be part of everyone’s daily vocabulary — certainly, it doesn’t roll off the tongue, and seeing one can look like reading a new language. But, what’s more commonly known as a UTM code, UTM parameter, or UTM tag, tells a data-rich story for digital marketers to dive into and dissect. ## Understanding UTM Codes A UTM code is the long string of information attached to a URL following a domain (or host, frequently ending in dot com) and, if applicable, path (child or sub page from homepage). Starting with a question mark and including letters and characters, this code tracks the effectiveness of different marketing tactics. ### What Is the Purpose of Using UTM Codes in Digital Marketing? Within the UTM parameters are words that signify where online the user came from, and how they interacted with different digital marketing products. Information gained from a UTM tag is used to better understand how different tactics drive users to the page, behaviors of specific users, and overall user trends. The key to gleaning these insights is feeding the information from a URL string into a web analytics tool, such as Google Analytics, or a measurement tool or dashboard within a marketing or advertising solution. ### How Do UTM Codes Help Track the Performance of Marketing Campaigns? Marketing teams determine the UTM codes that are used, meaning there are a finite number of parameters, and each tag is determined by marketers themselves. Think of this as limited, structured data being collected, as opposed to open-ended information. The information being collected is pre-determined based on how teams want to track results and parse the data collected for attribution. Examples of how UTM codes can provide actionable analytics include: - More precisely pinpointing traffic sources: Analytics often attribute how someone landed on your site either as direct (e.g., a user typed the URL into a search bar) vs. referral (clicked from another site). Through the use of multiple UTM codes or vendor/platform terms, marketers can hone in more granularly on how someone came to a page. - Comparing campaign effectiveness by type: Have a hunch that email drives more traffic than another method, or that owned content is outperforming paid? With UTM tracking, the proof is in the parameter. - Attributing a campaign across media: With multiple tags, you can track information in different ways to be able to slice and dice data more powerfully. - Measuring test results: See performance across different variables or options within a campaign or page (for example, how different calls to action or link placements may attract click-throughs). By tracking the performance of your marketing campaigns with UTM codes, you can gain valuable insights into how your customers are interacting with your brand and which specific channels and campaigns are most effective. While UTM codes provide structured manual data, certain performance advertising platforms like Realize now optimize results using codeless tracking, a no-code solution that allows marketers to create event- and URL-based conversions directly within Realize for faster, more accurate performance measurement without technical setup. ## The Five UTM Parameters UTM codes consist of three required and two optional parameters. While campaign source, medium, and name are needed, including the other two (campaign content and term) is a good practice. As a quick reference, the five UTM parameters are: 1. Campaign Source (utm_source) Required parameter to identify the source of your traffic (e.g., Google, newsletter, Taboola). 2. Campaign Medium (utm_medium) Required parameter to identify the medium through which the link was distributed (e.g., search, email, CPC). 3. Campaign Name (utm_campaign) Required parameter to identify a specific product promotion or strategic campaign (e.g., spring sale). 4. Campaign Content (utm_content) Optional parameter to differentiate campaigns for A/B testing and content-targeted ads. 5. Campaign Term (utm_term) Optional parameter used to note keywords for paid search campaigns. ### What Is UTM_Source and What Information Does It Track? The most important parameter, the Campaign Source (which will show in the URL as utm_source), is the most important of all sources, since it identifies your traffic’s origin. A required parameter, it tells where a click came from, be it a site, vendor, or platform. This could show up in the URL as terms like: - Website. - App. - Newsletter. Example: utm_source=taboola ### What Is UTM_Medium and What Does It Identify? Campaign Medium tag utm_medium notes the channel where the link appeared to the user. Examples could be a paid ad, organic placement, email, social media, and so on. This could show up in the URL as terms like: - cpc. - Email. - Social. - Ad type. Example: utm_medium=social ### What Is UTM_Campaign and How Is It Used? When you want to know why a click occurred, you will look to the Campaign Name parameter, utm_campaign. This required tag identifies and tracks a specific marketing initiative. If you’re promoting a Black Friday sale through various tactics, this code will tie the different efforts together across sources and mediums. This could show up in the URL as terms like: - Product name. - Sale name. - Promo code. - Welcome bonus. Example: utm_capaign=informational-article ### What Is UTM_Term and When Is It Typically Used? An optional tag, Campaign Term shows up in UTM parameters as utm_term. This can be used in paid search campaigns to distinguish which keyword(s) drove the click to conversion. Marketers can view these analytics to see which keywords are effectively bringing traffic to the page. This would show up in the URL as the keyword used in a paid ad. Example: utm_term=utm-codes-explained ### What Is UTM_Content and How Can It Differentiate Ads or Links? Campaign Content (utm_content) is an optional tag that tracks differences within an ad or piece of content. It’s often used with A/B testing, where the tracking term is tied to a specific element and version. This could show up in the URL as terms like: - Version letter. - Text element. - Image name. - Button color or placement. Example: utm_content=headline-a ## How UTM Codes Work UTM codes take the tags of structured data that you’re using for each parameter to create a complete URL. For the example above, a URL with all five UTM codes would look similar to this: https://www.yourlandingpage.com/page-title?utm_source=taboola&utm_medium=social&&utm_capaign=informational-article&utm_term=utm-codes-explained&utm_content=headline-a When a user clicks on this full URL, the data is captured and sent into your analytics platform. ### How Do You Create URLs With UTM Codes? While the formula above will work when creating UTM codes, it’s best practice to use a UTM builder tool. Such builders are available in marketing automation or ad platforms, analytics tools, and even Google. These tools will ask for the parameters of the UTM codes you want to include, then create the URL, including all the characters in the correct spots. This reduces risks of manual or human errors, misspellings, misplacements, and the like. ### Where Do You Typically Use UTM-Tagged URLs? Think of UTM tracking like this: Any time you could use more information about at least the three required UTM codes (Campaign Source, Medium, or Name), it’s worth considering UTM-tagged URLs. Marketers often implement this tactic when they’re running CPC or other paid media initiatives, sending communications around a promotion, directing people in influencer or affiliate marketing, providing a URL from an external source, or driving clicks to a campaign landing page. ### How Do Analytics Platforms (e.g., Google Analytics) Interpret UTM Parameters? When a user clicks a UTM code, they are redirected to the page URL that comes before the question mark and UTM parameters. Those UTM tags are then fed into your analytics tool and captured as user behaviors. Marketers can then use this information to better understand user behavior, and inform future strategy. While Google Analytics (both GA4 and UA) is a popular analytics platform, others — like Kissmetrics or Adobe Analytics — also support capturing UTM code data, as do CRMs, like HubSpot. To view UTM data in your analytics report, go to the analytics dashboard and find a report that captures behavior or sources. In Google Analytics, navigate to the Acquisition > Traffic acquisition report to view results for Session source/medium, Session medium, Session source, and Session campaign. ## Benefits of Using UTM Codes ### How Do UTM Codes Enable Accurate Campaign Tracking? A UTM tag feeds into your analytics platform, capturing additional data that can be viewed to better tie marketing efforts to user actions. Generally speaking, integrating UTM codes into your marketing strategy allows for greater granularity when analysing your traffic — the drivers and sources, what works best, and so on. ### How Do UTM Codes Help in Identifying the Specific Sources of Website Traffic and Conversions? UTM codes allow you to infer the impacts that either a single campaign parameter or multiple together have on conversions, and slice and dice that information into different views. For example: Does a Black Friday promotion perform better on email, and a demo video drive people to your site from social? Do more people click a button that has the words “save now” vs. “limited offer?” With these UTM codes, you have data-backed evidence to work with. ## Best Practices for Using UTM Codes Using UTM codes early and consistently will help with the quality of the data you’re collecting. Take time to develop a strategy (what tags to include, when to use UTM parameters, how the information will be most helpful) to ensure you are collecting useful data. ### How Should You Ensure Consistency in Your UTM Tagging Conventions? Get the most out of your UTM tagging practice by sticking to the strategy you’ve created. Changing direction or tactics can cost accuracy in reporting, time correcting course, or money in ineffective ad spend. Remember, consistency doesn’t have to be complicated to be effective: It can be as simple as appointing someone to oversee UTM governance and capturing it in a spreadsheet for all to reference. Document when and why the various parameters are used, and which tags are acceptable. If you use Google Tag Manager, consider documenting that strategy alongside the UTM tag strategy. ### How Can You Avoid Common Errors When Creating UTM Codes? When an error is introduced in a UTM code, it can be difficult to undo without possibly losing some data. Consider the adage “measure twice, cut once” when creating UTM codes and URL strings: Once it’s published online, the URL is out there, even if you think you deleted it before anyone has clicked. Manually creating the URL string can introduce human error, so as suggested earlier, consider a URL builder. Check, and check again that the correct URL string is in the right placement so you aren’t gathering incorrect data, and only use UTM codes for external links. ### What Tools Can Help You Generate and Manage UTM URLs? Check with your media platform or other tools you’re already using, or generate UTM URLs with Google’s Campaign URL Builder. Document the URLs you create and where they are used. As you learn about a tag’s success, include that information: These findings could be in the UTM spreadsheet you may have created, with different tabs or color coding. ## Key Takeaways UTM codes serve as data-rich extensions to a URL. They consist of three required parameters (source, medium, and campaign name) and two optional parameters (content and term) to capture additional information that’s fed into an analytics platform. These tags help marketers better segment and understand user behavior, including details on what drove a person to your website. This information can better inform future strategies by understanding what channels, versions of a piece of marketing, or incentives contributed to their actions. ## Frequently Asked Questions (FAQs) ### Are UTM codes case-sensitive? Yes, UTM codes are case-sensitive. It’s a best practice to always use lowercase to prevent data fragmentation. For example, if you generally use “Taboola” as your source, but a tag becomes “taboola,” these will become two different sources in reporting. ### How long do UTM parameters track data? While the URL with the UTM parameters can live on, analytics tracking tools often have a limited time (say, six months) where they will track the initial source. After that, a new conversion may begin for a user. ### Can I use UTM codes for internal links? No. UTM codes are meant to track information coming into a site from outside sources. Therefore, using these parameters internally could skew tracking and reporting of user behavior, and not attribute a source correctly. ### What is the difference between UTM codes and other tracking parameters? UTM codes are generally considered to allow marketers more control over how and what is tracked, given that these are the people who set the tags and determine when and how to use them, and can easily be captured in an analytics tool. Other options, such as cookies, log files, or IP tracking, each have their strengths and weaknesses, but also offer less flexibility in customizing what’s collected. --- ### Viewable Cost Per Mille: vCPM Explained URL: https://www.taboola.com/marketing-hub/viewable-cost-per-mille/ Last Modified: 2025-06-09 11:36:29 In online marketing, the acronym CPM stands for “cost per mille,” which translates to cost per thousand impressions (mille being the Latin word for thousand). CPM is a common advertising metric where an advertiser pays a predetermined amount for every 1,000 times their ad is served to a user, thus generating an impression. CPM is often used in campaigns focused on brand awareness, and to measure the impact ads have when reaching large audiences. vCPM is a more precise and actionable metric than basic CPM. It stands for “viewable cost per mille,” which is to say the price a marketer will pay for 1,000 ads that are actually seen by people online (or on an app), not just the people to whom it was served. vCPM is actionable because when you know your ads were actually seen, you can track their actual efficacy, or the lack thereof. ## Understanding Viewable Cost per Mille With so many different metrics (and terms) to keep a handle on in the marketing world, it can get confusing quickly. I’ll break down vCPM in a step-by-step way to help you understand it: ### How Does vCPM Differ From Standard CPM? Think of it like this: A bus with an advertisement for a local dentist drives by a busy street corner on which 15 people are crowded, waiting to cross when the light changes. Ten of the people are looking at their phones, one is staring at the clouds, and four see the ad on the side of the bus. That’s 15 people to whom an ad was served, but only four actual views. ### Why Is Viewability an Important Metric in Display Advertising? Viewability is a crucial metric in online display advertising because it’s the measure of whether an ad is actually seen by the intended audience, ensuring that advertising spend is not wasted on impressions that have no chance of being noticed. If an ad isn't viewable, it cannot deliver its message, rendering the impression ineffective. ## How Viewability Is Measured vCPM is not a one-size-fits-all way of tracking ad service and performance, so again I’ll go step-by-step. And, a quick note: Often ads that count as “viewable” were not necessarily actually seen by human eyes, but the chance they were is good enough for them to cross the bar. ### What Are the Industry Standards for an Ad to Be Considered "Viewable?" An ad is generally considered "viewable" if at least 50% of its pixels are visible to the user for a specific duration, typically one second for display ads and two seconds for video ads. This standard, defined by industry organizations like the IAB (Internet Advertising Bureau) and MRC (Media Rating Council), aims to ensure that advertisers are paying for impressions that are actually seen by their target audience. ### What Technologies and Methods Are Used to Track Ad Viewability? Ad viewability is tracked using a combination of technologies and methods. These include things like JavaScript libraries, ad verification tools, and industry standards like the Open Measurement SDK (OM SDK). Key techniques involve measuring pixel geometry, page geometry, and scrolling behavior, as well as leveraging server-side measurement. ### How Do Different Advertising Platforms Report on Viewability Metrics? While different advertising platforms utilize somewhat different methods and standards for reporting on viewability metrics, they generally focus on the percentage of impressions that are deemed "viewable." As explained above, viewability is often measured as a percentage of total impressions that meet specific criteria for being in view, typically a minimum of 50% of the ad being visible for a certain duration. ## Benefits of Using vCPM Bidding vCPM bidding offers several advantages for advertisers by focusing on viewable impressions, leading to a better understanding of an advertisement’s effectiveness and improved ROI (return on investment). It allows advertisers to pay only when their ads are actually seen, reducing waste and ensuring campaigns are reaching the intended audience. ### Why Would Advertisers Choose to Bid on a vCPM Basis? Advertisers choose viewable cost per mille bidding when their primary goal is brand awareness or exposure, rather than immediate conversions or sales. It allows them to set a maximum price they'll pay for each 1,000 times their ad is actually seen, not just displayed. This is particularly useful for new product launches or campaigns targeting a wide audience to build brand recall. ### How Does vCPM Help Ensure That Advertisers Only Pay for Ads That Have a Chance to Be Seen? vCPM helps ensure advertisers only pay for ads with a chance of being seen by defining a “viewable” impression as one where at least 50% of the ad is visible on the screen for at least one second for display ads, or two seconds for video ads, as noted before. This contrasts with traditional CPM, which pays for every impression, regardless of whether it's actually seen by the user — or very likely seen, at any rate. The vCPM model incentivizes publishers to place ads in viewable positions and rewards them for impressions that are actually seen by users. The overall effect is an improvement in display campaign efficiency and ROI by ensuring ads are actually seen by the target audience, leading to more effective engagement and better budget allocation. However, vCPM is not available for all types of display ads. Specifically, it is not available for Search Network only campaigns. vCPM is primarily used for display and video ads, particularly within the Google Display Network. ## Considerations and Potential Drawbacks of vCPM ### Is vCPM Always the Best Bidding Strategy for All Campaigns? vCPM is not always the best bidding strategy for ad campaigns. It’s most effective for campaigns focused on brand awareness and visibility, where the goal is to increase the number of people who see the ad. It's less suitable for campaigns that prioritize conversions or clicks, as vCPM focuses on viewable impressions rather than user engagement. ### How Might vCPM Impact Ad Inventory and Reach? Focusing on viewable cost per mille can impact ad inventory and reach by focusing on the visibility of ads rather than just their placement. This can lead to a more valuable and impactful ad inventory, as users are more likely to engage with ads they can actually see. ### Are There Potential Discrepancies in Viewability Reporting Across Different Platforms? Discrepancies in ad viewability reporting across different platforms are a common issue in the digital advertising world. This is due to various factors, including different methodologies for measuring viewability, variations in impression counting practices, and the use of different tools and technologies. ### What Are Some Factors That Can Influence Ad Viewability? Many factors can impact the viewability of ads, including ad placement, page load speed, user behavior, ad size, and device type. Ads placed "above the fold" (visible without scrolling) and larger, vertical ad units generally have higher viewability rates. Faster loading pages and more engaging content also contribute to higher viewability. Achieving high ad viewability rates presents several challenges, including ad fraud, ad placement, non-human traffic, and technical factors like page load speed and ad design. ## Implementing vCPM in Advertising Platforms ### How Do You Set Up vCPM Bidding in Platforms Like Google Ads? To set up vCPM bidding in Google Ads (or other ad platforms), you'll need to choose a bidding strategy that focuses on viewable impressions, set your maximum bid, and then potentially customize bids at the ad group or placement level. Customizing bids can involve setting the same maximum vCPM bid for all keywords and placements within a specific ad group or, for more precise control, you can also set individual vCPM bids for specific placements. ### What Are the Typical Settings and Options Available for vCPM Campaigns? vCPM campaigns offer settings and options focused on maximizing the visibility and impact of your ads. These include targeting, bidding, and creative optimization. Targeting can refer to audience targeting, where you reach out to people based on demographics, location, interest, and so on; device targeting, where you focus on mobile or desktop devices; keyword targeting, where you tie your ads to search queries, and more. ### How Should You Monitor and Adjust Your vCPM Bids? To effectively monitor and adjust vCPM bids, regularly analyze campaign performance metrics like viewable impressions, click-through rates, and conversions. Adjust bids and creative strategies based on this data to optimize your ROI. Always consider implementing A/B testing, seeing how different attribution models are working — or not working. ## Key Takeaways vCPM stands for viewable cost per mille and is a pricing model where advertisers pay based on the number of times their ad is actually seen (i.e., is considered viewable) by users, as opposed to just the number of times the ad is merely placed. Unlike traditional CPM, which charges for every 1,000 impressions regardless of whether they are visible, vCPM focuses on the true value of viewable impressions. That said, even ads considered “viewed” might not have been actually seen. Display ads are considered viewed after being at least 50% visible for one second, while video ads are considered viewed when allowed to play for two seconds. Did the user glance down from his or her computer to a phone, or look up from a phone to a friend’s face as that ad showed? Marketers can never know for sure. Still, vCPM helps to enhance transparency for advertisers and sets higher standards for publishers, who are incentivized to try for better ad placement because they can charge higher rates. ## Frequently Asked Questions (FAQs) ### What percentage of an ad needs to be visible to count as viewable? For most digital ads on a website, in a search engine results page (SERP), on an app, or on social media, 50% of an ad has to be fully visible for it to be considered viewable, and for at least one second. ### How does vCPM compare to other bidding strategies like CPC or CPA? vCPM differs from CPC (cost per click) and CPA (cost per acquisition) by focusing on the visibility of an ad rather than clicks or conversions. vCPM only charges advertisers for impressions that are actually (or very likely) seen by a user, while CPC pays for clicks, and CPA pays for specific actions like conversions. ### How can I improve the viewability of my display ads as a publisher? To improve display ad viewability, publishers should focus on optimizing ad placement, improving page load speed, and utilizing responsive design. Strategic placement, like above the fold positioning or placement on sidebars, increases the likelihood of ads being seen. Additionally, lazy loading and ad refreshes can enhance viewability. --- ### Meta Titles: How To Write Titles That Win in Search URL: https://www.taboola.com/marketing-hub/meta-title/ Last Modified: 2026-06-22 08:49:44 Meta titles might seem like an afterthought in your overall SEO strategy, but they pack a powerful punch when it comes to search rankings and attracting your target readers. Think of meta titles as your digital storefront, the first thing visitors see when your page appears in search results. According to Semrush, Google processes over 8.5 billion searches daily. That’s why optimizing your meta titles is a must to ensure your content marketing goals are a success. ## Understanding Meta Titles (Title Tags) A meta title, also known as a title tag, is an HTML element that defines the title of a webpage. It’s one of the first on-page SEO elements you fill in that serves multiple purposes: telling search engines what your page is about, providing users with a preview of your content, and influencing click-through decisions. The meta title isn’t always the same as your page’s main headline, which is an H1 tag, as each serves a different purpose. While your H1 is designed for visitors who are already on your page, the meta title is crafted specifically to signal to search engine results pages and social media sharing. Your meta title should complement your headline but doesn’t need to match it exactly. ### Where Does the Meta Title Appear in Search Engine Results Pages (SERPs) and Browser Tabs? Meta titles appear in three key locations: the clickable headline in search engine results, browser tabs, and on social media when the page is shared. Your meta title functions as a clickable headline in SERPs when someone searches for a topic related to your content. This looks like a blue, underlined link that users click to visit your page. This placement can make or break a user’s decision to choose your result over someone else’s on SERPs. When users have your page open, the meta title also appears in the browser tab, helping visitors identify your content when they have multiple tabs open. Think of it like a mental bookmark that helps users easily navigate back to your page when they’re toggling between multiple tabs. Finally, meta titles appear when pages are shared on social media platforms, though some platforms may override them with Open Graph tags. Still, if social media is a major part of your content marketing playbook, meta titles influence how your content appears across multiple social platforms and serve as a first impression of your brand. ### What Is the Primary Purpose and Importance of a Meta Title for SEO and User Experience? Meta titles serve as a critical ranking signal for search engines, helping algorithms understand your page’s topic and relevance to specific user search queries. Content and links are among the three most important ranking signals Google uses, and meta titles play a supporting role in the content signal. For users, meta titles set an expectation for what visitors will find or read about on your page. They need to accurately represent your content while being engaging enough to win clicks. An effective meta title has to balance accuracy with appeal to stand out from generic titles. ## Key Elements of Effective Meta Titles Writing effective meta titles involves understanding both the technical constraints and consumer behaviors that motivate people to click. The most successful content marketers treat meta title optimization as part art, part science. SEO data analysis is the first step in crafting creative meta titles that inspire action. For the most part, meta titles should be precise, clear, and succinct — 40 to 60 characters max — to ensure full visibility across desktop and mobile devices. This brevity means you don’t have a lot of room for error when trying to catch users’ attention. ### Why Is It Important to Include Relevant Keywords in Your Meta Title? Having the right keywords in your meta title signals to Google what your page’s relevance is to specific search queries. A keyword-relevant meta title can significantly boost click-through rates when it matches user search intent to a T. URLs that include keywords also have a 45% higher click-through rate versus those that don’t. The same principle applies more strongly to meta titles, which take up prominent real estate on search results. Keyword placement in a meta title is important, too, because it triggers bold formatting on SERP when users search that term. This helps bump up CTRs, but avoid overdoing it; keyword stuffing in meta titles may work against you. Aim to integrate keywords as naturally as possible (10 to 15 words, max), usually at the beginning of the title. This strategy leads to 1.76 times more clicks compared to one-word keywords (per the same study linked above), indicating that longer, more specific keyword phrases are the sweet spot for winning the SERP. ### Should Your Brand Name Be Included in the Meta Title? Where? The answer here really depends on your brand recognition and the specific page type. For well-known brands, including the name can increase CTRs and build trust. However, newer brands or those with limited character space should focus on value propositions and keywords to win on search. When you do include your brand name, though, put it at the end with a separator like a pipe (|) or dash (-). For example,“What Is a Meta Title? | Taboola.” The valuable keywords appear first while giving a nod to the brand at the very end. The key takeaway here is that your meta title should accurately reflect your page’s main topic. This is crucial for both SEO and to satisfy user intent. Misleading titles might earn you some clicks initially, but they’ll result in high bounce rates and tank your search rankings over time. Google’s algorithms are increasingly sophisticated at detecting (and penalizing) this type of bait-and-switch behavior, so be careful. ## Impact on Search Engine Rankings and CTR Meta titles significantly influence both search engine rankings and click-through rates — that’s why they’re such a big deal in SEO strategy. Understanding this dual impact helps content marketers better optimize titles to please the algorithmic gods and search intent. ### How Do Meta Titles Influence Organic Search Engine Rankings? While meta titles aren’t the strongest ranking factor Google looks at, they provide important context for search engines to understand page relevance. Google uses title tags as one of several factors to determine how well a page matches a user’s search query. The relationship between meta titles and rankings is sometimes tricky, with Google’s tendency to rewrite them. Google still uses the HTML title tag (rather than the displayed version) for ranking purposes, though, so your original title tag is still important even if Google serves something different to users. ### Can a Well-Optimized Meta Title Improve Your Click-Through Rate (CTR)? Yes, it can. Meta titles are often the deciding factor in whether users click on your results versus a competitor’s. Usually, the top result in Google gets 27.6% of the clicks and the top three results garner 54.4% of all clicks, according to a Backlinko analysis of Semrush data. However, don’t be discouraged, because well-optimized titles can boost lower-ranking pages, helping them capture more traffic. Moving up even one opposition in Google can increase your absolute CTR by an average of 2.8%, per the same study. However, moving from the No. 2 to the No. 1 spot results in nearly two-thirds (74.5%) more clicks. That’s why competition for top positions is intense. ### What Are Some Common Mistakes That Can Hurt Your Rankings and CTR? Several common meta title mistakes can ding your rankings and CTR, such as keyword stuffing, or forcing your keyword into the title tag in an unnatural way. Plus, Google tends to rewrite HTML title tags that are overstuffed with keywords, which can hurt user perception of your content quality. Another big no-no is creating duplicate meta titles across multiple pages. This misstep usually happens with boilerplate text or repeating titles across pages, which creates confusion for users. It also leads to pages cannibalizing others in rankings, so you might end up with a page ranking for content inadvertently. Other mistakes to avoid in meta titles include overly long titles, missing or vague titles (as well as meta descriptions), and poorly optimized titles. ### How Can You A/B Test Different Meta Titles to Optimize Performance? A/B testing your meta titles helps you achieve meaningful performance results — with the right tools and strategy. Platforms such as VWO and Optimizely allow you to visually experiment with meta titles with variations that assess conversion rates and SERP performance. As you A/B test meta titles, focus on variables such as keyword placement, emotional triggers, and brand inclusion. Pay attention to the length of test runs to account for search engine algorithm variations — plan for four to six weeks to capture meaningful results. Aside from clicks, you’ll want to track other metrics when you A/B test meta titles, including: - Click-through rates from search results. - Average position of target keywords. - Overall organic traffic patterns. - User engagement (time on page, bounce rate, etc.). ## Meta Titles in Paid Search Ads Meta titles are instrumental in paid search ad performance, too, but there’s a bit more nuance to how title optimization translates to Google Ads. ### How Are Meta Titles Used in Search Engine Advertising? In paid search, meta titles serve a slightly different purpose compared to organic results. Google Ads headlines are the primary text element users see, but the landing page’s meta title impacts Quality Score (QS) calculations and post-click user experience. Quality Score is a diagnostic tool that compares how well your ad quality compares to other advertisers, and it’s measured on a scale of one to 10. When it comes to your ad’s landing page, Quality Score looks at page load speed, mobile-friendliness, and content relevance. These are all areas where meta titles provide important context to Google. ### How Do They Contribute to the Overall Ad Quality and Relevance? Meta titles contribute to ad quality through their impact on landing page experience, one of three main Quality Score components, in addition to click-through rate and ad relevance. To ensure your paid content is seen as relevant, make sure you include appropriate SEO tags and meta data on all landing pages. Optimizing your meta title well means it’s aligned with your ad copy and target keywords to improve the post-click experience. This helps lower bounce rates and increase users’ time on page, another element that impacts the overall QS. A relevant campaign presents a landing page solution that perfectly matches the problem a user is trying to solve. Google’s landing page quality guidelines emphasize relevance, and meta titles serve as one signal of topical alignment between your ads and destination pages. Google’s top three landing page quality factors are relevant and original content, transparency, and navigability. ### What Are the Best Practices for Crafting Effective Meta Titles for Paid Ads? Your meta title should have primary keywords from your ad groups, including exact match and phrase match keywords that drive clicks. Similar to organic titles, paid search meta titles should be 40 to 60 characters, but paid search often requires more specific, conversion-focused language that shows the user they made the right click. Include elements that reinforce the value proposition from your ads, such as pricing information, guarantees, or unique selling points. Keep in mind you can customize meta title templates for different ad groups while being consistent in messaging and brand presentation. This allows you to scale more efficiently and improve performance, rather than creating completely unique titles for every single landing page. Here are some additional best practices for crafting paid ad meta titles: - Place the primary keyword near the beginning. - Include a clear value proposition or benefit statement. - Add the brand name (if it enhances credibility). - Include call-to-action language when needed. - Use geographic modifiers for local campaigns. ## Best Practices and Optimization Meta titles are not a one-and-done task; you have to measure performance and optimize constantly across your entire content portfolio. After all, content is a long game. ### How Often Should You Review and Update Your Meta Titles? There’s no black-and-white answer here, as it really depends on content type, competition levels, and performance metrics. Definitely move pages with falling CTRs or search rankings to the top of your priority list, though. Here are some general reviewing recommendations: - Quarterly: Baseline for all pages as part of broader SEO audits. - Monthly: High-traffic pages facing increasing competition. - Immediately: Pages with declining CTR or search rankings. - Every 30-60 days: New content after enough search data accumulates. ### What Tools Can Help You Analyze and Optimize Your Meta Titles? Several tools can help you better optimize titles and track their performance over time. Here are some to consider: - Google Search Console shows queries, positions, and CTR data. - Optimizely and VWO for A/B testing. - Ahrefs, Semrush, and Moz for SEO analysis. - Browser extensions for real-time pixel length and character counts. - Built-in CMS features for title optimization with immediate length/keyword feedback. ### How to Handle Missing or Duplicate Meta Titles Across Your Website Missing or duplicate meta titles can hurt your page performance. To identify these issues, do a thorough audit using tools like Google Search Console, focusing on high-traffic pages, conversion pages, competitive keyword pages, and recently published content. Once you identify missing or duplicate meta titles on pages, work on optimizing them. Here are some approaches worth thinking about across content categories: - Product pages: Include specific model numbers, sizes, key features, and purchase information. - Blog posts: Create compelling headlines promising specific value or unique takes. - Service pages: Clearly communicate services offered and geographic availability. - E-commerce sites: Differentiate similar products with specific variations and features. - Large sites: Implement dynamic title generation and schema markup integration. ## Key Takeaways Meta titles are an important part of your overall SEO content strategy, so don’t sleep on optimization. A well-crafted meta title can give your content pages the boost they need to rank better and help your target audience find your content online more easily. Remember to keep them short (40 to 60 characters, tops), and make the copy clear, concise, and compelling. ## Frequently Asked Questions (FAQs) ### Does Google always display the meta title I provide? No. Google frequently rewrites meta titles that appear in search results, though it still uses the HTML title tag for ranking purposes (not the displayed version). Google does this when titles are too long, irrelevant to the search query, stuffed with keywords, or misrepresent the page content. ### How important is the placement of keywords in the meta title? Keyword placement impacts both search rankings and click-through rates. URLs that include keyword-related terms have a 45% higher CTR than those without relevant keywords, and the same concept applies to meta titles. That said, keyword placement shouldn’t out-prioritize readability or compelling copy. Naturally integrate your primary keyword near the beginning of the meta title while keeping it engaging and relevant. ### What are some examples of high-performing meta titles? Meta titles that have high success rates have clear value propositions, relevant keywords, appropriate length, and compelling language that drives clicks. Run meta titles through A/B testing and optimize accordingly to ensure they’re effective for your specific audience and industry. Here are some examples of high-performance meta titles by content type: - Informational content: “Complete Guide to Email Marketing: 15 Proven Strategies for 2025” - Product pages: “iPhone 15 Pro Max 256GB - Free Shipping | TechStore” - Service pages: “Denver SEO Services: Increase Traffic 300% in 90 Days | Agency Name” - Blog posts: “Why 73% of Marketers Fail at Content Strategy (And How to Succeed)” --- ### Search Engine Results Page (SERP): A Beginner's Guide URL: https://www.taboola.com/marketing-hub/search-engine-results-page/ Last Modified: 2026-06-22 08:44:59 If you know exactly where you’re headed on the internet, you can just head right to that site or even to that specific webpage. If you’ve been there before, it will likely even autopopulate for you as you begin to type in the site’s name. On the other hand, if you’re searching for something — from a great pizza restaurant in Phoenix, to the definition of mid-century modern design, or an explanation of ad fatigue — you’re not going to head to a specific website or webpage. Instead, you’ll open a search engine like Google or Bing. When you type your search terms into the browser’s bar (think “best pizza restaurant in Phoenix,” e.g.) and hit “return,” the next page you will see is the search engine results page. ## What Is SERP? A Search Engine Results Page (SERP) is the page displayed by a search engine after a user enters a query, typically featuring three main components. At a glance, you might think a SERP is pretty simple. After all, isn’t it just a list of relevant websites based on the term you used in your query? Not entirely, actually — and in fact, there’s quite a lot going on with a search engine results page, from paid placements to organic results to ads and more. Plus, these days, an AI (artificial intelligence) overview now appears at the top of many SERPs as well. ## The Evolution of the SERP Google's Search Engine Results Page (SERP) has undergone a profound transformation, evolving from a simple list of "ten blue links" to a dynamic, information-rich interface driven by significant advancements in algorithms and language models. Initially, SERPs primarily displayed organic search results determined by Google's PageRank algorithm, which prioritized links. Early updates like Panda (2011) and Penguin (2012) significantly shifted focus towards rewarding high-quality, original content and penalizing manipulative link-building tactics, respectively. This marked a move towards improving content quality and user experience. A major leap occurred with the Hummingbird update (2013), which allowed Google to understand the meaning and context of queries rather than just individual keywords. This was crucial for handling more natural, conversational language. This semantic understanding was further enhanced by RankBrain (2015), Google's first AI-powered ranking signal, which used machine learning to interpret ambiguous queries and improve results based on user interaction data. The integration of advanced language models became even more prominent with BERT (Bidirectional Encoder Representations from Transformers) in 2019. BERT enabled Google to understand the nuances, context, and intent behind complex search queries more accurately, leading to more relevant results for long-tail and conversational searches. More recently, the Multitask Unified Model (MUM) in 2021 took this even further, allowing Google to understand information across multiple languages and formats (text, images, audio, video) simultaneously, aiming to answer complex queries in a single search. These algorithmic shifts have directly influenced the proliferation of SERP features. Examples are listed below. The most recent and significant evolution (at the time of writing) is the introduction of AI Overviews (formerly SGE), which provide AI-generated summaries and answers directly at the top of the SERP, synthesizing information from multiple sources. This development aims to provide faster, more comprehensive answers, although it has raised discussions about the "zero-click" phenomenon and the continued importance of high-quality source content. In essence, Google's SERP has transformed from a list of pointers to an answer engine, driven by increasingly sophisticated AI and language models designed to understand user intent more deeply and deliver diverse, highly relevant information directly on the results page, often reducing the need for users to click through to a website. This continuous evolution necessitates that SEOs and marketers constantly adapt their strategies to maintain visibility and engage with users in this dynamic search landscape. Currently Google also offers a feature called AI Mode, which is only available in the US and in English, and is in Beta phase. Users can opt in through Google Experiment Labs. ## SERP Results (Types) The primary purpose of a SERP is to display the results of a search query on a search engine, like Google, providing users with a list of relevant websites, information snippets, and other content. The primary types of results appearing on a search engine results page are paid results (advertisements) and organic results (natural, non-paid results). In addition, modern SERPs include various SERP Features that provide rich, immediate answers and diverse content types, as explained below. ### Organic Search Results Organic search results are non-paid results, often displayed as blue links, that are determined by the search engine's algorithms. The better a website ranks in organic search results, the more people are going to find it. A high organic rank is the goal of all online advertisers and marketers. Websites earn a place in organic search results by being deemed relevant, high-quality, and authoritative by search engine algorithms. This is achieved through various factors, including content relevance, backlinks, on-page SEO, and domain age and authority. The better maintained a site is, with fresh content created by humans, not AI, and the longer a site stays around, the better its organic search placement will be. Organic results on a SERP are typically displayed below any paid advertisements or sponsored results, which occupy the first few spots. So, too, do local search results on a SERP typically appear above the standard organic search results. They often show up in a section showcasing three to five local businesses with relevant information, including addresses, phone numbers, and directions. Additionally, a map view may also appear, highlighting the location of businesses within the area. Organic search rankings are influenced by a complex web of factors, broadly categorized into on-page, off-page, and technical SEO. These factors aim to provide users with relevant, high-quality results that also provide a good user experience. It can’t be stressed enough that high-quality, informative, and engaging content is essential, with writing and media that people will genuinely want to consume. A good user experience — like fast-loading pages and easy site navigation — also helps, as does careful keyword selection. Backlinks, which are links placed on other websites pointing to yours, also do wonders for organic search rankings. ### Paid Search Results The top results you see when you run a web search may not be the best fit for your query, but there they are anyway, because someone paid for them to pop up. Paid search ads are distinguished from organic search results primarily by their placement on search engine results pages. Paid ads are typically displayed at the top or bottom of the SERP, and they are revealed by one more telltale factor: They’re labeled as “Ads” or “Sponsored,” while organic results appear naturally, based on relevance and quality. Advertisers pay for clicks on paid search ads using a pay-per-click (PPC) model. This means they are charged a fee each time a user clicks on their ad. The cost per click (CPC) is determined by an auction system where advertisers bid on keywords. Advertisers can set a budget for a given ad or ad campaign, and once the allocated money is exhausted, their ads are pulled and other ads are given priority. ### SERP Features SERP features are the various elements — including the standard blue links (or purple links, after they’ve been clicked) — that appear on a search results page. These features aim to enhance the search experience by providing additional information or functionality, often helping users find answers faster. Common examples of search engine results page features are designed to provide users with more immediate and informative results, often displayed above the organic search results. Here are some examples: - Featured Snippets: Direct answers extracted from webpages, often appearing at "Position 0." - People Also Ask (PAA): Related questions that users frequently ask. - Local Packs: Maps and business listings for local searches. - Knowledge Panels: Comprehensive information boxes for entities. - Rich Results: Enhanced organic listings with images, ratings, or other data (e.g., product snippets, recipe cards). - Image and Video Carousels: Visually oriented content directly on the SERP. - Shopping Results: Integrated product listings. - Top Stories/News: Real-time updates for trending topics. - AI Overviews:  AI-generated summary of search results. SERP features — like featured snippets, knowledge panels, and rich results — significantly impact the visibility of both organic and paid search results. They can boost visibility by highlighting content directly on the SERP, potentially capturing user attention more effectively than traditional listings. However, they can also push organic results down the page, potentially reducing their visibility. Yes, you can influence whether your content appears in SERP features. While you can't guarantee appearance (or lack thereof, though few people would hope for that), you can significantly increase your chances by understanding user intent, optimizing your content, and implementing strategies that signal to search engines that your content is valuable and relevant. To try to get yourself featured in SERP knowledge panels, focus on establishing a strong online presence, ensuring consistent information across various sources, and optimizing your content for relevant keywords. This includes creating a Google Business Profile, building a strong presence on trusted websites like Wikipedia, and using schema markup to provide structured data to search engines. To try to get tagged in a local pack, focus on optimizing your Google Business Profile (GBP), ensuring accurate and consistent NAP (Name, Address, Phone number) information, and building local backlinks. Also, encourage and respond to reviews, target local keywords, and ensure your website is user-friendly and fast. ## Analyzing and Understanding SERPs The more you know about SERPs in general, but also about how they differ across a few different search engines, the better equipped you’ll be to make search engine results pages work for your business or brand. ### Keyword Research and Competitive Analysis SERP analysis is a powerful tool for both keyword research and competitive analysis. By examining the top-ranking pages for specific keywords, you can gain valuable insights into your competition, identify ranking opportunities, and understand search intent. You can also use this keyword research process as a way to identify potential new keywords and phrases you should be using. ### Google vs. Bing Search engine results pages can differ significantly between search engines like Google and Bing due to variations in ranking factors, user interface, and the inclusion of specific features like AI integrations. While both offer organic search results, paid ads, and sometimes knowledge panels, they differ in their emphasis on certain aspects and the overall presentation of the results. Bing tends to lean more heavily into images and videos at the top of the SERPs, whereas Google uses more short snippets of written content. ### User Intent Search engine results pages can vary notably depending on the type of search you are conducting, such as informational, navigational, or transactional searches. For example: - Informational searches, like "how to bake a cake," yield results with articles, videos, and featured snippets that answer questions. - Navigational searches, such as "Reddit login," display the specific website as the top result, often with sitemaps. - Transactional searches, like "buy iPhone 16," feature product pages, shopping ads, and local listings to facilitate purchases. Understanding SERPs is crucial for digital advertisers, as they directly impact website visibility, organic traffic, and click-through rates. A higher ranking on a SERP leads to increased organic traffic, which translates to more potential leads and sales. Furthermore, understanding SERPs allows for better optimization of content, leading to more effective advertising campaigns and improved user experiences. ## Key Takeaways A SERP, or Search Engine Results Page, is the web page you see after you search for something on a search engine like Google or Bing. It displays a list of relevant websites and other search results, potentially including paid ads. A SERP can feature images, snippets of copy, maps showcasing local businesses and attractions, and more. Advertisers compete for top positioning in search engine results pages both by running paid ads and by trying to get their brands higher up in organic search results. A web page ranks well due to having good content, excellent navigability, and authority based on time and frequent updates. SERPs can also be great tools for marketers to study, as they can conduct keyword research, see what the competition is doing, and more. ## Frequently Asked Questions ### What is a featured snippet and how can I get one? A featured snippet is a special search result box that appears at the top of Google's search results, providing a quick answer to a user's query. It's a condensed version of content from a top-ranking page, and often appears in the form of a paragraph, list, or table. Getting a featured snippet can significantly boost your website's visibility and click-through rate, as it appears at "position zero" — above organic search results. You don’t pay for these snippets — you earn them by maintaining a site with great copy and content, including relevant and well-incorporated keywords. ### What is a knowledge panel? A knowledge panel is a prominent information box that appears on Google's search results pages for people, places, organizations, and so on. It provides a quick snapshot of key information about the entity, like a brief description, images, and relevant details. These panels are powered by Google's Knowledge Graph, which is a massive database of structured information. ### What Is the "People Also Ask" (PAA) Box? The "People Also Ask" (PAA) box is a Google SERP feature that displays a list of related questions to a user's initial search query. These questions are typically displayed below the main organic search results. When a user clicks on a question, the answer is revealed in a drop-down format, and a link to the source page is provided. The PAA box can also expand with new related questions as users interact with it, and it can be a powerful research tool for marketers looking to enhance the sites and pages they manage. ### How might AI impact the future of SERPs? AI is already rapidly transforming SERPs, prioritizing user experience and direct answers, while also impacting how businesses approach search engine optimization (SEO). AI-powered features like AI Overviews provide concise, synthesized answers directly within the search results, potentially reducing reliance on traditional organic clicks. This shift necessitates a focus on creating high-quality, authoritative content and optimizing for AI-generated SERP features. --- ### Lead Scoring: What It Is, Why It's Important URL: https://www.taboola.com/marketing-hub/lead-scoring/ Last Modified: 2025-07-24 10:08:16 What if there were a way for your marketing team to identify its most qualified prospects, understand their engagement journey, nurture them with tailored information, and present the sales team with the accounts most likely to convert? There is, and it’s called lead scoring. Read on to learn how to incorporate it into your marketing plan. ## What Is Lead Scoring? Lead scoring is the process of assigning numerical values (scores) to potential customers (leads). Leads earn points based on their behavior related to your brand, product, or service. The marketing and sales teams assign points based on how important each action is in terms of making a purchase, where the lead is in the sales funnel, timing of (in)activity, and so on. Leads earn — and lose — points based on their engagement with different assets and actions. Once a lead reaches a certain threshold determined acceptable by the sales and marketing teams, the prospect is turned over from marketing to the sales team as an MQL (marketing qualified lead). ## Why Is Lead Scoring Important? What Are the Benefits? Simply put, lead scoring helps demonstrate the interest and likelihood of a prospect making a purchase. This allows both the marketing and sales teams to better focus on accounts that are more likely to convert. This can mean a better return on marketing investment, shorter time to make a sale, more effort on hotter leads, less marketing and sales budget needed, and data-informed decision making. Despite helping qualify leads, only 44% of businesses were found to use lead scoring to sort these highly interested accounts. If you’re part of the majority not using lead scoring, you could be leaving money on the table, and overtaken by competitors using this tactic. After all, two out of three (68%) marketers in one survey noted that lead scoring contributes to their revenue. ### Improved Marketing ROI When leads receive a score based on engagement, they also show a trail of which actions they took to receive that score. Marketers can see what contributed to scores — i.e., what’s really moving the needle to nurture leads toward becoming a qualified sales lead. Marketers can also see where leads stalled in this process, dropped altogether, or may have backtracked, which can be parts of the scoring journey to improve underperforming assets. This all informs marketing tactics and budgets. ### Shorter Sales Cycles When the sales team is receiving only highly qualified leads, the team can focus on prospects that are more ready to buy. The benefits are twofold: - Higher likelihood of sales related to overall prospects that sales receives. - Shorter window to close a sale, since prospects have been vetted by marketing. This allows the sales team to be more efficient and effective in their jobs. ### More Alignment with Sales and Marketing Establishing a lead-scoring system agreed upon by the marketing and sales teams means that each team must buy into the process, metrics, and shared definition of success. This creates understanding and alignment into when and how a prospect is qualified. ### Data-backed Decisions Creating a strong scoring model involves looking at your funnel’s effectiveness and making decisions on what’s working, how hard it’s working, and its role in moving people toward purchases. A strong scoring model allows you to adjust, add, or modify the overall journey and experience to improve outcomes. ## Key Components of a Lead-Scoring System ### Demographics and Firmographics Information Data and facts related to what people and organizations are interacting with are known as demographic and firmographic information. This can include details such as their role in the organization, what industry they are in, the location they do business in, company size, or revenue. The closer a fit these (and similar traits) are to your ideal customer profile, the higher a chance they are to be a warm target for your services, and a better fit your customer base. ### User Behavior Traits User actions and activities are behavior traits — think of this as how people interact with your brand. This can be measured as website activity (what content they looked at and how long they spent with it), email or social media interactions (such as opening, clicking, replying, or forwarding to others), form completions, or content downloads. Metrics that involve more time commitment (over a certain amount of time on a page, filling out a form, sharing information with others, etc.) show heightened interest, and are typically weighted higher in lead-scoring models. ## Types of Lead-Scoring Models When evaluating the best lead-scoring model for your organization, consider things like what data you currently have and collect, the assets that can be weighted, what measurements matter to the marketing and sales team, the duration of a sale, and key behaviors. There are multiple ways to score leads, including creating hybrid or blended approaches. ### Rules- or Points-based Lead Scoring In the straightforward points approach, marketing and sales teams determine a numerical value for marketing efforts. Together, the group defines how much weight each interaction is worth, based on how important it is to a sale. For example, booking a product demonstration might be weighted much more than just following a brand on social media, if the team determines that action is worth more to them. Once a lead has acquired a certain number of points — as previously agreed upon by marketing and sales to signal a hot lead — the information is passed from the marketing team to the sales team. Sales can then use the information (total points earned, actions users took to receive the points, and so on) to prioritize and refine sales efforts. ### Predictive Lead Scoring Information such as past user behaviors and company or public data can be used to analyze future potential sales. Through the use of artificial intelligence (AI) and machine learning (ML), a predictive lead-scoring model identifies which accounts or users are most likely to convert. The use of this technology decreases the time marketing and sales teams need to spend in setting up a scoring model, and allows for more time to create experiences and assets, or have meaningful conversations that move a user to become a customer. ### Demographic Lead Scoring Demographic lead scoring involves collecting a person’s or organization’s traits — such as industry, job role, company budget or revenue, business size, contact information — to determine qualified leads for sales. This approach can be helpful when targeting your ideal customer profile. ### Behavioral Lead Scoring In behavioral lead scoring, data is collected and assigned a score based on a user’s actions. This data, including single actions or a collection of activities, is used to indicate where someone is in the buyer journey. Examples of what to collect for behavioral lead scoring include what users register for and when, and what they download, click on, stop at, go to next, save and share, and so on. ## How to Build a Lead-Scoring Model A lead-scoring model should accurately incorporate your business objectives and ideal customer behaviors. Together, marketing and sales should create a system that creates mutual success for both teams. ### Define Your Ideal Customer Profile and Buyer Personas Identify what a successful conversion would be, and how that person might get there. Consider where your current customers fall, and adjust expectations accordingly. Understand current and future pain points, goals, their role in the buying process, and what assets and support potential customers need to make an informed purchase, as well as any roadblocks they face. ### Study User Behaviors Look at analytics and proof points you have, and measure them against your marketing efforts. List every piece of content that a prospect could come across, and consider how important it is to ultimately making a sale. ### Assign Points Values Based on Information Track scores based on how much of a fit a person or organization is for your business goals. The closer their demographics and firmographics are to your ideal customer profile, the higher the value. Give higher points values for actions people take closer to purchase, or that are more important to your qualifying process. If a lead isn’t a fit (based on role, company information, etc.) or loses interest (doesn’t open any recent emails, stops visiting the website), subtract points from their total score. This helps keep the active prospects closer to a handoff from marketing to sales. ## Lead-Scoring Best Practices Once marketing and sales have established a model and scoring system that aligns to the ideal customer profile and buyer journey, consider additional lead-scoring best practices to enhance the process and results. ### Integrate With a MAP or CRM Once the upfront work is done, consider adding an integration, such as a marketing automation platform (MAP) or customer relationship management system (CRM). This integrated solution can score users, reduce the chance for errors, alert marketing and sales in nearly real time of a prospect’s progress, share information among programs, and manage a lot of requests at once. ### Add Automations for Active Management Layer in automation for always-on improvements and tweaks. This doesn’t mean a human touch won’t be needed, but adding automation enhances the model that sales and marketing have established. For example, filling a form could trigger an alert to the marketing team that the prospect is getting warmer. ### Define Thresholds to Transfer From Marketing to Sales Lead-scoring thresholds are the score that prospects must reach to be passed from the marketing team to sales, as agreed upon by the two groups. Setting a threshold is crucial to success, ensuring that only high-quality leads make it to sales, allowing sales to better focus efforts. It’s also vital to periodically review the accuracy and quality of leads being passed to sales. This will help ensure that scoring is not passing lower-quality leads to sales, or reveal if thresholds are set too high and keeping potential opportunities from engaging with the sales team. ### Optimize the Entire Ecosystem As previously mentioned, lead scoring is not a one-time activity. Ideal customer profiles may shift, product offerings can change, and external factors (like SEO, third-party data, or AI) can affect the overall experience and how people consume information. Be sure to audit behaviors, learnings, and even the initial scoring model. Is it working today to the best of its ability? Are there patterns that weren’t accounted for previously? During the initial six to 12 months of using your model, review it as frequently as monthly or quarterly. Moving forward, review and adjust your lead-scoring model at least once a year. ## Key Takeaways Lead scoring is a method of filtering interest in a product or service by assigning numerical values to the user based on intent, behaviors, and demographics or firmographics. Once a marketing prospect reaches a certain threshold in the model, they are passed to the sales team as a prospect interested in converting. This allows both marketing and sales to focus on users who are more interested and engaged, and nurture them towards becoming customers. An effective lead-scoring model is one that both marketing and sales come together and agree on, and aligns to the organizational ICP and buyer journey personas. ## Frequently Asked Questions (FAQs) ### How do you assign points in a lead-scoring model? The sales and marketing teams should work together to determine what level of intent a certain action or asset elicits, and create a score based on it (where a more desirable activity is worth higher points). Together, these teams will also define when a lead is handed from marketing to sales, based on points accrued. ### What data should you use for lead scoring? When implementing lead scoring, consider the following types of data: - Demographic: Information about the user, such as location, job experience details, age, gender, etc. - Firmographic: Information about the organization, such as industry, revenue, or headcount. - Behavioral: Information based on actions a user took, such as opening an email, engaging in pop-up messaging, or downloading content. ### How do you score B2B leads vs. B2C leads? While a B2B sales cycle can last well over a year and be decided by committee, a B2C cycle is often shorter, may be made by one person, and can even see a lot of emotional or impulse buys. Therefore, each scoring model needs to consider these nuances and adjust how much value to give different demographics or behavioral inputs. Make sure to align this to your ideal customer profile and buyer journey. ### What do you do when low-scoring leads convert? Sometimes low-scoring leads may convert, or high-scoring leads fail to. Use this as an opportunity to acknowledge there can be outliers, but also look to see if you should update or adjust your lead-scoring model. ### What are common mistakes to avoid in lead scoring? Lead scoring cannot be decided one time by one person. Create alignment among marketing and sales as to what constitutes a hot prospect, don’t over-rely on any one form of information without assessing the entire demographic and behavioral makeup, and remember to include lead decay. Consider a schedule to review how accurate your scoring values are, and assess if any scoring items should be added or removed from the model. --- ### Buyer Persona: Importance and Types URL: https://www.taboola.com/marketing-hub/buyer-persona/ Last Modified: 2025-06-03 12:02:59 When your marketing team is asked about who uses your products, given directives on a deliverable, or developing the next campaign, are they leaning into the end users you’re targeting, ensuring everything supports at least one type of customer? If not, it may be time to revisit, refine, or reintroduce profiles of your prospects to support your marketing and sales efforts. Representative profiles of your audiences will enable sales and marketing teams to better understand who they are creating for and talking to, equipping them to guide people successfully through the marketing funnel. This is done through creating and laddering efforts up to buyer personas — a detailed profile of your ideal customer types. Below, I’ll explain how. ## What Is a Buyer Persona? A buyer persona is a general representation of different customer types. Unlike a user persona — a profile of the person who uses the product — the buyer persona focuses on the purchase decision maker. Think of this as a biography of audience segments, compiled from actual customer data you have collected, your documented business objectives, generally available market research, and more. Buyer personas often include the following information to help marketing and sales better understand their audiences: - Demographics: Age, gender, marital status, location, income, education, job details, etc. - Psychographics: Values, interests, habits, lifestyle, attitudes, motivations, pain points, etc. - Behaviors: Purchasing history and habits, online behavior, brand interactions, media consumption, reviews and testimonials, etc. - Goals and challenges: Motivators, aspirations, obstacles, pain points, etc. Typically, marketers identify three to five customer types, and create write-ups using the information collected. These buyer personas give sales and marketing teams a composite profile of a single type of audience member. This helps them better understand how this user type relates to the full buyer journey, resulting in campaigns, plans, and deliverables that will support these buyer types. ## Why Should Advertisers Create a Buyer Persona? Buyer personas provide clarity and focus on the target audience, empowering your sales and marketing teams to better empathize and connect with your users. They also help you qualify leads that are more likely to convert, for an impactful and effective performance marketing strategy. According to the Marketing Insider Group, using buyer personas resulted in: - 56% of companies generating higher-quality leads. - 36% of companies creating shorter sales cycles. - 24% of companies developing more leads. Companies that segment their database by buyer persona exceed lead and revenue goals by 93%, Marketing Insider Group reports. Buyers also respond to a personalized purchase experience: 94% of marketers in HubSpot’s “The State of Marketing” report note that offering this customization impacts company sales. However, only two in three marketers say they have the high-quality data on their target demographic necessary to create these game-changing experiences. The key is understanding the right mix of buyer personas for your organization. ## Types and Examples of Buyer Personas Generally, aiming for three to five distinct buyer personas allows you to target a variety of audience segments. If you’re new to buyer personas, don’t have a lot of user insights, or are tight on time, even one or two personas helps ensure you’re focusing marketing and sales efforts efficiently. Consider different user mindsets when developing buyer personas. While there is a set of common user segments that sales and marketing teams rely on, it’s not a one-size-fits-all approach. Typical ways to break down buyers include competitive, spontaneous, methodical, humanistic, or negative. Here’s a breakdown of each type: ### Competitive A competitive buyer values results, data, and proof points they can quickly see. Often direct and to the point, they want the facts. They look for products and solutions they can count on, that are built to last, and give them an advantage. Consider a persona that includes quantifiable metrics and benefits, shows why your offering is superior, anticipates the competitive buyer persona’s questions, and provides product proof points. ### Spontaneous A spontaneous buyer is someone who makes a quicker buying decision, often based on how they feel in the moment or on the immediate return on investment. They look for clear and direct confirmation that this purchase is what they want now. Consider a persona that focuses on motivators, solutions, or differentiators in the market, or an immediate payback. In marketing, this persona often reacts to strong visuals and emotive information, and limited-time or time-sensitive offers. ### Methodical A methodical buyer values thorough research and understanding before committing to a decision, and that includes info about your product or service and that of your competitors. They consider and compare features and technical details, focusing on how the purchase works. Determine the most important information they seek to make an informed, confident decision when creating a methodical buyer persona. Consider case studies, fact sheets, value calculators, or other real-life proof. ### Humanistic A humanistic buyer values taking time to come to a decision, based on emotional factors such as empathy, connection, or values. They want to feel a sense of trust and authenticity with the product or company. Transparency supports their decision-making process. Understand their values, purchase history, or previous actions to determine what information to use to tell your brand story. These personas want to know about your business, and that you can back up your stories with actions. ### Negative A negative buyer persona is the makeup of the audience segment you’re not targeting. They may be very engaged, but don’t lead to a sale. These personas take a lot of company resources on the journey to purchase and ultimately don’t result in a closed sale. They distract sales and marketing teams from more likely prospects, and can be a drain on profits. Glean insights for this persona by looking at trends in information such as potential sales that were closed and lost, highly engaged digital users that don’t convert, abandoned carts, returns, complaints, and more. ## How to Create a Buyer Persona Documenting and sharing buyer personas with the sales and marketing teams will help make sure everyone is aligning work to a set of agreed-upon user types. Remember, these are general makeups of semi-fictional prospects, buyers, and customers, so there will be variances in the real world, but this is a good starting point for understanding your audiences. ### Research and Collect Information Gather information you have on hand — analytics, first-party data, feedback, etc. — as well as current and potential customer interviews, competitor analysis and audits, industry benchmarks, or social listening findings. Analyze the findings and segment information into the persona groups you create to better understand the current state of your customers and potential addressable market. ### Visualizing a Buyer Persona Formalized buyer personas are commonly created as documents that look similar to a resume, often including a name, an image, and a backstory. Key information and topics about the buyer and their buying habits or motives are listed in the persona documentation. It’s also a good idea to add an image to help visualize the buyer persona, along with a background blurb. ### Information in a Buyer Persona Persona names often align or allude to the persona type they represent. Through the information you have collected to create your buyer personas, create a composite of that user segment, which often includes demographics, psychographics, behaviors, and goals, as explained above. ### Validating a Buyer Persona Once you have finalized your buyer personas, share these with the teams who know your user segments the best and will be using these write-ups going forward. Ask for feedback: What parts of the personas do they challenge, or is anything missing that would help them do their job better? Take the feedback and update the personas as needed. Revisit the profiles over time: Is there new or different information that should be incorporated into the personas? Are there outdated examples, or different users you should consider as your organization or product line changes? ## Key Takeaways Sales and marketing teams can better understand the people that are purchasing their products or services through buyer personas — semi-fictional biographies of customers and prospects based on a user segment. These profiles are created with quantitative and qualitative information based on current customers and prospects, market data, and more, and provide demographic, psychographic, and behavioral information about each audience segment. By using personas, sales and marketing teams can better understand the people they are connecting with, which can lead to a shorter sales cycle and increased revenue. ## Frequently Asked Questions (FAQs) ### What is a negative buyer persona? A negative buyer persona is a representation of the type of buyer you aren’t going after. It's a fictional biography backed up by actual customer and market data of a buyer that isn’t the right fit for your audience, product, or service, so the sales and marketing team understands who they’re not talking to. ### What goes into persona development? Personas are created using a variety of data and quantitative and qualitative inputs, including demographics, psychographics, behaviors, and goals and challenges. ### Buyer persona vs. user persona: What’s the difference? A buyer persona is a semi-fictional biography of the people who purchase a product or service, and is used by sales and marketing teams to influence efforts during the buyer journey. A user persona is a representation of the type of people who use the product or service you provide, which helps to inform product and design teams. While that could be the same person, it likely could be different people, and influence different roles and tasks within marketing and sales functions. --- ### Omnichannel Marketing: A Beginner's Guide for Performance Advertisers URL: https://www.taboola.com/marketing-hub/omnichannel-marketing/ Last Modified: 2025-06-05 06:48:15 As the online advertising landscape becomes increasingly fragmented, omnichannel marketing is more important than ever. When implemented strategically, it can help you connect with customers in the ways they want to be reached. It’s important to know what’s involved with omnichannel marketing, what its core principles are, and how you can benefit from an omnichannel marketing campaign. ## Understanding Omnichannel Marketing ### What is Omnichannel Marketing and What Are Its Key Characteristics? Omnichannel marketing is defined by marketing that cohesively exists across all possible channels and meets the customer at every stage of the funnel. The core principle and goal of an omnichannel approach is to create a seamless customer experience across every touchpoint and channel, including websites, apps, physical stores, and every other place a customer might see or hear an ad. Key elements include consistency, convenience (offering the customer a chance to engage, no matter where they are), personalization (done through gathering data across the customer journey), and a sense of where the customer is now in their buying journey. This is why you might put something in an online cart but not buy it, and then see an ad for that same product the next day. Omnichannel marketing approaches the customer from all angles and tries to encourage them to move down the marketing funnel. It’s becoming increasingly more important for businesses as customers engage with media in so many forms and on so many platforms. There’s also the fact that if a brand’s competitors are on TikTok, Instagram, the digital kiosk at the market, the print newspaper, and everything in between, that brand better be there as well. ### How Does Omnichannel Marketing Differ From Multichannel Marketing? While, theoretically, it could be as simple as the difference between omni (all) and multi (many), the meaning goes deeper than that. Omnichannel is about bringing the customer a unified experience across all touchpoints, whereas multichannel aims to promote the products across a variety of channels without a personalized integration. With a multichannel approach, a product might have a different voice or style for its website than, say, for a magazine ad. With omnichannel, the voice is unified and is designed to work with the brand’s messaging on other channels to keep customers engaged in the product’s world. ## Benefits of Omnichannel Marketing ### How Does Omnichannel Marketing Enhance the Customer Experience? Omnichannel marketing enhances the customer experience by delivering consistent messaging across all channels, so the customer feels like they’re dealing with an actual entity, and not a random series of brands being thrown at them. ### Can Omnichannel Marketing Drive Higher Conversion Rates and Sales? Omnichannel marketing can drive higher conversion rates and sales by tracking consumer data across the customer journey. The strategy is largely informed by data, which can tell you how and when your consumers engage with brands. As the approach maintains consistent messaging and reaches the customer at multiple touchpoints, it promotes brand recognition, customer engagement, and loyalty. The strategy gives marketers a more holistic view of the customer journey via mapping it along all touchpoints. It identifies where customers interact with a brand, analyzes the data to see where friction occurs, and then develops plans to work through the challenges, and land a conversion. ## Key Components of an Omnichannel Strategy ### What Are the Different Channels That Can Be Integrated in an Omnichannel Approach? The different channels that can be integrated in an omnichannel approach include SMS messages, radio and podcast ads, video ads, display ads, in-store displays, in-store and outdoor interactive kiosks, social media ads, and email marketing newsletters. ### How Do You Ensure Seamless Transitions Between Different Touchpoints? You can ensure seamless transitions between different touchpoints by employing consistent messaging that makes it feel like the brand is a dependable source, no matter how or where the consumer encounters it. One way to maintain consistent branding and messaging across all channels is to create a style guide that includes certain words and phrases, as well as graphics, fonts, colors, and other aspects of visual presentation. Maintain clear and firm rules about how you speak to your target audience, whether as a trusted friend or something more formal. Decide whether your voice is humorous, serious, or a combination of both. As you work out the nuances and create a clear voice, it will become easier to maintain a consistent tone when launching your omnichannel campaigns. Make sure to save and analyze consumer interactions along the way, so that you have a sense of their behavioral history and you can see where they might have lost or gained interest. You can also set up your campaign touchpoints so that the consumer can engage with a brand on one platform and complete the purchase on another if necessary. This way, they can buy whenever the mood hits, and not just when they happen to be on one particular channel or device. ### How Important Is Personalization in an Omnichannel Strategy? Personalization is very important in an omnichannel strategy (and, indeed, in any marketing strategy). It’s also easier than ever before, given the amount of data marketers can collect using the various technologies available to not only gather data, but organize it, and in some cases suggest a next move. Nearly 75% of consumers say they are more likely to buy from brands that deliver personalized experiences, and they spend 37% more with those brands, according to a recent study from Deloitte. Since an omnichannel strategy is based on reaching the consumer through as many avenues as possible and basing interactions on previous behavior, it offers a particularly good opportunity to achieve a high level of personalization. ## Implementing an Omnichannel Approach ### What Are the Initial Steps Involved in Developing an Omnichannel Strategy? To get started with developing an omnichannel strategy, you’ll want to map the customer journey, define how you will measure success (ROI, certain KPIs, etc.), create a unified customer data strategy with easily accessible data, arrange a flexible tech stack to support your personalization strategy, and be prepared to keep testing and iterating. ### How Do You Map the Customer Journey and Identify Key Touchpoints? In order to map the customer journey, find out what channels your customers use, where they are likely to find your brand, where they either stop without making a purchase, or where they decide to make a purchase. You can experiment by offering personalized discounts for people who abandon their carts or are returning after an absence of shopping. You want to get clear about your business goals and which metrics are most important to you in determining whether your campaign is a success. Create a data strategy that will give you a full view of the customer, so you can track their interactions across various touchpoints with your brand’s messaging across the channels. As you get to know your customer, you might prioritize different channels and types of messaging over time. ### How to Measure the Success and ROI of Your Omnichannel Efforts? You’ll want to take a few key things into account when measuring the success and ROI of your omnichannel efforts. For example, you can track key performance indicators like conversion rates, cross-channel engagement, customer lifetime value, customer retention, customer acquisition costs, and return on ad spend. Pay attention to which areas of your campaign drive conversions and engagement (revenue attribution), and how much your campaign has earned versus how much you’re spending. You can look at the campaign as a holistic experience and then break down where you’re seeing positive movement. In order to do this, make sure your tech stack is giving you what you’ll need. A good mix includes: cross-channel analytics to track customer behavior across the platforms; a customer relationship system to provide a holistic view of a customer’s history and enable personalization; customer data platforms to unify data from all touchpoints; and API integration for real-time data sharing across the platforms. The silos that generally exist between sales and marketing can hamper productivity and campaign success, so it’s wise to break them down as much as possible. By fostering a collaborative culture where both sides share the same goal, utilize one unified source of data, and engage in open and easy communication, you’re more likely to get the teams to behave like one solid force, rather than separate entities. ## Challenges and Best Practices ### What Are Some Common Challenges in Implementing Omnichannel Marketing? Some common challenges in implementing omnichannel marketing include maintaining brand voice and messaging throughout all channels, integrating data across all channels, dealing with outdated technologies that won’t support omnichannel marketing, handling attribution measurement issues, and ensuring you’re getting quality data that can help you make the right decisions. ### What Are the Best Practices for Creating a Cohesive and Seamless Customer Experience? In order to create a cohesive and seamless customer experience, you’ll need to make sure you can overcome data integration and channel management issues. You can do this by making sure your data isn’t siloed and is instead easily accessible to all departments. You can also connect your sales, marketing, and customer service teams so everyone has access to the same information when viewing customer behavior on your channels. You’ll also want to agree on a specific brand message so your customer has a consistent experience no matter where they encounter your brand. One way to do that is to leverage customer data effectively, by paying attention to what resonates with your customer and what influences their engagement with your brand. From there, you can personalize your approach for specific situations. And, of course, you want to stay on top of emerging technologies (like AI) that can help you better track and plan for evolving customer behavior. As you get to know your customers, and as customers overall change their expectations and needs, you want to stay agile and open to changing your methodologies. ## Key Takeaways Omnichannel marketing is a necessary part of doing business in today’s marketing landscape, where consumers are encountering your brand on multiple different channels. By integrating various technologies and creating a consistent messaging plan across every touchpoint, marketers can reach consumers in highly effective ways that can encourage greater campaign success and ROI. ## Frequently Asked Questions (FAQs) ### What are some examples of successful omnichannel marketing campaigns? Starbucks has seen success with its loyalty rewards app as part of an omnichannel marketing approach. The app lets customers check their balances and reload their cards whether in-store, on a website, or on the app, with real-time updates across every platform. Walgreens has also benefited from an omnichannel approach by allowing users to optimize their in-store experience via online choices. Customers can quickly fill out refill requests and place orders on their mobile device, and then pick up the item in a local store. While this might seem like a “digital first” experience, it ties in directly with making their in-store experience easier and more streamlined. ### Is omnichannel marketing only for large businesses? Omnichannel marketing is for businesses of all sizes. While larger companies might have the means to make implementation easier, smaller businesses have everything to gain from connecting with the consumer across every possible channel. ### How much does it cost to implement an omnichannel strategy? The cost of implementing an omnichannel strategy will vary from brand to brand, as it depends on how many channels and platforms a brand is seeking to integrate into its strategy, and how much data it is looking to collect and analyze. The more data and the more touchpoints, the higher the costs might be. ### What is the future of omnichannel marketing? The future of omnichannel marketing involves more personalized experiences and increased use of AI to analyze data and aid in marketers’ decision-making. Immersive experiences, such as augmented reality and virtual reality, will likely also increase as technologies develop and consumers crave new and innovative forms of engagement. ### How do you train your team for an omnichannel approach? In order to train a team when implementing an omnichannel approach, it’s important to develop clear and cohesive brand messaging and commit to deliver that across all platforms. You’ll also need to train the team to understand the different customer relationship management systems and various kinds of data they will encounter throughout the omnichannel campaign. The team will need to learn how to align with other departments on shared goals, and how to effectively understand consumer interaction across the various touchpoints on the way to conversion. --- ### Search Engine Optimization: The Importance of SEO for Your Business URL: https://www.taboola.com/marketing-hub/search-engine-optimization/ Last Modified: 2026-06-22 08:50:15 One of the most important advertising spaces in digital marketing is the Google search results page, and search engine optimization (SEO) is an affordable and effective way to get there. A successful SEO strategy involves creating unique content using relevant keywords that search engines recognize, but only to an extent that does not compromise the quality of brand messaging or user experience. So, it’s a delicate art, but an important one to master to gain new customers. ## What Is SEO? SEO is the pairing of technical knowledge of how search engines work with quality content that consumers need. When websites answer questions with relevant information and optimize this content with common keywords people use, it helps content appear more often, and higher up, on search engine results pages. That is why search engine optimization is vital for directing people to your website, and by extension, helps promote your products and services. ## Importance of SEO for businesses SEO is vital for business, because the ongoing optimization allows companies to gain visibility while generating traffic, leads, and conversions without spending much money (a few subscription fees for search engine optimization tools, and whatever your content budget is, e.g.) “SEO is how people find your business online if they don’t already know your brand name and you’re not paying for Google or social media ads,” explains Michelle Symonds, an SEO specialist at Ditto Digital. By ensuring that your website is displayed when someone does an online search relevant to your business, companies find new customers from the right locations, she adds: “That is why it’s important for any business.” ## Crawling, Indexing, Ranking: How Do Search Engines Work? Search engines work like a digital librarian, explains Symonds. First, bots scour billions of pages to identify new content and changes to existing content, which in the SEO world is referred to as crawling. Once a business’ page is crawled, the search engine uses keywords and topics to organize content in an easily retrievable database of all the information on the internet, or an index. Finally, when someone goes to the digital librarian or search engine with a question, they can find the most relevant content from this index. This determines a website's rank in search engine results. “Ranking uses complex computer algorithms, comprised of hundreds of different factors, to decide the order of the web pages retrieved for any particular search query,” Symonds says. A number of Google Core updates have informed these algorithms in recent years. ## What Are the Most Recent Google Core Updates and How Do They Affect Search Results? ### Helpful Content Update Introduced in 2022 to reduce spam and irrelevant backlinks, the Helpful Content Update was refined in 2024 to prioritize quality content and user experience over low-quality, keyword-padded pages. These updates “rewarded genuinely helpful content and cracked down on unnatural links,” says Paul Jozsef, founder of the marketing agency Digital Practice. Likewise, they built on Google’s ranking system, which emphasized expertise, authoritativeness, and trustworthiness (E-A-T) by adding another E to the equation: experience, or E-E-A-T. ### AI-generated Spam Update In March 2024, Google updated its algorithm to crack down on websites scaling content with AI and AI-generated spam. This update also targeted sites using expired and irrelevant domain names to manipulate the algorithm. These major changes also reduced the search relevance of “low-quality, SEO-first pages,” Jozef says. ### Product Review Update The product review update in February 2023 rewarded detailed analytical product reviews that, again, reflected experience, expertise, authoritativeness, and trustworthiness. This update also penalized vague reviews and forced marketing professionals to update content with more details. ## How Is AI Affecting Search Results? ### SERP Snippets AI-generated SERP (search engine results pages) snippets — i.e., the quick summaries that appear at the top of these pages — have reduced organic traffic among users looking for quick, surface-level answers because they can get that without leaving the search engine page. Still, customers looking for more detailed information need to see your SERP snippets at the top of search results to click on your website. So, creating content that emphasizes experience, expertise, authoritativeness, and trust is still essential for SEO. ### AI-powered Search Engines AI-powered search engines like ChatGPT and Perplexity have changed how people look for information. Although Google’s indexing capabilities make it easier to optimize for keywords, ChatGPT doesn’t crawl or organize information the same way. Instead, these search engines rely on large language models to answer questions. As a result, companies should create conversational, personalized content geared towards intent-based searches when optimizing for AI-powered search engines. ### Optimization Efficiency AI-powered tools like Surfer SEO can provide fast, detailed analysis of keywords, make content recommendations, and analyze user data for better audience targeting. These tools can also quickly generate content. However, because the Google algorithm favors originality, AI-generated web pages produced without human editing and oversight are more likely to harm a company’s SEO strategy than help it. ## 4 Types of SEO Techniques ### On-Page On-page SEO is a type of search engine optimization strategy that refers to everything a company can control on its website. This includes content quality in general, but also keyword usage, headings, and other meta tags, internal linking, alternative text that describes images, and URL structure. Correctly optimizing these elements can all help improve rankings. ### Off-Page / Link Building Off-page SEO is about getting other credible websites to link to your site in their relevant content, also known as backlinking. Collaborating with other businesses and content creators by using their links or providing opportunities for guest blog posts can help boost backlinks. ### Local Local SEO is essential for local, physical businesses. It ensures that when people search for a product or service nearby, they know where to go. The best way to optimize local SEO is to create a Google Business profile to gain online reviews and research local keywords to incorporate into relevant content. ### Technical Technical search engine optimization involves anything related to how your website functions. To improve technical SEO, companies need to make sure their websites are efficient and easy to use. This also includes optimizing code to provide search engines with additional information and decreasing page load time, which is one technical aspect of Google Core Web Vitals signals. ## SEO Best Practices By implementing key SEO best practices, you can enhance your website's visibility, attract organic traffic, and ultimately drive business growth. From strategic keyword research to technical site improvements, a well-executed SEO strategy is crucial for cutting through the noise and connecting with your target audience. Here are some recommended practices. ### Crafting High Quality Content Since the Google algorithm de-prioritizes vague pages that overemphasize keywords, creating high-quality content is the best way to optimize for SEO. That means content should address your ideal customers' pain points and educate them on how your products and services can solve their problems. Quality content should also be driven by E-E-A-T principles. Incorporating customer reviews, testimonials, case studies, backlinks, and expert insight can highlight experience, expertise, authoritativeness, and trustworthiness and boost your website's ranking for relevant searches. ### Utilize Long-Tail Keyword Searches Long-tail keywords are phrases with three words or more that are more descriptive and indicate that the individual searching is further along in the marketing funnel. Longer-tail keywords tend to have lower competition as well, so they can lead to higher rankings and conversion rates. For example, a parenting brand may use long-tail keyword searches like “best strollers for babies” or “how get a newborn to sleep” which can tell business more about user intent. Tools like Semrush can help businesses research long-tail keywords with higher search, and lower competition and keyword difficulty. ### Embrace AI, But Don’t Depend On It With more consumers going to AI-powered search engines for answers instead of Google, adapting to emerging technologies like artificial intelligence has become an inescapable aspect of SEO. To optimize for AI search, focus on long-tail keywords that emphasize user intent. Creating informative content that clearly answers queries, incorporating multimedia, and adding code to your website’s HTML that describes content, can also improve AI search rankings. Although AI can generate content, both Google and AI search engines deprioritize general content that lacks expertise and human oversight. At this stage of innovation, it's best to utilize AI to expedite content research and outlines. ### Think Locally It is becoming more competitive to rank for national or international search terms, but local, personalized search terms can improve SEO, especially for small to medium businesses (SMBs). By creating a Google business profile and filling out each section with detailed, accurate information, adding images, and engaging with customers through reviews can help companies optimize for local SEO. Tools like Google Keyword planner can give brands insight into what keywords and phrases the local audience is using in their searches. ### Maintain Authority Once your website is ranking well, it’s important to keep that positioning by maintaining authority. Investing in PR outreach can help brands network with other reputable sites, who can link to your content. Boosting your backlink profile communicates to Google that your company is an authority on certain topics, and should be ranked higher than sites with fewer backlinks. With more overhead, businesses can put a portion of their ad budgets towards original research, case studies, and expert testimonials. Backlinks combined with a strong E-E-A-T content strategy is an effective way to stay on top through SEO. ## SEO Tools ### Google Analytics Google Analytics (GA4) is a free tool that helps brands see where their website traffic is coming from, such as organic search, social media, paid ads, or other referrals. This helps with identifying which channels to focus on optimizing. Other tools like Google Search Console can be integrated with GA4. This shows what queries and keywords consumers are using to find your site. Businesses can then create content that speaks to these searches, which can boost rankings and SEO. ### Semrush Semrush has a database of over 25 billion keywords, providing valuable insight into what people are searching for related to a certain subject area. This tool also helps brands identify trending topics and which high-volume keywords have lower keyword difficulty, to improve content strategies and overall SEO. ### Ahrefs Like Semrush, Ahrefs helps businesses identify important keywords. This comprehensive platform also provides insight into what keywords competitors are ranking for, showing opportunities for optimization. Ahrefs also helps with critical backlink analysis by tracking the number and quality of backlinks on your website, flagging broken backlinks, and identifying areas where anchor text could be improved for SEO. ### Surfer SEO Surfer SEO analyzes top-ranking pages and makes recommendations based on the keywords and structure of top performers. The AI-powered resource also identifies keyword-rated words to enhance search further and gives content a score to gauge relevance and highlight areas for improvement. ## How to Measure SEO Success ### Ranking Appearing higher up in search results is the goal of search engine optimization, so where your content ranks is a clear indicator of SEO success. The higher your ranking, the better your content's chances of getting eyes on it, which tends to increase organic traffic. ### Impressions Impressions, or the number of times your website appears in search results, is another valuable metric for measuring SEO success. Like ranking, impressions directly impact brand visibility and overall reach. Experts agree that anywhere from 500 to 1,000 impressions a month indicates your SEO strategy is working. ### Click-through Rate While ranking and impressions indicate whether your website is showing up in search results, how high up it is, and how often, click-through rate (CTR) shows how many people are clicking on it. A high CTR is a sign that your content is relevant and engaging, whereas a low CTR with good ranking and impressions can be an opportunity to re-strategize about the quality of your messaging. ### Organic Traffic Organic traffic, or the number of people visiting your site without targeted advertising, is a good signal that your SEO strategy works. If your website is ranking well, receiving a sizeable number of impressions, and people are clicking on content, this should also be reflected in organic traffic. ## Key Takeaways SEO is a marketing technique that combines knowledge of what search engines prioritize with unique, informative content, so it ranks higher on results pages: It’s vital for educating new customers about your business when searching for solutions online. Search engines like Google rank content that emphasizes experience, expertise, authoritativeness, and trustworthiness higher than vague websites that rely on keywords only. Advances in AI are making some aspects of optimization easier, such as keyword research, but innovations like AI summaries of SERP snippets can reduce CTR. Certain metrics, such as ranking and organic traffic, inform companies whether their SEO strategy is working or should be reassessed. ## Frequently Asked Questions (FAQs) ### SEO vs. PPC: What’s the difference? SEO refers to the organic process of getting your website to rank higher for searches related to your content, products, and services. This is achieved through creating high-quality content that naturally incorporates relevant keywords frequently used in Google searches. Pay-per-click, or PPC, involves paying for ads or sponsored content to appear at the top of search results. Both marketing strategies have similar goals — mainly to appear on search engine results — but SEO does not require much overhead to get started. ### What role does content marketing play in SEO? Content marketing involves strategically creating and sharing information that helps people, but does not promote a brand directly. The goal is to educate consumers, and when done correctly, content marketing gives businesses more authority on particular topics, which boosts visibility in search results. Thus, effective content marketing is essential for SEO, and the strategies work best together. ### How do I conduct a link audit on my website? Conducting a link audit is easy with tools like Google Search Console or Ahrefs. Compile your link data before examining each link individually and identifying any broken links or areas for improvement. Finally, create an action plan to fix issues, such as updating and increasing backlinks and internal links. ### How does SEO affect or contribute to performance marketing? Although performance marketing refers to paid campaigns, effective SEO strategies can enhance these efforts by increasing the number of people who enter your marketing funnel. ### How frequently should I update my content for search optimization? Since relevant content is a vital part of search engine optimization, experts recommend conducting updates every few months at a minimum. However, updating content as often as possible, even weekly, is optimal. Content should also be promptly updated for SEO if there is a sudden decline in traffic or rankings. --- ### Vertical Banner Ads: What Are They? How Are They Used? URL: https://www.taboola.com/marketing-hub/vertical-banner/ Last Modified: 2026-03-23 11:10:11 Whether you’re an advertiser or just a regular internet user, one thing we can all agree on is that the digital landscape right now is enormous. With so many apps, ads, and services competing for clicks, it’s an assault on users’ senses, and difficult for advertisers to keep up with the latest flashy ad mediums while finding a way to stand out. Throughout all the internet evolution of the last few decades, a constant companion has been the banner ad, specifically vertical banner ads. They may look similar to the early days of the internet, with the same shape and basic premise, but don’t write them off just yet: There’s a lot this ad format can deliver, and a lot you’ll be able to do with it. As an advertising copywriter, I see the value in these. They’re not as invasive as a video ad, but offer more than a static print ad, and still give a fair shot to get your message and product across in enough time to entice a user to click. Plus, they can get really creative, too, so make sure you have fun with it. First, though, let’s learn a little more about their structure and how they work. ## What Is a Vertical Banner (Also Called a Skyscraper Ad)? Vertical banners, or skyscraper ads, are tall, narrow digital advertisements on the lateral margins of web pages. You’ve definitely seen them before, sitting on the sidelines, cycling through animated slides, and ending on a call to action (CTA) asking for your click. They’re one of the oldest forms of online advertising, but they’re still around because they bring a lot to the table. They’re persistent without being overly obnoxious, easily visible during scrolling, and offer sustained brand exposure, making them a reliable method for capturing user attention and getting your targeted messaging through. ## Standard Sizes for Vertical Banners These are the most common vertical banner sizes you’ll find and have access to creating: ### 120x600: The Standard Skyscraper The most widely adopted vertical banner format by far, it fits into sidebars well and gives enough vertical space for content and broad reach. ### 160x600: The Wide Skyscraper If you need a slightly wider option, this may be for you. The wide skyscraper provides greater creative flexibility without significant intrusion on the user experience. ### 300x600: The Half Page Ad (HPA) Significantly bigger and taking up way more space than the other skyscrapers, the HPA is a larger, attention-grabbing (and really hard to ignore) format. It’s most used for ads that are working with detailed visuals and information, requiring careful placement consideration that wouldn’t otherwise work in a slimmer ad format. ### Other Less Common Sizes There’s definitely other variations of sizes out there, and I’ll get into a few more later, but focusing on the standard, wide, and half-page sizes are what I’d advise when getting started. ## Benefits of Vertical Banners ### Persistent Visibility and Sustained Engagement When your ad remains visible during scrolling, it significantly enhances brand exposure and the chances that users will remember your message. ### Minimally Intrusive Placement Strategy With its lateral positioning in the sidebars, vertical banners bring a less disruptive advertising experience to the user, compared with alternate methods like pop-ups. ### Creative Expression The vertical format allows for comprehensive and engaging creative executions. You can get clever and fun with these, utilizing the format and even its limitations to create an ad that users will find funny, meaningful, and, most of all, memorable. ### Brand Reinforcement and Recall When an ad is memorable, consistent, and non-disruptive, it’s more likely to get positive brand recognition from users over time. ### Utilizing a Website’s Content Think about using the content or subject matter of the web page as part of your advertising strategy with a vertical ad. Thoughtful placement and integration can boost an ad’s recall and recognition. ## Considerations to Take When Advertising With Vertical Banners ### Maintain a Balance Between Visibility and User Experience Prioritize a positive user experience alongside advertising exposure. This means keeping eyesores like seizure-inducing flashing lights or overly repetitive CTAs to a minimum (or better yet, not at all). ### High-Quality Creative Design There may not be a ton of space to work with here, but an impactful design that’s easy to read and understand is essential to getting your users to take action. ### Mobile and Other Devices People aren’t just on desktops anymore, and you’ll need to assume that a good percentage will be viewing your ad on phones, tablets, or some other connected device. That’s why easy adaptation across various screen sizes is crucial, so it’s not a scrambled mess of broken images when it loads. ### Ad Blocking Software Ad blocking software is everywhere now, even built into some browsers. It’s a tough one to get around, if at all, so it’s best to keep in mind the potential impact of ad-blockers as you craft your campaign’s designated reach. ### Monitoring Metrics After all the above is checked off, maybe the most important thing is to make sure that your banners are actually being seen by the target audience. If not, it’s time to take a step back and see where the blockage is. Tracking core metrics is crucial for understanding the effectiveness of your vertical banners, but to truly grasp the impact they’re having, it's best to go beyond the basics. Choose an ad platform that offers in-depth features like comprehensive analytics dashboards, which can provide you with deeper insights into metrics like user engagement and conversion paths that were initiated by your banners. ## Vertical Banners vs. Other Banner Ad Types Before getting started on verticals, take a minute to get to know these other various banner sizes and shapes: ### Leaderboard (728x90) Horizontal, typically top-of-page, with different visibility characteristics. ### Medium Rectangle (300x250) Versatile, near-square format, often within content, with less sustained visibility. ### Wide Skyscraper (160x600) vs. Standard Skyscraper (120x600) The difference here is primarily the width and visual prominence on the page. ### Mobile Banners (Various Smaller Sizes) Specifically designed for smaller mobile screens. ## Vertical Banner Ad Placements Size and creative content are just two major pieces of the puzzle when it comes to creating a successful vertical ad. Another crucial area is placement — here are a few examples of where vertical ads can go to be most effective: ### Website Sidebars Good ol’ tried and true, this is the conventional and most frequent placement for sustained visibility. It’s a snug spot that’s noticeable, fitting for the banner’s shape, and visible without being intrusive. ### Adjacent to Long-Form Content Similar to the above, you can also make your ad more effective when thematically relevant to the surrounding text. ### Within Website Navigation These ones are a little unconventional, as they target users while they’re actively scrolling through the site. They capture attention, but can be seen as annoying, too. ### Below the Fold Much like in a newspaper, a below-the-fold ad might be bigger, more effective, and engaging for those who see it, but not everyone is going to, and that’s part of the risk. ## Best Design Practices for Vertical Banners ### Clear and Concise Messaging Prioritize a singular, focused key message for clarity and fast delivery. You’ve got about two seconds before you lose a viewer’s attention. ### Utilization of High-Quality Visuals Even if you have a killer tagline and copywriting, don’t skimp on relevant, high-resolution graphics. These significantly enhance engagement and make things look more professional and respectable. ### Prominent and Action-Oriented CTA Concise, directive language ensures that the user will know where to go and what to do if they want to take action. If it’s not clear, they may give up and move on. ### Consistent Brand Identity Integration Reinforce brand recognition through consistent use of visual elements, so viewers can learn to easily recognize your style at a glance. ### Optimization of File Size for Rapid Loading You can have the best vertical ad ever, but if it doesn’t load properly, it’s pretty useless. Make sure to minimize file size in order to enhance user experience and boost your ad’s performance. ### Adaptability Across Devices I mentioned this above as well, but don’t forget about scalability (or develop device-specific creative variations). Bottom line, viewers should be able to easily see it no matter the device they’re on. ## Vertical Banner Ads Measurement (Metrics) Once your ad is up and running, how do you know if it’s successful? Here are some key performance indicators (KPIs) to keep track of: ### Impressions These are the total number of times the banner is displayed, letting you know the reach you’re getting from it. ### Click-Through Rate (CTR) CTR is the percentage of impressions resulting in a user taking action and clicking. ### Conversion Rate Similar to CTR, but a conversion rate tells you the percentage of clicks that led to a desired ultimate goal, like a purchase. ### Viewability This is a key one to measure how well your placement is working. Viewability shows you the percentage of the ad that’s visible to users, clueing you in to its efficiency. ### Cost Per Click (CPC) Vertical banners can bring in sales and revenue when successful, but how are you supposed to know how much it’s going to cost you? The average cost per user click is crucial info for budget management. ### Return on Ad Spend (ROAS) This shows you how profitable your ad is compared with what you’re spending. ROAS measures the revenue generated for every dollar spent, specifically on your vertical banner advertisements. It's calculated by dividing the total revenue attributed to vertical banner ads by the total cost of running those ads. A higher ROAS tells you that the vertical banner ad campaign is doing its job in generating revenue. ## Optimization for Vertical Banner Performance ### Desktop Optimization Always be A/B testing creative to see which versions resonate with users more. Along with that, stay on top of refining targeting, monitoring placement efficacy, and enhancing landing page relevance. ### Mobile Optimization Strategies Consider alternative formats for users on mobile devices, and optimize for touch interaction since that’s becoming more and more commonplace. Be sure to prioritize concise messaging, and test mobile-specific landing pages too. ## How Do Vertical Banners Perform in Programmatic or Real-Time Bidding? Vertical banners are widely available in programmatic and RTB environments, often with competitive bidding. These programmatic platforms can help with offering precise targeting, real-time optimization, and detailed performance analytics. ## What Are Common Mistakes to Avoid With Vertical Banners? If there’s one overarching piece of advice you’ll hear me say over and over, it’s to keep things straightforward and concise. You’ve got a short amount of time to get your message across, so stay away from things like: - Overly complex and cluttered design. - Generic and uncompelling messaging. - Low-quality or irrelevant visuals. - Neglecting mobile responsiveness and display optimization. - Lack of a clear and prominent call to action. - Not implementing comprehensive performance tracking. - Deployment in irrelevant placement contexts. ## Key Takeaways After decades, vertical banners still remain a simple but strategically valuable tool in digital advertising, offering wide visibility to users across devices, apps, and interests. Understanding things like standard specifications, effective design principles, and key performance indicators is essential for creating successful campaigns. But, make sure to prioritize clear messaging, high-quality visuals, and relevant placements, with continuous testing and optimization. ## Frequently Asked Questions (FAQs) ### Vertical banners vs. web page sidebars: What’s the difference? They’re related, but different. A web page sidebar is a structural component of a website's layout. It’s a dedicated vertical section, typically on the left or right, designed to hold various elements. Think of it as a container or a column within the page's overall design, one that can house a wide range of content, from navigation menus and social media feeds to calls to action and, of course, advertisements. A vertical banner, on the other hand, is a specific type of advertising content that’s within that web page sidebar (or sometimes another vertical area on a web page). It's a visual advertisement specifically designed to attract attention and convey a marketing message. While a vertical banner ad often resides within a sidebar, the sidebar itself is a more structural element that can contain various non-advertising content as well. The key distinction between the two is that a vertical banner is specifically for advertising, while a sidebar is more for general content and navigation. ### What are the latest trends in vertical banner advertising in 2025? Digital advertising is constantly evolving, and vertical banners are no exception. On the surface they may look the same as past years, but it seems like several key trends are popping up: - Enhanced interactivity: Static banners are becoming less engaging and inspiring, and easier to ignore. Expect to see more vertical banners incorporating interactive elements like quizzes, polls, mini-games, and dynamic content feeds to capture user attention and encourage interaction directly within the ad unit. - Advanced personalization: Leveraging user data for more sophisticated personalization is something that’s increasing among advertisers across the board. Vertical banners will likely adapt their messaging and visuals based on individual user behavior, browsing history, and real-time context, aiming for greater relevance and effectiveness. - Augmented reality (AR) explorations: This one’s been brewing for years now, and while still in early stages for standard display ads, expect initial integrations of AR experiences triggered by vertical banners, especially on mobile devices. AR for vertical ads has real potential, and could involve technology like interactive overlays or virtual product demonstrations. - Privacy-focused approaches: As privacy rules get tighter, advertisers are turning to contextual ads and first-party data to serve up relevant vertical banners while minimizing their dependence on third-party tracking — all while prioritizing user trust and compliance. - AI-driven creative optimization: Artificial intelligence is creeping into everything, and will absolutely play a larger role in automatically generating and optimizing vertical banner creatives. AI tools can analyze performance data and suggest instant variations in design and messaging for continuous improvement. - Seamless content integration: Native advertising is influencing vertical banners, with designs that aim to blend more seamlessly with the surrounding website content, reducing ad blindness, toning down flashy out-of-place ads, and enhancing user engagement. - Cross-device adaptation: With diverse screen sizes and device types emerging all the time, vertical banners will need to adapt more intelligently to provide optimal viewing experiences across desktops, tablets, and new formats like foldable screens. ### How does audience targeting impact vertical banner results? Audience targeting is a crucial cornerstone of effective vertical banner advertising, and should be a major part of your strategy. By selecting the specific groups of users who are most likely to be interested in your products or services, you can significantly increase the chances of enhancing your campaign’s performance. When it’s successful, the impact is felt all around: - Increased relevance: When your vertical banners are shown to a relevant audience, the messaging and visuals are more likely to resonate with their needs and interests, leading to higher engagement and a more positive perception of your brand. - Improved click-through rates (CTR): Targeted ads are more likely to generate clicks, since they’re shown to users actively seeking (or interested in) related information or products. A higher CTR is a clue that your ad is grabbing the attention of the right people. - Higher conversion rates: By reaching users already predisposed to your offerings, you increase the likelihood that their clicks will result in desired actions, such as purchases, sign-ups, or inquiries. Targeting helps connect with users further down the sales funnel. - Reduced wasteful spending: If you’re not using effective targeting, a significant portion of your ad impressions might be shown to users who have no interest in your products, causing you to waste your budget. Targeting ensures that your advertising spend is focused on reaching potential customers. - Enhanced ROI: Make your campaigns as cost-effective as possible by improving engagement, CTR, and conversion rates while reducing wasted spending. Valuable data insights: Analyze the performance of your banners across different audience segments. This helps provide valuable data on which demographics, interests, and behaviors are most responsive to your messaging, and gives you a roadmap for continuous refinement of your targeting strategies and creative content. --- ### App Installs: Definition and Best Practices URL: https://www.taboola.com/marketing-hub/app-install/ Last Modified: 2026-06-30 07:40:32 If you’re a marketer tasked with boosting mobile app installs, one of the best ways to target the right audience is through app-install ads. These specialized digital ads are designed to drive installs by putting your app in front of the people most likely to download and use it. In this guide, I’ll break down how app-install ads work, highlight the benefits across several popular ad platforms, and share some best practices. ## What Is an App Install Ad? An app-install ad is a digital advertisement specifically designed to encourage users to download and use a mobile app. App-install ads tend to be visually appealing, using high-quality images and video, with clear messaging and a direct call-to-action (CTA), often as a download button. Most major ad platforms — like Meta, Google, Realize, and social media channels like Pinterest, X, and Snapchat — offer app-install ads. Unlike traditional digital ads, app-install ads are optimized to track and measure app downloads. ## 6 Best Practices for App Install Ads ### 1. Use eye-catching visuals At the risk of sounding obvious, always use high-quality images and videos that showcase your app’s functionality and value. ### 2. Focus on the benefits, rather than the features Don’t just tell users what your app can do — let them know how it will make their life easier, or solve their problems. ### 3. Avoid long copy It’s best to use clear and concise headlines with direct CTAs, e.g., Install Now, Download Free, as these work best on mobile devices. ### 4. Target the right audience Regardless of the ad network you’re using (Google, Facebook, Realize), make sure to leverage their targeting capabilities. This will help you reach users based on location, interests, behaviors, etc. ### 5. A/B test your ad creatives Run two versions of your ads to test different headlines, visuals, and CTAs. This way, you can see which version drives the most installs. ### 6. Track user actions post-install Use tools like Meta’s SDK, or MMPs to track in-app behavior, such as sign-ups, purchases, and retention. ## What Do App Install Ad Platforms Do? App install ad platforms don’t just place your ads, they provide you with tools that are designed specifically for driving app downloads. Here are some key features offered by these platforms: ### Drive Installs Across Multiple Channels When you work with an ad platform that serves ads across multiple channels, like Meta or Google, your app install ads can appear in everything from social feeds to search results, video streaming platforms, and, in the case of Realize, the open web. ### Accurate Tracking of App Installs Most ad platforms can connect to MMP or SDK tools, which can track when someone installs your app and the actions they take afterward. You need this data to know which ad or platform drove the install or to effectively re-target users who didn’t fully convert. ### Use AI to Find High-Quality Users The major app install platforms offer AI-powered optimization, which leads not only to more app installs but also helps to find the highest-quality, aka most profitable, users. Those are the ones who are more likely to take valuable actions in the app, like subscribing to a membership or making a purchase. ### Audience Segmentation App install platforms can segment audiences based on the type of mobile device they’re using, OS version, location, interests, or even behaviors. Having this information at your disposal makes it easier to promote your app to users who are more likely to convert. You can also avoid wasting valuable ad budget on low-value users. You can consult our recommended list of app install ads for 2025 with features and benefits. ## Key Takeaways App install ads are an excellent way to find new mobile app users, especially when you run them on platforms that are optimized for mobile engagement, like Facebook, YouTube, Snapchat, or Realize. Each platform has its distinct benefits, from cross-platform promotion (Meta, Google) to real-time visibility (Twitter/X) to open web access (Realize). Most platforms also offer advanced targeting and tools for install tracking. Whether you’re launching a new app or scaling an existing one, app install ads are an instrumental part of your growth strategy. ## Frequently Asked Questions (FAQs) ### What is an app install ad on Twitter (X)? Twitter, or X, app install ads use app cards to promote mobile apps directly in user timelines. The cards include your app’s icon, ratings, and a download button, and are designed to drive app installs. ### How do I run an app install ad on Facebook Ads? Meta Ad Manager allows you to create an app install ad using its “App Promotion” objective. Head to Ads Manager and select the ad account associated with your app. Select “Create” and the “App Promotion” objective from there. You can opt for a manual app promotion campaign or automate the process by choosing “Advantage+ app campaign.” Once you’ve completed the various sections, including budget and schedule, audience, placement, ad creative, etc., click “Publish” to take your app ad live. ### How do I create an app install campaign in Google Ads? Google Ads offers ad campaigns to promote your app on Google-owned properties like Google Search, Google Display Network, YouTube, etc. Here are the steps to follow: 1. From the “Campaigns” page, choose “New Campaign”. - Choose “App promotion” as your campaign type. - Select the app you wish to promote from the Apple App Store or Google Play Store. - Choose a name for your campaign. Google recommends including the OS in the campaign name (iOS or Android). - Configure your campaign location and language settings. - On the Budget and Bidding page, set your average daily budget, focus, and target users. - Finally, create your ad assets (headlines, descriptions, images, videos, etc.), and publish your campaign. --- ### 6 Best App Install Ads Platforms URL: https://www.taboola.com/marketing-hub/best-app-install-ads-platforms/ Last Modified: 2026-01-13 10:24:30 In today's hyper-connected world, having a fantastic mobile app is just the first step. The real challenge lies in getting it into the hands of your target users. With millions of apps vying for attention, a robust app install advertisement strategy is no longer a luxury – it's a necessity. But where do you begin? The digital advertising landscape is vast and ever-evolving, with numerous platforms promising to deliver the elusive "install." From social media giants to search engine powerhouses and specialized ad networks, each platform offers unique strengths, targeting capabilities, and audience reach. This blog post will cut through the noise, providing an in-depth look at the best platforms for app install advertisements. We'll explore the pros and cons of industry leaders like Google Ads and Meta (Facebook & Instagram), delve into the unique opportunities presented by Twitter and Snapchat, and touch upon other powerful contenders. Our goal is to equip you with the knowledge to make informed decisions, optimize your campaigns, and ultimately drive the growth your app deserves. Let's dive in! ## 6 Top Advertisement Platforms to Drive App Installs ### 1. Facebook App Install Ads Facebook allows marketers to run in-depth app install ad campaigns across Meta properties, including Instagram, Messenger, and on third-party apps and websites through Facebook Audience Network. Also, by linking your app via the Meta Software Developer Kit (SDK), you can track installs, retarget users who didn’t complete the install and optimize ad delivery based on real-time performance. Key Features:  - Advanced targeting. - Lookalike audiences. - Advanced install and in-app targeting via Meta SDK. - Run ads across multiple platforms (e.g., Facebook and Instagram) in a single campaign. ### 2. Google App Install Ads Google app campaigns use automation and machine learning to remove the work from ad creation. Instead of creating individual ads, you provide the creative assets, like text, headlines, images, and videos. From there, Google generates the ads and tests them across Google properties, such as Google Search, YouTube, Gmail, etc. Google’s AI can also optimize the ads to drive performance. Key Features:  - Can advertise across Google-owned properties. - Automated bidding and ad optimization. - Cross-channel targeting while managing a single campaign. ### 3. X/Twitter App Install Ads Twitter is now X — however, its app install ads are still often called Twitter App Cards. App Cards allow advertisers to drive app downloads directly from user timelines on X. When users click on the card, they are taken directly to the appropriate app store. You can target by interest, keyword, device, and location using static images and videos, and X will ensure that your ad is placed in relevant timelines in real time. Key Features:  - Leverages real-time visibility in X’s fast-moving timelines. - App Cards are customizable and include install buttons in the ad. - Cards are visually appealing, which can lead to higher engagement rates. - Users can view an app's critical details in a single tweet. ### 4. YouTube-App Install Ads YouTube app install ads are managed through Google Ads. Because they’re in video format, they are very effective if you want to showcase your app’s functionality. You’ll need to upload your video creative into an App promotion campaign and choose YouTube as your placement. You can set the ads to appear before, during, or after the video content. Google will automatically serve your ad to relevant audiences to increase the chances of installation. When users click on the ad, they are sent directly to the app store. Key Features: - Video ads allow you to showcase your app’s functionality. - Great medium for high-impact storytelling. - Leverages Google Ads’ tracking and bidding capabilities. - High conversion potential with CTAs that link directly to the app store. ### 5. Pinterest App Install Ads Pinterest uses Promoted App Pins, which are designed to blend seamlessly alongside the platform’s regular pin formats. The ads appear in users’ feeds just like any regular pin, but they have an “Install” button that allows users to download the app. To run an app install ad on Pinterest, simply upload your ad creative and set your targeting preferences, based on interests, keywords, or custom audiences. It’s very important to ensure your app is well-represented in your ad creative. Also, remember to include a link to your iOS or Android app store. Key Features: - Native ad experience can lead to higher engagement. - Ideal for visual-first and lifestyle brands. - iOS users can install the app without leaving Pinterest. ### 6. Snapchat App Install Ads Snapchat is a social media platform geared toward younger audiences. Snap Ads are full-screen vertical videos that lead to the app store where users can download your app. Snapchat recommends that advertisers create at least one video for their ads, and to keep it short (3-5 seconds is recommended). To create your first ad, head to the Delivery section and select the “App Installs” objective and your Impressions goal. Select your Customer List Audience and choose an age and gender. From there you can set your ad budget and duration. You can also connect your app with Snapchat to gain further insights into user behavior. For example, general MMPs (mobile measurement partnerships) allow you to see what actions Android Snap users take on your app after they respond to your ad, while the SKAdNetwork allows you to keep track of app activities on iOS devices (iOS14+). Key Features:  - Ideal for reaching Gen Z and Millennial audiences.  - Works well for entertainment, gaming, and lifestyle apps. - Reach people who have taken actions in your app by creating a Mobile App Custom Audience. ## Key Takeaways Successfully navigating the competitive landscape of mobile app installs requires a strategic approach to advertising. As we've explored, platforms like Google Ads and Meta (Facebook & Instagram) offer unparalleled reach and sophisticated targeting, making them cornerstones of many app install campaigns. However, neglecting the unique opportunities presented by lower-funnel, performance platforms, which cater to distinct demographics and engagement styles, would be a missed opportunity. To truly maximize your app's growth, consider harnessing the power of AI-powered software that leverages predictive analytics and intent-based targeting to deliver unparalleled ROI for your app install advertising. --- ### KPI's (Key Performance Indicators): Types, Tracking, Measurement URL: https://www.taboola.com/marketing-hub/key-performance-indicators/ Last Modified: 2025-07-24 10:45:22 You and your team might have some of the best ideas out there when it comes to sales and marketing, and you might be doing great work. But, if you’re not properly tracking your work, you might never know it. You might even shift to a less effective strategy because of that lack of knowledge. On the other hand, candidly, there’s always a chance your marketing efforts just aren’t working, your ad spends are often wasted, and you and the team need a serious shift. But, again, if you’re not properly tracking your progress (or lack thereof), then you may never know any of that. In both cases, using key performance indicators, or KPIs, can help keep you on track and ever improving. ## Defining and Understanding KPIs KPIs are the metrics a team agrees upon that will be regularly tracked and assessed. They set an expectation for performance, be that in terms of revenue, enrollment, reach, product development, and so much more. ### How Do KPIs Differ From General Metrics or Statistics? KPIs differ from general metrics and statistics because they are specifically tied to a company's strategic goals, whereas metrics are broader data points used for analysis. You may use stats and data to help create your KPIs, but it won’t go the other way around. In marketing and advertising, KPIs enable data-driven decision making by providing quantifiable metrics that measure the effectiveness of campaigns, allowing marketers to track progress toward goals and optimize strategies based on real-time data and historical analysis. ### What Are the Key Characteristics of Effective KPIs (SMART Criteria)? Effective KPIs are characterized by the SMART criteria: Specific, Measurable, Achievable, Relevant, and Time-bound. These criteria ensure that KPIs are clear, quantifiable, attainable, meaningful, and tied to specific deadlines. Using SMART criteria helps avoid ambiguity, confusion, and conflict, leading to more effective goal setting and progress tracking. KPIs are essential for measuring digital advertising success in particular because they provide quantifiable metrics that track campaign performance and guide data-driven decision making. They help to assess the effectiveness of marketing strategies, optimize campaigns in real-time, and demonstrate tangible value to clients. ## Types of KPIs in Digital Advertising Specific key performance indicators are relevant to different advertising channels, since each channel will have its own unique goals and metrics. For instance, social media KPIs might focus on engagement, while email marketing KPIs could emphasize open rates and click-through rates. ### What Are Common KPIs for Measuring Campaign Reach and Awareness? Common KPIs for measuring campaign reach and awareness include impressions, engagement, website traffic, branded search volume, and share of voice. Impressions show how often content is displayed, engagement tracks interactions, and website traffic indicates how many people are visiting the site. Branded search volume measures how frequently people search for the brand, while share of voice compares the brand's visibility to competitors. ### What KPIs Are Used to Track Engagement and Interaction With Ads? KPIs to track engagement and interaction with ads include click-through rate (CTR), conversion rate, engagement rate, and cost per click (CPC). These metrics help assess the effectiveness of advertising campaigns and provide insights for optimization. Conversion-focused KPIs measure desired actions by quantifying the percentage of users who complete a specific defined action, often referred to as a conversion. This action could be anything from making a purchase to filling out a form or signing up for a newsletter. By tracking the conversion rate, businesses can see how effectively their website, marketing campaigns, or product is driving users toward the desired goal. ### What Are Important KPIs for Evaluating Cost-Efficiency and ROI? To evaluate cost-efficiency and ROI, key performance indicators should focus on profitability, cost control, and resource utilization. Important KPIs include Gross Profit Margin, Operating Expense Ratio, Return on Investment (ROI), Cost Performance Index, and Cost Savings. ## Identifying and Selecting Relevant KPIs ### How Do You Align KPIs With Your Specific Advertising Goals and Objectives? To effectively align KPIs with advertising goals, start by clearly defining your advertising objectives, then select relevant metrics, and finally, ensure these KPIs are specific, measurable, achievable, relevant, and time-bound (remember: SMART). This process ensures you're tracking the right data to measure progress toward your advertising goals. ### What Factors Should You Consider When Choosing the Right KPIs? When selecting key performance indicators, prioritize alignment with business goals, ensuring they are measurable, actionable, and relevant to your industry and target audience. Consider using the SMART criteria to guide your selection. Avoid vanity metrics and focus on KPIs that provide actionable insights to drive decision making. ### How to Prioritize KPIs Based on Their Impact on Business Outcomes To prioritize KPIs, businesses should focus on those that directly align with strategic objectives, have the greatest impact on desired outcomes, and are feasible to measure and track. This involves understanding the organization's goals, considering the impact of each KPI on key areas, and assessing the feasibility of data collection and analysis. To effectively involve stakeholders in KPI selection, start by clearly identifying them and their interests. Then, engage them in discussions, workshops, or surveys to gather their input and ensure the selected KPIs are relevant and meaningful. Regularly communicate the selected KPIs, their rationale, and how they will be used, fostering transparency and accountability. ## Tracking and Measuring KPIs The best-picked KPIs on the planet don’t matter a bit if you’re not tracking them closely, so pay attention so you can take action, and don’t overdo it. For a single marketing campaign, it's generally recommended to track between five and 10 KPIs. While the exact number can vary based on the campaign's goals and complexity, focusing on a manageable set of metrics helps ensure clear results and actionable insights. Watch out, too, for common challenges you may run into when measuring digital performance, which range from improper attribution to ad fraud to data overload and so much more. ### What Tools and Platforms Can Be Used to Track Digital Advertising KPIs? To effectively track digital advertising KPIs, businesses can utilize a combination of tools and platforms. These include website analytics platforms like Google Analytics, social media analytics within platforms like Facebook, LinkedIn, and Instagram, and ad-specific tools like Google Ads and Meta Business Suite. KPIs should be monitored and reviewed regularly, typically monthly or quarterly, to ensure that they align with business goals and provide actionable insights. The frequency can be adjusted based on the specific KPI and how quickly things can change within the business. ### How Do You Set Up Accurate Tracking and Reporting for Your Chosen KPIs? To accurately track and report on chosen KPIs, start by defining clear goals, then establish a robust data collection and analysis process, potentially using software or tools to automate reporting and create dashboards. Regularly review and refine your KPIs and reporting methods to ensure they remain aligned with your evolving goals. To ensure data accuracy and reliability in KPI tracking, organizations should establish clear data collection standards, implement robust validation processes, and regularly audit data quality. This includes defining KPIs clearly, validating data sources, cleaning and organizing data, and analyzing data with caution. Regular reviews and updates of data are also crucial. ## Analyzing and Acting on KPI Data ### How Do You Interpret KPI Data to Gain Meaningful Insights? By identifying patterns and correlations in the data collected, companies can gain valuable insights that inform future marketing decisions. For instance, if a particular marketing KPI, such as email open rates, shows a consistent decline, it may indicate the need to revamp email marketing strategies or content, or to shift to another approach. On the other hand, studying your KPI data may confirm that an approach is working, prompting you to lean into it even more. ### How to Identify Trends and Patterns in Your KPI Performance To effectively identify trends and patterns in your KPI performance, you need to define clear goals, track relevant metrics, and analyze data using various techniques like time series analysis, comparative analysis, and data visualization. Regular review and adjustments to your KPIs, along with communication of results to stakeholders, are crucial for continuous improvement. Be sure to let people know when you see that improvement: To communicate KPI performance effectively to stakeholders, use clear and intuitive visualizations like charts and graphs, provide regular updates on results, and tailor communication to the specific needs and interests of the audience. Focus on presenting key takeaways and insights, and actively solicit feedback for continuous improvement. ### When and How Should You Adjust Your Advertising Strategies Based on KPI Results? Advertising strategies should be adjusted regularly based on KPI results to optimize performance and maximize ROI. Adjustments should be made when KPIs consistently show underperformance, when new data indicates a need for changes, or when external factors necessitate a shift in strategy. The frequency of these reviews can vary, but regular monitoring is crucial for staying on top of campaign performance. A/B testing (a method of comparing two versions of a product, service, or website to see which performs better) can be applied to a defined KPI — and it should be. By testing different variations and measuring their impact, businesses can make data-driven decisions to improve their KPIs and ultimately boost their bottom line. Using an ad platform that offers AI-powered A/B testing for real-time optimization can be immensely helpful with this. ## Key Takeaways A KPI (or key performance indicator) is a quantifiable metric used to measure how effectively an individual, team, or organization is progressing toward achieving a specific objective or goal. Essentially, it's a benchmark that helps track performance and demonstrate success. Examples include profit margins, customer acquisition, return on investment, and more. KPIs are crucial for businesses because they provide a structured way to measure progress, track success, and ensure alignment with strategic goals. They act as measurable benchmarks, helping companies understand where they stand and identify areas for improvement. For any given marketing campaign, it’s a good idea to have between five and 10 KPIs, and it’s critical that you track them closely and take action when changes are needed, based on what you find. ## Frequently Asked Questions (FAQs) ### What is a good KPI for a brand awareness campaign? One good KPI for a brand awareness campaign is called “share of voice.” This metric measures how much your brand is being talked about and how it compares to your competitors. Other key metrics include brand mentions, social media engagement, and website traffic. ### How do you calculate return on ad spend (ROAS)? To calculate return on ad spend (ROAS), divide the total revenue generated by your ad campaign by the total cost of the ad campaign. The result is typically expressed as a ratio, such as 4:1 (meaning $4 of revenue for every $1 spent), or as a percentage (400%). ### What is the difference between a leading and a lagging KPI? Leading KPIs, also known as lead indicators, focus on future performance by predicting potential outcomes, while lagging KPIs, or lag indicators, reflect past performance and the results of past actions. Leading KPIs allow for proactive adjustments, whereas lagging KPIs offer a retrospective view and help evaluate the effectiveness of past strategies. ### What are some common pitfalls to avoid when using KPIs? Common pitfalls to avoid when using KPIs include selecting the wrong KPIs, measuring too much, failing to act on data, and relying on industry benchmarks without considering individual business needs. Also, insufficient data quality, inadequate communication, and ignoring reviews/adaptations can hinder effective KPI implementation. --- ### Advertisements: Your Idea-to-Execution Guide URL: https://www.taboola.com/marketing-hub/ad/ Last Modified: 2025-06-22 11:12:02 Advertisements are everywhere in our lives, and have been for some time now, to the point that you encounter them everywhere you go — both online and in the real world. In fact, the amount of ads that people are exposed to in just one day is estimated at around 5,000, reaching all the way up to 10,000 — that’s up from between 500 and 1,600 in the 1970s. While it’s true that we’ve grown accustomed to tuning a lot of them out, they’re still there, vying for our attention (and money), from TV commercials to the billboards we see while driving, and now the digital landscape that’s always finding innovative ways to reach us. There’s a reason advertising is such a huge industry: When a campaign has all the right elements, it works. Sometimes even when it doesn’t, an ad will find a niche in pop culture and take off on its own as a meme or memorable phrase. As someone who’s been in the advertising world for two decades, I probably pay more attention to every ad I see than most people, and am always interested in the strategy and creative (or lack thereof) behind it. A good ad goes deeper than a catchy jingle, funny tagline, or eye-popping colors — these messages are carefully crafted to elicit thoughts, feelings, and ultimately actions out of the viewer. Knowing all the moving parts of advertising is a valuable skill in today’s cluttered world, and not just for us ad nerds. Whether you’re looking to be more of a savvy consumer, thinking about advertising your own business, or just someone intrigued by how companies get their messages across, I’ve broken it down here to give you a holistic look at everything that makes an ad come to life. ## What Is an Advertisement? Starting at the heart of it all: At its core, an advertisement is a message that a business or organization pays to share with a particular audience. That message, which can take many forms, typically aims to promote a product, a service, or a general idea you want to get across. You want to target the demographic that’s most likely to take action, and the largest number of people within that group. ## Why Are Advertisements Important? You might gloss over hundreds of ads in a day, unaffected by most of them (though keep in mind, you also might not be their target.) I’ve been around campaigns that were so meticulously crafted that a test audience just one year outside the demographic didn’t respond to it, while the main group did. If advertising didn’t work, you’d certainly see a lot less of it, but it absolutely does, and that’s why businesses spend so much time and money on creating the right campaign to send out into the world. Ads play a crucial role for the companies creating them, the people who see them, and the product’s reputation. ### Getting Noticed I’ve not only been writing ads for my whole career, but have also seen things from the other side when doing advertising for myself — everything from copywriting services to the music I write and record. Being on that end of things provides a whole different perspective, and can be incredibly frustrating for any client who believes in their product, but isn’t breaking through the noise. Even for someone like me who’s spent years in the ad world, selling your own product and getting attention can be a whole different game. When a new product hits the market, it might be life-changing for a certain subset of people, but how would those potential customers even know it exists? That’s where targeted advertising is still the best way to cut through and inform people, with the goal of encouraging them to make a purchase, click a link, sign up, or whatever the desired action is. ### Brand-Building Think about some of the brands you recognize instantly. Maybe they’ve been around your whole life, remind you of a happy moment, and have a distinct "feel" or "personality.” That’s branding, and companies spend lots of time, money, and research building their product’s persona. Whether it comes across as seeming fun, reliable, innovative, or caring, consistent advertising helps create and reinforce a brand’s identity in the minds of consumers. It reaches outside normal advertising channels, too: I once had a creative director tell me that in the age of social media and 24-hour news, “how a company behaves is its advertising nowadays.” ### Standing Out In lots of industries, there’s no shortage of companies offering similar things. Think about areas like drinks, insurance, and apparel: The sheer number of options can be instantly overwhelming even to those who are actively looking to buy. Advertising helps a business highlight what makes its products or services different (and ideally better), encouraging consumers to choose them over the competition. ### Educating and Informing Advertising is usually associated with sales, but that’s not always the case. Sometimes, an ad is more informational, getting the word out about a new feature, explaining how a service works, or even raising awareness about a cause or issue. ### Supporting Media Channels Have you ever been watching a show or listening to a podcast when you’ve heard something like “support for this program” or “thanks to our sponsor today”? Many of the websites, TV channels, and radio stations we consume for free rely on that advertising revenue to keep operating, helping fund the entertainment and information we take for granted. ### Highlighting Available Options Ads can be a valuable source of information, and this knowledge helps consumers make more informed decisions about what to buy or use. They expose us to a wider range of options than we might otherwise encounter, leading us to discover those new products or services that better meet our needs, further our interests, or make life a little easier. Not only that, but they often highlight sales, discounts, and special offers we may not have known about otherwise, giving consumers the chance — and motivation — to save money on the items they’ve been looking to buy anyway. ## Understanding the Core Elements of an Ad Every ad, regardless of its format or where you see it, has a story behind it. Assembling that final product was a process, composed of these components all working together in the hope that the ad delivers its message effectively: ### The Message This is where it all starts. Before the ideas even begin, identifying the core message you want to communicate is the center of everything. Whether it’s about the features of a product, the benefits of a service, or the core values of a brand, a clear and concise message is essential for the audience to understand what the ad is trying to say. ### The Audience There are a lot of people in this world, and your first instinct may be to cast the widest net possible to get the largest number of viewers. But, that’s not necessarily the most efficient way, and advertisers carefully think about who the audience you’re trying to reach is — the more specific the better. Understanding the target audience (such as their age, interests, needs, and habits) is a crucial part of hitting your sales goal, and advertisers tailor their message, visuals, and the channels they use to reach these niche groups. ### The Medium A target audience isn’t the only thing that advertisers have to narrow down — you also have to think about how the ad is going to reach them, and where they’ll see or hear it. Nowadays, there are plenty of options: They include traditional channels like TV, radio, and magazines, or digital platforms like social media, websites, podcasts, and email. Choosing the right medium depends heavily on who the advertiser is trying to reach, and how it’s going to help execute the creative idea and overall message. ### The Goal Every ad has a purpose. Reaching potential customers is the most obvious one, but it could also be driving traffic to a website, directly encouraging sales, or even just increasing brand awareness. No matter what, having a clear objective helps the advertiser see if their ad is successful during the campaign. ### The Call to Action (CTA) The CTA, or “call to action” is where this effort wraps up. It’s the final push to get viewers to take that last step. An effective ad will tell you exactly what the advertiser wants you to do next, such as "Visit our website," "Call now for a free quote," "Learn more," or "Buy now." It may seem like a no-brainer since it’s usually not the most clever or funny part of an ad, but a clear call to action guides and motivates the audience on what to do next if they're interested. ## Which Ad Format Should I Use? Back before the internet days, you had your four main options for advertising, but the digital world has expanded the selection significantly. Here’s a breakdown of the most common ad formats: ### TV With a TV ad, especially during a widely televised event, you’ll be able to reach a large audience, but it can be seriously expensive to produce and broadcast. ### Radio A radio ad is generally much more affordable than TV, and can target specific geographic areas or listener demographics. But, with the advent of podcasts and music services, the number of radio listeners aren’t what they once were. ### Print Print, like newspapers and magazines, can reach even more niche audiences. Much like radio, though, it’s changed since the pre-internet days, as readership has generally declined or moved online. ### Out-of-Home (OOH) (Billboards, Posters) These are great for high visibility in specific locations, and can also be used in clever ways that enhance the creative message of your ad. ### Search Engine Marketing Unlike trying to blanket the widest audience possible, search engine marketing (SEM) ads target users who are actively searching for specific keywords related to the product or service. It’s an excellent option for pinpointing the right potential customers and getting them to click through. ### Social Media Ads Another great option for honing in on your target audience, social media can narrow it down by age, interests, demographics, online behavior, and more. ### Display Ads (Banner Ads) The earliest ads on the internet were display ads, and they’re still active as ever today. These visual ads appear on websites and apps, often rotating through slides or an eye-catching animation with a CTA at the end. Find out more about display ads here. ### Video Ads With a similar format to a TV ad, video ads often play right before a YouTube or social media video. They can be an excellent and creative way to deliver a message quickly, but can also be invasive to a viewer and create a negative association with your brand. ### Email Marketing For people who have signed up for an advertiser’s email list, this direct communication reaches an audience you know wants to hear from you, and is on the lookout for the latest sales, offers, and updates. ### Native Advertising If you’ve ever seen an ad that doesn’t quite look like an ad, that’s native advertising. These are designed to blend in with the surrounding content on a website or platform, making them less invasive and distracting. ### Influencer Marketing Influencers are people on social media who have a dedicated online following. Collaborating with them to promote products or services gives advertisers a trusted spokesperson with a built-in target audience who can give their fanbase a more personal review. It’s a more authentic feel hearing it from the influencers, rather than from the company via a traditional ad. ### Podcast Advertising Podcasts are more popular than ever, and also more loaded up with ads. That’s because they’re a solid option for reaching engaged listeners within specific topic areas. Having the podcast host read ad copy increases the chances that it’ll be heard by their fanbase, and it’s trusted since it’s a familiar voice. The hosts often riff off-script, too, making it less rigid and corporate-sounding, and keeping the listeners interested and tuned in. Pre-produced advertisements that run on their commercial breaks still work, too — even if listeners hit the 15-second skip button, they’ll most likely still catch the tail end of the ad (or it’ll catch them at a time when skipping isn’t possible, like while driving). ### Interactive Ads Old-school advertising media like TV and radio were passive, meaning you sat there and got advertised to. But, with current constant connectivity, ads allow users to interact with them, such as through polls, quizzes, or previewing a game. ## How to Create an Ad Which Will Meet Your Marketing Goals (8 Steps) When you see a simple billboard or banner ad, it may not seem like a whole process went on to get it up and running. Creating an effective advertisement involves a journey that goes through multiple rounds, though, and involves different departments and teams. Here’s a look at some of the key steps along the way: ### Step 1: Clearly Define Your Goals This is the main point the entire campaign hinges on, and the central message that everything needs to come back to if it’s going to work. The goal is everything that advertisers hope to achieve with this campaign, and the more specific the better. For example, is the goal to increase brand awareness by a certain percentage? Or drive a specific number of sales? Get a certain number of people to visit their website? No matter what, it all starts here. ### Step 2: Thoroughly Understand Your Target Audience Next up, advertisers think about who you’re trying to reach — and not just who, but what about them, such as their needs, wants, and pain points (and how this product can solve those). Thinking about behaviors like where they spend their time online and offline can be enormously helpful when trying to reach the right crowd, as the more you know about your audience, the better you can tailor your message. This is why ad platforms that utilize AI to predict user intent — going way beyond basic demographics and interests to deliver truly personalized ad experiences — are likely to see more success. ### Step 3: Develop a Compelling and Clear Message Once an audience is identified, advertisers can think about what you want to say. It needs to be quick in order to cut through the clutter and compete with people’s short attention spans: An ad should be easy to understand, memorable, and highlight the benefits of a product or service, focusing on the problem it solves for them, or how it can make their lives better. ### Step 4: Choose the Right Advertising Formats and Channels Even if you narrow your audience down to the exact group that’s most likely to buy, the next obstacle is figuring out where to reach them, and that can be a challenge. The good news is that there are a ton of different ways to do that nowadays. Selecting the ad formats and platforms that align with your strategy and budget are a crucial part of a campaign, and when done right, really pay off. ### Step 5: Create Engaging and Effective Ad Creative This is where the creative team comes in and works their magic. Copywriters, art directors, and designers will think up a few different ideas, and present them to the creative director, who selects one (or more) ad creatives they think works best for hitting the main message and resonating with the audience. It needs to be attention-grabbing, visually appealing (if applicable), and consistent with the brand's overall look and feel. ### Step 6: Set Your Budget and Determine Your Campaign Timeline Decide how much you're willing to spend on your advertising campaign and how long you want it to run. Even if you don’t have a huge budget like a major corporation, you can still create and launch an effective ad campaign by using the right channels and tools. ### Step 7: Implement and Launch Your Advertising Campaign The launch is your campaign’s big moment that all these other steps have been building up to. Once your ad is ready, it’ll go out into the world on the platforms you’ve chosen. But, that’s not where it ends! In fact, this is the start of when you’ll… ### Step 8: Monitor Performance and Make Necessary Optimizations Once your ad is live, you won’t just sit back and hope it works. Instead, you’ll track key metrics (like clicks, views, and conversions) to see how it's performing in the real world. For online ads, be prepared to make adjustments on the fly to improve their effectiveness — it’s a crucial part of what helps you understand which parts are working and what needs to be altered, both now and for future campaigns. To this end, be sure to choose an ad platform that offers deep insights into user engagement and the overall effectiveness of your campaigns, tracking lower-funnel, performance metrics, as well as a series of other helpful indicators to see how users are interacting with your ads and the content around them. ## Evaluating the Effectiveness and Impact of Your Ad Using analytic tools helps you keep up on what’s working and what’s not, so you can always be uncovering valuable insights into how to fine-tune your ads for even better results. Here are a few things to know on how to assess the effectiveness of your online ads: ### Keeping Up with KPIs: Numbers That Matter So, how can you tell if your online ads are actually doing their job? Important metrics, often called key performance indicators (KPIs) are measures that tell you how well your ad is performing. An important one to keep an eye on is impressions, which is how many times your ad popped up on screens. Another is click-through rate (CTR), which tells you what percentage of people who saw your ad actually clicked on it. A higher number here generally means your ad was a success, caught their eye, and prompted action. ### Conversions Conversions are also a big one. Did your ad lead to the desired action? Did people make a purchase? Did they sign up for a newsletter? Did they fill out a contact form? Conversions directly link your ad to your marketing goals, i.e., the actions you want people to take after clicking, such as buying something or signing up for an email list. The conversion rate tells you what percentage of those clicks actually turned into the goal you wanted users to complete. ### CPC/CPA Cost per click (CPC) is how much you pay each time someone clicks, and cost per acquisition (CPA) tells you how much it’s costing you to get someone to complete that click (or whatever the desired action is that you want them to take). ### Return on Ad Spend Return on ad spend (ROAS) helps you see if the money you're spending on ads is actually bringing in more money than it's costing. Reviewing these numbers regularly in your ad platforms gives you a pretty good idea of how things are going. ### Return on Investment Did you make more money from the ad than you spent on running it? Similar to ROAS, return on investment (ROI) is a key indicator of the financial success of your advertising efforts. ### Brand Awareness and Recall Did your ad increase awareness of your brand? Do people remember seeing your ad later on? While this one’s definitely harder to measure directly, brand awareness is a long-term benefit of advertising. ## Key Takeaways An advertisement is a paid message from an identified sponsor aimed at promoting a product, service, or campaign, and they’re crucial for businesses to raise awareness, build brands, and drive sales. Understanding the core elements of an ad (message, audience, medium, objective, CTA, etc.) is an essential part of creating effective campaigns, as is choosing the right ad format, which depends on your specific goals, target audience, and budget. A structured approach to creating ads, from defining goals to tracking results, is necessary for success, along with evaluating the effectiveness of your ads through key metrics, which allows for optimization and insights to keep improving it. ## Frequently Asked Questions (FAQs) ### What persuasive techniques exist for ads? Advertisers can use different methods to persuade audiences, including appealing to emotions, celebrity endorsements, creating a sense of urgency or scarcity, leveraging social proof (i.e. showing that others like and use the product), and presenting logical arguments or data. ### How do you identify your target audience’s needs? Simply put: Research and analysis. This can include conducting surveys and focus groups, analyzing competitor audiences, creating detailed profiles of your ideal customers (buyer personas), examining website and social media analytics, and actively listening to conversations online related to your industry and potential customers. Having a well-researched profile of your typical target audience is going to help all departments make decisions along the way. ### How do you A/B test your ads? A/B testing, also known as split testing, involves creating two or more slightly different versions of your ad and showing them to similar segments of your audience. By tracking which version performs better (such as by getting more clicks, views, conversions, etc.), you can identify which elements of your ad are most effective and use that information to optimize your campaigns. You can also supercharge the process with platforms that offer AI-powered A/B testing, which lets you automatically test multiple variations of your ad creative, headlines, and other elements across different audience segments, and optimize in real time. This approach to A/B testing helps you quickly identify the highest-performing ad combinations, saving you time and resources while maximizing your advertising results. --- ### Understanding Ad Groups: Structure Your Ads to Reach the Right Audience URL: https://www.taboola.com/marketing-hub/ad-group/ Last Modified: 2025-06-30 08:58:42 In the ever-evolving landscape of digital marketing, ad groups serve as fundamental components for orchestrating effective online advertising campaigns. Whether you’re a novice exploring the basics of advertising for your small- or medium-sized business or a seasoned professional seeking to refine existing ad strategies, a thorough understanding of ad groups is paramount. Let's delve into a detailed examination to help you understand the definition, significance, structure, performance monitoring, and optimization of ad groups. ## What Is an Ad Group? An ad group is a collection of advertisements that are organized around a common theme or objective within a digital advertising campaign. In platforms such as Google Ads, ad groups encompass a collection of ads that share the same keywords. The primary purpose of an ad group is to facilitate the management and targeting of related advertisements to optimize both relevance and performance. This structure allows advertisers to isolate different aspects of their campaigns, ensuring a more strategic and focused approach. ## Importance of Ad Groups ### Organization and Control The organization of ad campaigns into ad groups offers substantial structural benefits. By categorizing ads based on specific themes or target demographics, advertisers can maintain greater control over the performance and management of their campaigns. This also enables marketers to streamline their strategies and ensure that each ad aligns with its designated objective. ### Relevant Targeting Ad groups allow advertisers to craft targeted messages that resonate with particular audiences, or ideally, audience segments. By clustering ads around a specific set of keywords, marketers can ensure that their advertisements are highly relevant to users’ search queries. This relevance can significantly enhance user engagement, leading to increased click-through rates (CTR) and conversion rates. ### Performance Insights Ad groups facilitate the acquisition of data-driven insights regarding the performance of advertisements. By monitoring metrics such as CTR, conversion rates, and cost-per-click (CPC), advertisers can evaluate the effectiveness of each ad group. This analytical approach enables marketers to identify successful strategies while also pinpointing areas that may require improvement or refinement. ## Structure and Components of Ad Groups ### Keywords At the core of every ad group lies a set of carefully selected keywords. These keywords serve as the foundation upon which ads are built and are critical in determining the overall visibility of an advertisement. A well-researched keyword list can substantially impact the success of an ad campaign by connecting relevant queries with targeted ads. ### Captivating Ads Each ad group comprises one or more advertisements that are designed to be triggered by the associated keywords. These ads must be specifically tailored to appeal to the target audience and should contain compelling copy and visuals that encourage users to engage with the ad. ### Targeting Settings Targeting settings define the demographic and geographic attributes of the audience that will see the ads in the ad group. Advertisers can specify factors such as age, gender, and location to ensure that their messages are delivered to the most relevant audience segments. You can also take things a step further by using an ad platform that offers AI-powered, intent-based targeting, where users are targeted by their predicted behavior, rather than simply their demographic group. ### Ad Extensions Ad extensions enhance the functionality and appeal of advertisements. These additional features provide users with more information about the product or service being advertised, such as links to specific pages on a website, location details, or contact information. Utilizing ad extensions can lead to improved visibility and CTR. ## How to Monitor the Performance of Your Ad Groups ### Setting Up Tracking Establishing robust tracking tools is essential for effectively monitoring the performance of ad groups. Tools such as Google Analytics or native platform dashboards allow marketers to capture and analyze data on key performance indicators (KPIs), including conversions, clicks, and impressions. ### Analyzing Key Metrics A systematic analysis of key metrics is critical for gauging the effectiveness of ad groups. Advertisers should routinely examine CTR, conversion rates, and CPC to gain insights into user behavior and engagement. ### A/B Testing A/B testing is an integral part of your advertising plan. Implementing a variety of comparative tests within ad groups provides valuable insights into which elements resonate most effectively with the target audience. By comparing variations of the same ad, marketers can identify effective strategies and refine their messaging accordingly. ## How to Optimize Ad Group Performance ### Refine Keywords Regular refinement of the keyword list is crucial for maintaining relevance and performance. Advertisers should conduct periodic reviews of their keywords, pausing those that are underperforming and integrating new keywords that align with current trends or user intent. ### Enhance Ad Copy Optimizing your ad copy is another vital part of improving ad group performance. Testing different headlines, calls to action, and descriptive elements can yield significant improvements in engagement. Ensuring that the ad copy aligns well with the target audience's interests is necessary for maximizing this impact. ### Utilize Negative Keywords Incorporating negative keywords into ad groups can help exclude irrelevant search terms, preventing ads from appearing for queries that are unlikely to convert. This practice can lead to a more efficient allocation of budget and an overall improvement in ad relevance and CTR. ### Allocate Budgets Wisely Finally, prudent budget allocation is key to optimizing ad group performance. Advertisers should continually review the performance of their ad groups and adjust budgetary allocations based on performance metrics to increase the overall effectiveness of their campaigns. ### Key Takeaways Ad groups facilitate the management and analysis of advertising efforts within a campaign, since grouping related ads enhances targeting effectiveness and user engagement. As with individual ads, performance monitoring is critical to a campaign's success, and continuous optimization is required. ## Frequently Asked Questions (FAQs) ### What is an ad group example? An example of an ad group could be a digital marketing agency focusing on "SEO Services." Although the marketing agency offers several types of services, this ad group would contain keywords related only to SEO (search engine optimization) alongside tailored ads promoting various aspects of that agency's SEO offerings. This ensures relevance and coherence to specific searches and target customers. ### What are groups in Active Directory? Active Directory groups are collections of user accounts or other groups, which allow administrators to perform certain activities en masse (like allowing permissions) rather than having to perform the same actions for each individual user. In relation to ad groups, Active Directory usually refers to distribution groups or collections of users from specific demographics. ### What is the difference between a campaign and an ad group? A campaign encompasses a broader organizational structure that may include multiple ad groups, each focused on specific objectives or themes. An ad group, in contrast, is a more targeted subdivision within a campaign, concentrating on particular keywords and audiences to enhance the precision of your advertising efforts. --- ### Conversion Rates: Tangible Growth Beyond Traffic URL: https://www.taboola.com/marketing-hub/conversion-rate/ Last Modified: 2025-05-26 11:25:25 In today’s highly competitive market, success isn’t only about driving more traffic to your website. Converting that traffic into action, whether it’s a sale or an email newsletter sign-up, is just as important as getting eyes on your content. The metric for tracking all of this? Conversion rate. Every click counts, no matter the size of your business. But, what happens when you see an increase in other key performance indicators but not your sales? That’s when it becomes critical to assess how many people are completing the actions that you want them to take when they see your brand online. Understanding conversion rate, and how to optimize it, is essential for ongoing growth. ## What Is Conversion Rate? Conversion rate is the percentage of users who take a desired action on your website or on an ad that you might be running. This could be a purchase, but it could also be a newsletter sign-up, a digital download, filling out a contact form, or even watching a demo video. What’s considered a conversion will depend on your business goals, but it could look like an e-commerce store customer making a purchase, or a service provider receiving a booking for a consultation. Ultimately, your conversion rate reflects how effective your website or ad campaign has been at turning interest into action. ## Why Conversion Rate Matters Conversion rate is more than just another metric to track, it’s a window into how well your marketing is performing. A high conversion rate means that your messaging resonates with your target audience, that the user experience on your site is good, and that you have a compelling offer. On the other hand, a low conversion rate could indicate issues in your marketing funnel in moving users from the top to the bottom, that your ads aren’t targeting the right people, or that your creative isn’t that compelling. Understanding and improving your conversion rate can generate an immediate increase in revenue if you do it correctly, particularly if you’re already driving high levels of traffic to your site or landing pages. ## Conversion Rate Calculation Calculating your conversion rate is fairly straightforward: Conversion Rate = (Number of Conversions ÷ Total Visitors) x 100 For instance, if your website had 2,000 visitors last month and 100 of them made a purchase, your conversion rate would be 5%. ## What Is a Good Conversion Rate? There’s no one-size-fits-all benchmark for conversion rates — they can even vary within each industry sub-category. But, for some averages, these key industry standards can help you gauge where you are right now: ### Gaming Average conversion rates in the gaming industry are around 1.5-3%, with mobile apps converting at a slightly higher rate due to low-barrier-to-entry downloads. ### Health Health and wellness sites, particularly those selling fitness-related plans or supplements, average around 3-5% year round, with seasonal spikes at New Year and in the month leading into summer. ### Automotive Conversion rates for the auto industry are lower, around 4-6%, due to longer sales cycles and high-value sales. However, lead form conversions or test drive bookings are often slightly higher. ### Tech Tech products average around 2-5% conversion rates, but these can vary significantly depending on what type of tech you sell and what you’re tracking as a conversion. Demo sign-up conversions are often higher, and for top-performing campaigns can be around 10%. ### Travel Travel websites often convert around 2-4%, with mobile users typically converting at a lower rate than desktop users due to the complexity of travel arrangements and booking flows. ### Finance Insurance, loans, and banking sites typically see 4-7% conversion rates. This is particularly the case when contact forms are short and offers are personalized for the user. ## Factors Affecting Conversion Rate ### Website Speed and Mobile Experience Slow-loading websites are one of the top reasons that users abandon a page before converting. Even a one-second delay in page load time can lead to significant drops in conversions. On mobile devices, where users are often browsing quickly and on smaller screens, this can be even more of an issue. Optimizing for speed means compressing your images before you upload them, reducing scripts loading on the site backend, and using mobile-responsive design templates. Many tools for building landing pages that are tailored for this type of high-level performance can eliminate this technical guesswork for smaller teams. ### Targeting Match You can’t convert users who were never interested in your content to begin with. That’s why audience targeting is vital to get right. Switching to intent-based targeting, over identity-based, is one of the best moves you can make to increase your conversion rate. This means that your ads are more likely to reach users who are already in a decision-making phase of the marketing funnel. ### Creative Fatigue Even the best ad creative has a shelf life. When potential customers see the same creative over and over again, the performance naturally declines due to overexposure. Not only does this reduce your click-through rate and conversions, but it weakens the perception of your brand. Refreshing visuals and headline copy regularly means that you can offer something new to your audience. ### Lack of Clarity If a user can’t tell within a few seconds what your offer is and why it should matter to them, they’re more likely to leave. Ambiguity in product descriptions, unclear pricing, or vague calls to action (CTA) can all reduce the likelihood of a conversion being made. ### Checkout or Sign-up Issues There’s nothing more frustrating than trying to check out or sign up for something and it doesn’t work. Long forms, surprise fees, or required account creation are all some of the biggest reasons that users abandon carts or leave a sign-up halfway through. ## How to Improve Conversion Rate ### Run A/B Tests A/B testing allows you to compare different variations of elements in your ads or landing pages like heading, images, button placement, or product descriptions to see what resonates more with your target audience. For instance, changing the CTA from “Buy Now” to “Get Yours Today” might boost urgency and increase clicks. ### Build Better Landing Pages Your landing pages should be laser-focused on the specific goal you want a user to complete when they get to that page. Instead of sending traffic to your homepage, create campaign-specific pages that speak directly to your target audience and their intent. These pages should be as distraction-free as possible, load quickly, and they should also be visually aligned with the creative of the ad that brought the user to that landing page. There are now numerous tools that can help you build landing pages quickly, from drag-and-drop options to AI-assisted page generators. ### Leverage High-Visibility Ad Formats High-visibility ad formats like full-screen vertical videos, interactive carousels, and rich displays often stop users from scrolling and draw attention to your offer. The most important thing here is to match your messaging with the creative and the format you’re using. For instance, motion to highlight product features in a video ad, or a carousel to create a narrative around your offer. ### Use Trust Signals People often trust other people more than brands, so including product reviews, testimonials, case studies, and user-generated content in your landing pages can significantly improve your credibility. Visual trust indicators like SSL certificates, secure payment icons, and media mentions also help reassure potential customers who have never worked with you. ### Retarget Strategically Retargeting campaigns can be highly effective and re-engage users who showed interest but didn’t convert the first time. To be effective, though, these ads must go beyond a generic reminder and offer something that the first round of ads didn’t. Use dynamic retargeting to show ads for the product that the specific user has in their abandoned cart or was considering. Offer a discount or an added benefit to entice them back and make the conversion. ## Tools Used to Optimize Conversion Rate ### Landing Page Builders Tools that let you build and deploy custom landing pages, especially those that require no coding, allow small teams to test landing pages more quickly and generate new pages faster. ### Predictive Targeting Platforms Solutions that go beyond traditional interest targeting, instead targeting by intent, help get your message in front of users who are already in the decision phase of their sales journey. ### Creative Optimization Tools Platforms that are built with A/B testing capabilities usually offer options for you to rotate assets before creative fatigue sets in for your audience. Repurposing social content into new formats is one of the best ways to get more from assets you already have. ### All-in-One Assistants AI tools can help manage campaigns from creation to optimization, making it easier for your team to handle tasks that would typically require multiple platforms or specialists. ## Key Takeaways Understanding and optimizing for improved conversion rates is essential for any business. While traffic is still important, what you do with that traffic is ultimately what determines your profitability. By understanding what impacts your conversion rate and where you can make improvements, you can optimize existing and future campaigns to give you measurable impact within weeks. ## Frequently Asked Questions (FAQs) ### How do you calculate conversion rate for e-commerce? You divide the number of purchases made by the total number of visitors, then multiply by 100 to give you an overall percentage for your conversion rate. ### What are some common reasons for low conversion rates? Low conversions often happen as a result of poor targeting, slow website load times or performance, unclear calls to action, or issues around the checkout or sign-up. ### How can A/B testing help improve conversion rates? A/B testing allows you to quickly identify which version of a page, or which on-page elements, are performing better. From there, you can make data-informed decisions that help you increase conversions. ### What are the best practices for optimizing landing pages? Using clear headlines, fast-loading designs, and strong calls to action are the best options for your landing pages. Matching the content to your ad also helps provide consistency. ### How does mobile optimization affect conversion rates? A mobile-friendly design is critical, as a poor mobile experience can lead to higher bounce rates on your site and landing pages, and lower conversions overall. --- ### Closed-Loop Marketing: What It Is, How It Works, How to Implement URL: https://www.taboola.com/marketing-hub/closed-loop-marketing/ Last Modified: 2025-05-26 11:16:38 Small to medium-sized businesses (SMBs) in particular often face an uphill marketing battle — tight margins, rising customer acquisition costs (CAC), and an oversaturated digital ecosystem can make your job even more difficult. This isn’t helped by the feeling that many big platforms like Google and Meta are plateauing, leaving teams scrambling for scalable strategies that generate real, measurable growth. In this landscape, closed-loop marketing marketing can be the practical solution to many an ROI problem. There’s no vague metrics here: Instead, closed-loop marketing gives you the ability to connect the dots between marketing activities and real revenue outcomes. ## What Is Closed-Loop Marketing? Closed-loop marketing is a data-focused marketing methodology that directly links marketing campaign efforts to customer actions and revenue outcomes. In other words, it “closes the loop” by providing data from the sales and conversions side back into the marketing side of the business. The process starts by tracking a potential customer’s journey from their first interaction to making a purchase or other form of conversion event. This data is then used to assess which marketing activities are most effective, allowing marketing teams to double down on what’s working well and drop efforts that are less effective. The loop is closed because the performance insights from the bottom of the sales and marketing funnel directly inform decisions made at the top of the funnel. ## How Does Closed-Loop Marketing Work? Having the ability to integrate your marketing platforms, analytics, and sales data is essential for closed-loop marketing. When a lead converts into a paying customer, their journey must be traceable back through every touchpoint. That means every email they opened, ad they clicked on, and landing page they visited should all be trackable so it’s possible to identify what actually drove the sale. By creating feedback loops from real customer behavior, marketing teams can continually refine creative, targeting, and budget allocation for advertising. Alongside this, any content marketing, email marketing, or SEO efforts can be adjusted based on data. When you’re working with a smaller budget and an even smaller team, every action must move the needle and that’s exactly what closed-loop marketing aims to achieve. ## Why Closed-Loop Marketing Is Important When acquisition costs are continuing to rise for businesses of all sizes, it’s vital that resources are used as effectively as possible. Closed-loop marketing shows you exactly which campaigns, creatives, and channels are driving the most conversions and where cuts can possibly be made. This is particularly important when working with a small team with limited time and budget. When every click, view, and visit can be accurately tied to a tangible outcome, marketing efforts become less about guesswork and more about refined strategy. This gives teams the confidence to invest more in what’s working and pull back quickly on what isn’t, without needing to wait weeks to see results. Closed-loop marketing is also a great strategy for reducing internal friction that could arise between marketing and leadership. With shared data to review, subjective decision-making can be simplified. Closed-loop systems allow for rapid testing and adjustment, ensuring that campaigns are efficient and results remain steady across platforms being invested in. In other words, closed-loop marketing keeps growth sustainable and scalable. ## How to Implement Closed-Loop Marketing ### 1. Understand your Buyer's Journey Before you can close the loop, you need a clear understanding of the entire journey. This involves designing specific marketing campaigns and content strategies tailored to each stage of the customer's path, from broad awareness tactics aimed at introducing your brand, through nurturing consideration with valuable content, to targeted conversion efforts that drive action. A robust full-funnel strategy ensures you engage potential customers effectively at every touchpoint and establishes the necessary data points for tracking their movement, which is not so feasible for advertisers to achieve in the fragmented data and platform landscape. ### 2. Connect Your Systems You can’t run a closed-loop marketing strategy without integrating and connecting all of your sales and marketing platforms to share data. This includes ad platforms, analytics tools, and any customer data software you’re using. This integration is now fairly seamless, depending on which tools you use. Shopify, WordPress, CRM tools, and analytics tools like Google Analytics or Google Search Console have options for syncing within each platform. This foundational connection enables you to track customer paths from first click to final sale. ### 3. Use Intent-Based Targeting Rather than focusing solely on demographic data or identity-based segmentation of your audience, intent-based targeting prioritizes user behavior instead. Predictive models can be used to read user intent and allow for more accurate audience targeting, which is especially useful when social and search channels aren’t scaling as efficiently as they previously were. ### 4. Repurpose Creative With limited time and budget, repurposing high-performing content across other channels is a must. Tools that allow for fast adaptation, like importing social content into high-visibility formats such as carousels or vertical ads, helps reduce creative fatigue on your team and increases ad longevity without the need for new weekly content. ### 5. Optimize Landing Pages Driving traffic is only half of the story — converting that traffic is just as important. Quickly building landing pages that align with your creative and overall messaging helps maintain consistency across your marketing channels and minimize audience drop-off. A/B testing these landing pages is also key to improving overall conversion rates. ### 6. Automate as Much as Possible Leverage automation within your tools to manage campaign builds, budget pacing for ads, and optimization in real time. ## Benefits of Closed-Loop Marketing ### Improved ROAS and Lower CAC When you’re able to double-down on proven channels and creative, this type of strategy reduces guesswork and maximizes every dollar spent. This often leads to higher return on ad spend and a more efficient customer acquisition funnel. ### Faster Insights Rather than waiting weeks to assess overall performance, closed-loop marketing can provide real-time feedback for decision-making within days. You can pivot your approach faster and see improved results within two or three weeks. ### Scalable Growth Instead of spreading your time and budget across every possible channel, without ever really investing in any one or two, closed-loop marketing means that you can scale in the areas that are proving to be the most beneficial. This approach reduces the noise that can easily overwhelm a small marketing team, giving you the ability to concentrate on where it matters most. ## Key Takeaways For small to medium-sized businesses looking to grow in crowded marketplaces, a closed-loop marketing strategy is essential. By linking marketing efforts directly to outcomes, your team can focus more on performance and real results. It’s no longer simply about doing more, but doing what works, and faster. ## Frequently Asked Questions (FAQs) ### What is the key difference between traditional and closed-loop marketing? Traditional marketing often relies on assumptions and educated guesswork, taking information from broad metrics. Closed-loop marketing integrates every part of the sales and marketing system, which allows you to tie actual revenue outcomes to tracked customer data. ### How does closed-loop marketing improve marketing and sales alignment? By showing which campaigns result in real conversion, both sales and marketing teams can prioritize what works best and collaborate on strategies that are tied directly to performance, rather than opinion. ### What tools are essential for closed-loop marketing? Integrated analytics, CRMs, ad platforms, and landing page builders are all essential pieces of the puzzle that make up a closed-loop marketing system. ### How do you measure the success of a closed-loop marketing strategy? Key performance indicators should be tracked for any integrated system. This could include ROAS and CAC if running paid ads, increases in revenue across all digital channels, and faster testing times for landing pages or ads. ### What are some examples of closed-loop marketing in action? A good example of closed-loop marketing would be a direct-to-consumer brand repurposing some of their top performing Instagram posts into a carousel ad. With predictive targeting, this ad can narrow focus on specific audiences, and conversions can be measured directly as users click on the ad to be taken to a landing page. --- ### Canonical URL: What Is It? How Does It Affect SEO? URL: https://www.taboola.com/marketing-hub/canonical-url-what-is-it-how-does-it-affect-seo/ Last Modified: 2026-06-22 08:32:39 For many growing websites, duplicate content is a common issue that can lead to SEO problems down the road. Whether it’s product pages with similar descriptions, sorting parameters in URLs that are slightly off, or multiple paths leading to the same content, these issues can significantly dilute your ranking power in search engines. Canonicalization, or canonical URLs, are one of the best ways to manage this and help search engines better understand which page is the preferred version of a specific piece of content. This is essential for ensuring that your site maintains strong signals without self-competing for space in search results pages. When used correctly, canonical URLs streamline your website structure and keep you off the hook for duplicate content penalties. ## What is a Canonical URL? A canonical URL is the preferred version of a webpage that you want a search engine to index and show to users in search results pages. This is typically designated by an HTML tag inside the head section of a page, the rel=”canonical” tag. These tags tell search engines which version of the page is considered the primary one, versus any possible duplicates or similar pieces of content elsewhere on the site. The goal of a canonical URL is to consolidate link signals and avoid duplicate content penalties if there are multiple URLs on the site with the same or similar content. SEO value through backlinks and other ranking signals can then pass through to this single page with the canonical URL, rather than being spread across multiple similar pages. You may be wondering why you would even keep duplicates of content on your site to begin with. For many businesses, especially those that are product-based, pages may be very similar and contain small variants, such as a change in product color, but the rest of the content is the same. The use of canonicals allows websites to keep all the variant pages online, without risking duplicate content penalties from search engines. ## Types of Canonical URL ### Self-Referencing Canonical URLs A self-referencing canonical points to the URL of the page that it’s already on, to ensure consistency and explicitly note that this is the preferred version of a page. This is particularly important to set up if alternate paths also exist, for example: https://example.com/blog should have a self-referencing canonical set up going to https://example.com/blog, along with the alternate URL of https://blog.example.com also having a canonical set up for https://example.com/blog. Even if your site doesn’t currently have duplicate pages in place, adding self-referencing canonicals at the start helps to prevent future indexing issues. ### Cross-Domain Canonical URLs If your site syndicates content or publishes information from a partner site, using a cross-domain canonical is useful. This tells search engines that the original content lives on a different domain from the one it’s being hosted on, so they can direct SEO credit to the original source. For sites that publish elsewhere, cross-domain canonicals help republish this content with correct attribution. ### Parameter-Based Canonical URLs Parameters can be set up for all kinds of reasons, from tracking a campaign to filtering products. Even though the content remains the same as the non-parameter page, URLs can often be treated the same by search engines. Setting up a canonical to the original source, such as example.com/blog/seo-tips, protects against SEO dilution when adding a parameter such as example.com/blog/seo-tips?utm_source=newsletter. ### Mobile and Desktop Canonical URLs If your website uses separate URLs for the mobile and desktop versions, rather than responsive resizing, you’ll want to canonicalize your URLs. The rel=”canonical” tag should be placed on the mobile version and point back to the desktop site, along with a desktop site tag of rel=”alternate” going to the mobile version. ## How to Set Up Canonical URLs ### Add a Canonical Tag in HTML The most common way to add a canonical to your URLs is through HTML tagging, updating the coding to include <link rel=”canonical” href=”https://example.com/preferred-url” />. ### Set Up Canonicalization on Dynamic URLS If new URLs are automatically created by filtering or sorting, such as on product variations, it’s important that you set up canonicalization pointing to the base page within your CMS settings. For Wordpress or Shopify sites, there are plugins that can help you with this. ### Set Up HTTP Headers in Non-HTML Content For non-HTML pages like PDFs, you’ll want to set up canonicals here, too. These can be added into the header file of PDFs or other non-HTML content on the backend of your site. ## Canonical URL Best Practices ### Always Use Absolute URLs Every canonical you use should be an absolute URL, not a relative path. This means including the full protocol and domain e.g. https://example.com/page instead of simply /page. This ensures that there’s a clear and unambiguous reference for search engines to follow when indexing. ### Be Consistent with Formatting Always choose a standard format that you can re-use when creating new canonicals. For instance, decide whether you want to use the www or non-www version of your domain. Inconsistencies can cause duplicate indexing and fragment the SEO efforts you’re implementing on your site. ### Let Paginated Pages Self-Reference Many site owners make the mistake of pointing paginated content like blog archives back to page one of their blog. Instead, each paginated page should be canonicalized to itself as a self-referencing canonical URL. This helps preserve the unique value of each archived page and gives search engines a better understanding of your content depth. ### Avoid Pointing Canonicals to Redirects Any canonicals you put in place should point to live, indexable URLs, rather than any 404 pages or 301 redirected pages. Doing so can create confusion as there are then multiple steps that search engines must go through to find live content for them to index. ## Key Takeaways Using canonical URLs should be part of your SEO strategy. In most cases, you’ll never need to manually update these and your CMS or plugins will take care of it for you. But, it’s important to understand what canonicals are and how they work in case you ever do need to make manual updates yourself. While they may be a more behind-the-scenes element of SEO, canonical URLs have a significant impact on your site visibility in search results, the stability of your SEO rankings, and the overall health of your site. ## Frequently Asked Questions (FAQs) ### How do canonical URLs prevent duplicate content issues? These URLs signal to search engines which version of a page should be indexed, consolidating SEO value onto one page and preventing penalties for duplicate content. ### What happens if you don’t use canonical URLs? Search engines may index multiple versions of the same on-site content, which can dilute ranking signals for SEO and potentially lower your site’s visibility in search results. ### When should you use a canonical URL? You should use a canonical when there are multiple URLs with very similar or identical content on your site. Filtered pages, product variations, or syndicated articles all create this type of content, so a canonical should be implemented. ### Can canonical URLs point to a different domain? Yes, you can point canonical URLs across domains, which is typically what sites publishing syndicated content use to prevent indexing issues. ### How do search engines interpret canonical URLs? Search engines treat canonicals as a strong suggestion, but never an absolute direction, about which page should be indexed and ranked. --- ### Breadcrumbs: What Are They? How Do They Work? URL: https://www.taboola.com/marketing-hub/breadcrumbs/ Last Modified: 2026-06-22 08:34:28 Good website navigation is the foundation of any online user experience. Without this, users can’t find what they need quickly and efficiently. Optimal navigation also improves user engagement and reduces bounce rate. One of the most overlooked areas of website navigation is the breadcrumb trail. Much like in the classic story of Hansel and Gretel, the breadcrumb trail is designed to lead users through the site’s infrastructure — essential for letting them know where they are and how to make their way back again. ## What Are Breadcrumbs? Breadcrumbs are a type of secondary navigation that displays where a user currently is on a site, mapped within the hierarchy of the full site structure. They’re typically found at the top of a webpage, just below the main navigation or header. As noted, the name comes from the Hansel and Gretel fairy tale, in which two children scatter breadcrumbs on the forest floor to help them find their way home. In web design, breadcrumbs perform a similar function by helping users find their way back to the homepage of a site. They’re typically formatted something like this: Home > Blog > SEO Tips > What Are Breadcrumbs? ## Types of Breadcrumbs ### Hierarchy-Based Breadcrumbs The most common type of breadcrumb is a hierarchy-based layout. This is where the user’s location is based on the site hierarchy or a folder structure based on the overall site composition. This is commonly used for e-commerce and news or blog sites. Home > Electronics > Smartphones > iPhone 16 ### Attribute-Based Breadcrumbs E-commerce sites also use attribute-based hierarchies, where breadcrumbs are determined based on the attribute or feature that the user has selected. Home > Men > Shoes > Size 10 > Black ### History-Based Breadcrumbs If the user has been on the website for some time, they may see breadcrumbs that reflect their specific history on the site. Home > Blog > About Us > Contact > Careers ## Why Are Breadcrumbs Important? ### Improved User Experience One of the biggest reasons that breadcrumbs are important for website owners is that they improve the overall user experience on the site. Breadcrumbs serve as a visual map to help users understand where they are on the site and how they got there. For instance, if someone lands on a subfolder page from a search engine result, they may not know where they are on the site if there’s no breadcrumb in place. The context a breadcrumb provides can be useful for first-time visitors who don’t know the site structure. With breadcrumbs in place, users don’t have to rely solely on onsite navigation to direct them around the site. Instead, they can simply click one or two levels up from their current page to browse more content. ### Reduced Bounce Rates Since breadcrumbs make it easier for users to find related content, this encourages them to browse further and reduces the overall bounce rate on the site. A user who might otherwise leave the website after viewing only one page is more likely to click through to a broader category or a parent topic when a breadcrumb is visible. This is one of the best approaches for keeping people on your website for longer and gives them more opportunities to convert via a purchase or sign up. ### Enhanced Search Engine Optimization (SEO) Breadcrumbs help Google and other search engines get a better understanding of the website’s structure, which ultimately can help support other search engine optimization techniques you might be using. Google can even display breadcrumb trails in its search results in place of longer URLs. This is a cleaner presentation overall and helps enhance click-through rates (CTR). Internal linking is a critical part of SEO on a site, and having breadcrumbs in place supports this by distributing authority across your site. ### Lower Cognitive Load Cognitive load refers to the mental effort that a user must exert trying to process information. Without clear navigational cues, users can quickly become overwhelmed and lost, especially on larger websites. Breadcrumbs reduce the overall cognitive load that a user must experience by giving them a quick reference point for where they are and where they can navigate to next. ## Breadcrumb Best Practices ### Keep It Simple Use clear and concise labelling at all times and, if possible, mirror existing language on your website to maintain consistency. Avoid jargon or any page names that are too long, as this can be confusing. ### Start with the Homepage Always start your breadcrumb trail at the homepage to create a better sense of structure and uniformity for users, wherever they are on your site. ### Use Separators Common separators like > or / make it easier for users to read the breadcrumbs and know what the hierarchy looks like. Choose the style that best fits your website design and user expectations. ### Make Breadcrumbs Clickable Other than the current page that the user is on, all elements of the breadcrumb trail should be clickable. This means that a user can navigate back to any part of the hierarchy easily. ### Ensure Responsive Design Breadcrumbs should still be visible on mobile devices, but auto-responsive to resize on different screen dimensions without breaking. ## How to Implement Breadcrumbs ### Use a CMS Plugin For sites built on popular CMS’s like Wordpress or Shopify, there are plugins that can be added to allow for breadcrumb integration without any hard coding. Tools like Yoast SEO offer an SEO-focused breadcrumb option. ### Hard-code in HTML or CSS Developers can also hard-code breadcrumbs into a custom site structure using simple HTML or CSS. This can look like: ### Build a Specific Framework For frameworks like React or Angular, developers can use UI library resources that include breadcrumb components to add them to a site. ## Key Takeaways Breadcrumbs are a powerful, yet often underused, feature of website navigation. Their benefits are numerous and can make a significant impact on how a user journeys through your website. By implementing breadcrumbs strategically, you can create a more intuitive and engaging user experience moving forward. ## Frequently Asked Questions (FAQs) ### How do breadcrumbs improve website usability? Breadcrumbs help users to understand where they are on a website and make it easier for them to move to other places within the site’s hierarchy. ### Do breadcrumbs help with SEO rankings? Yes and no. While breadcrumbs are an important way to optimize your site for search engines, they don’t automatically benefit you in terms of rankings. There are hundreds of factors that impact where a site ranks in search results, with breadcrumbs and easy-to-use navigation only being one part. However, as user engagement on your site is also a ranking factor and breadcrumbs can support this, they also indirectly help improve your site SEO in this way. ### What is the best way to structure breadcrumb navigation? Hierarchy-based breadcrumbs are typically the best option for most sites, starting at the homepage and then using clearly labeled, clickable links from there. ### Where should breadcrumbs be placed on a website? Ideally, breadcrumbs should be placed just below the main navigation on the page, or the header. It’s important that breadcrumbs are visible as the page loads. ### What are some examples of bad breadcrumb implementation? Bad breadcrumbs are typically those that aren’t clickable, are missing the homepage in the trail, are inconsistent in their structure, or are overloaded with irrelevant attributes that can confuse users. --- ### Best Ad Creatives in Health: What Works for Conversions URL: https://www.taboola.com/marketing-hub/health-ad-creatives/ Last Modified: 2025-12-01 20:02:34 Health advertising spans a wide range of companies, including consumer health businesses, wellness apps, pharmaceutical giants and startups, over-the-counter medication providers, telehealth services, and more. Some companies have to follow regulatory compliance and ethical standards in their marketing, depending on the industry and where they’re promoting products. Health businesses succeed in their advertising by focusing on how they can improve users’ lives. In a crowded field, there’s plenty of opportunity to get creative visually and verbally, educate their prospects, and build brand authority and promise with every campaign. Health ads can provide important education and awareness, playing a role in public health and understanding disease. Digital marketers across health companies have an opportunity to connect to emotions and tell human stories. ## 7 of the Best Ad Creatives in Health (and Why) Health advertising has to stand out to cut through user and creative fatigue. Display ads, landing pages, engaging social, and varying types of ad formats will all play a role in success. Here are seven great health ad creatives that broke through the noise, to inspire your own work. ### 1. Pfizer’s Preventive Care If you’ve got the budget for a celebrity spot in your ads, make sure it’s fun and gets the message across. Pfizer’s Covid vaccination ad with American football player Travis Kelce showed off the athlete’s strength, and his endorsement for getting regular vaccinations. It was well-timed, as Covid vaccination rates showed steady decline in the U.S., and brought light to a topic that can be mundane or easily forgotten. It was also light-hearted, bringing the humor (Kelce’s mom) that pharma companies don’t always embrace in their advertising. https://youtu.be/oklKtQS3JCY Brands can use health ads like this to educate viewers on the importance of preventive care, as well as reinforce their own authority in a particular market — in this case, Pfizer reminded audiences that they were an inventor of the Covid vaccine. ### 2. Brushing Basics Colgate’s work in underserved communities around the world, donating toothbrushes and toothpaste to kids who don’t have dental care access, was on display in its “Bright Smiles, Bright Futures” campaign. It’s a fine line to walk for a brand to humbly brag about charitable work, but Colgate’s ad creative used joyful videos to focus on smiling kids to tell their story. https://www.youtube.com/watch?v=7-TZCUGeqQY Using video, a voiceover, and a few key numbers, such as the number of years they’ve worked in South Africa and number of kids they’ve helped, created an impactful ad that backs up Colgate’s brand promise — and reinforces why a basic act like tooth brushing is essential. ### 3. A Mascot, Done Well Remember the Nasonex bee? The animated mascot, voiced by Antonio Banderas, actually broke through the crowded field of over-the-counter allergy medications. The bee’s adventures, either led or stymied by seasonal allergy symptoms, were entertaining, informative, and made the brand name recognizable. Drug mascots or cartoons can fall flat or induce cringe, but the bee hit it just right. With lots of competition in this particular subset of medications, Nasonex found a way to stand out in the pharmacy aisle. https://www.youtube.com/watch?v=Lj7Nsbp8fQ8 ### 4. Destigmatizing Mental Health Care During a time of increased mental health care needs, a shortage of therapists, and an increase in online healthcare platforms, BetterHelp stepped in with a broad campaign across video, social media, TV spots, and podcast ads. Its “Open Up” ad got real with a woman facing a crisis, then working through her panic with an online counselor. BetterHelp’s ads opened the door for more, similar telehealth therapy services that promise easy, fast access to counselors. BetterHelp quickly established its brand as a modern, useful service for tech-savvy prospective patients. Its platform connects patients with licensed, established mental health counselors, and by early 2025 counted more than 5 million users. ### 5. Introducing a New Vaccine How to tell the story of a new vaccine that can prevent sexually transmitted cancers? The team behind the launch of Merck’s Gardasil had to both introduce a product and educate a swath of the public on a disease and treatment many may never have heard of. The advertising around the product was a mix of both branded and unbranded, focusing primarily on educating parents on why they should vaccinate their kids and at what age. Merck has used a mix of approaches over the years of marketing Gardasil, talking directly to teens as well as parents. This is an example of advertising that succeeded both in revenue — almost 40% growth in sales to $5.7 billion in 2021 — and in public health, with nearly 60% of adolescents vaccinated within a few years. Healthcare marketing like this can promote understanding of a topic among prospective patients and promote long-term wellbeing. ### 6. Broadening Beauty Standards For many years, print and TV ads for beauty products were pretty standard: Airbrushed, softly lit women photographed in closeup. Or TV ads which showed something similar, with off-camera fans gently blowing a flawless woman’s hair. Beauty products company Dove shifted those trends when it launched its “Real Beauty” campaign to take on female self-image. It brought images of all kinds of women, of all sizes, ages, races, and more, showing them head-on, often without makeup — something creative in the beauty industry hadn’t shown before. Dove’s initial success led to more iterations of the concept across TV, social media, billboards, and more, most recently encouraging girls to stick with sports with the #keepherconfident tag. They’ve taken on self-criticism, encouraged female role models and, in the process, raised Dove’s profile and rebranded it as a female-first, empowering brand while also rewiring typical beauty standards. ### 7. Generating User Content In the beauty arena, especially skincare, user-generated content (UGC) is hugely popular and profitable for businesses. When agency Evolut analyzed a number of beauty ads, more than a third of the successful ads were in this UGC format — more than single image, video, or carousel formats. This LANEIGE still is a perfect example of the UGC trend, having actual users and/or influencers use the product live, whether it’s a product demo, unboxing video, or before-and-after makeover. Consider how to incorporate this type of authenticity to appeal to your users and attract new prospects. ## Key Takeaways Health advertising is a crowded field, spanning the gamut from prescription medication to consumer wellness products. To stand out, focus your creative on the human impact, the positive outcomes of the product or service, and use the opportunity for education and facts wherever possible. --- ### Ad Copy: Writing Effective Messages That Sell URL: https://www.taboola.com/marketing-hub/ad-copy/ Last Modified: 2025-05-26 09:24:03 Businesses are in a constant race to compete for consumers’ short attention spans today. One way they do it is with killer advertising copy — the carefully crafted text that promotes a company’s products or services, designed to capture attention, communicate value, and drive action, usually sales or subscriptions. Global ad spending is expected to cross the $1 trillion mark for the first time this year, with 75% of those dollars going toward digital ads, according to a report from eMarketer. With stakes so high, your brand needs to nail ad copy with effective, clear messaging. Here’s a look at why ad copy is important, how it’s used, and how to write ad copy that makes your brand unforgettable. ## What Is Ad Copy? From headlines and body text to calls-to-action (CTAs) and taglines, ad copy is everywhere. The No. 1 purpose of ad copy is to persuade the reader, viewer, or listener to take a specific action, whether that’s making a purchase, downloading an app, subscribing to a service or newsletter, or just learning more about what a brand has to offer. Good ad copy isn’t just pretty prose that’s there to inform — it has to engage, persuade, and align with a brand’s voice and values. It shows (not just tells) the unique selling proposition (USP) of a product or service while addressing the audience’s pain points, and how it can solve them. Strong ad copy tells a compelling story with limited space, creating an emotional connection while selling a specific solution to a specific problem. ## Why Is Ad Copy Important? Here’s a look at why ad copy can make (or break) a brand’s reputation and trust with potential customers: ### 1. Drives Conversions, Revenue Ad copy directly correlates to conversion rates and, ultimately, revenue. Well-written messages that resonate with your target audience can lift campaign performance and move the needle for your advertising efforts. In digital advertising, it’s even more critical that your brand’s ad copy gets it right the first time: The average cost per action in Google Ads across all industries is about $49 for paid search and $75 for display ads, which adds up fast. Meanwhile, 61% of businesses pay $0.11 to $0.50 per click on Google Ads. ### 2. Creates Brand Awareness and Recognition Nike’s “Just Do It” and “Run Like A Girl” slogans and accompanying ads have made it a household name in sports apparel. Consistent, memorable ad copy doesn’t always need to be splashy to work: Through repeated exposure to bespoke messaging, consumers become familiar with a brand’s voice, values, and offerings. This recognition builds trust and credibility, which sets your brand apart from competitors and helps your business grow over the long term. ### 3. Differentiates From Competitors In saturated markets, ad copy can make your brand stand out from a sea of sameness where other players are hawking the same products or services. By using ad copy that emphasizes the problems you can solve for your target audience, the ad becomes less about your company and more about the consumer you want to transact with. That’s how you differentiate yourself from the pack. ### 4. Targets Specific Audiences Impactful ad copy speaks directly to specific audiences — their needs, wants, dreams, fears, and pain points. It just gets them. Tailoring your messaging to different demographics or buyer personas shows that your business understands its target audience, helping it build relevance and engagement, and helping drive higher conversion rates. In turn, this leads to more revenue. ## How Is Ad Copy Used? In the digital space, ad copy appears across numerous formats and channels, including: ### Search Engine Marketing (SEM) Search ad revenues climbed to $102.9 billion in 2024, up nearly 16% over the previous year, making it the top player in online ad spend. Search ads rely heavily on concise, keyword-rich ad copy that aligns with user intent. Google Ads, Bing Ads, and other search platforms display text ads alongside search results, requiring copywriters to create succinct messages that pack a punch with limited characters. ### Social Media Advertising Social media ad spend trails just behind search at $88.8 billion, but saw 36.7% YoY growth. This explosive surge in social ad spending coincides with the rise of popular platforms like TikTok, YouTube, Instagram, and LinkedIn. Each of these major social platforms has its own ad formats and audience preferences, which require platform-specific copywriting approaches. Fun, short-form videos that have spawned TikTok trends, for instance, don’t necessarily translate as well on LinkedIn, which is more text-dependent for a professional audience. ### Display Advertising Display ads combine visuals with copy, appearing on websites, apps, and social media platforms. Display ad copy has to complement design elements to capture attention and drive clicks, often with limited real estate to get the job done. Despite a clear shift toward digital ads, traditional advertising channels are still relevant for many brands. These include: ### Print Advertising Newspapers, magazines, brochures, and direct mail all demand tailored copywriting approaches that match the medium’s audience and reading experience. For instance, many luxury brands invest heavily in sleek, sexy ads in high-end publications, because they know their target audience of affluent, high-net-worth individuals reads them. ### Broadcast Advertising Radio and television commercials depend on clever scripts that engage listeners or viewers within short timeframes, often under a minute. For example, look at brands that soar (or flop) with their Super Bowl ads. While the upfront spend on those campaigns reaches the millions, the ROI can be incredible for a brand that gets it right. ### Out-of-Home (OOH) Advertising Billboards, transit ads, and other outdoor formats also demand ultra-concise copy that people can digest quickly while on the move. Travel to any big city and you’ll be bombarded with OOH ads that have the power to stop you in your tracks. ## How to Write Great Ad Copy Most copywriters have formal training in a writing or creative field, but some of the best copywriting doesn’t require intensive education or a fancy pedigree — it simply requires solid research and putting yourself in the audience’s shoes. Here are some tips on how to write ad copy that converts: ### 1. Understand Your Audience Effective ad copy starts with taking the time to know your audience. Research their demographics, pain points, desires, buying behaviors, and language preferences. Create detailed buyer personas to guide your copywriting messages so they resonate with the intended audience. ### 2. Highlight Benefits, Not Just Features Too many companies get wrapped up in tooting their horns and espousing cool features of their service or product, but that’s not what consumers care about — they want you to make it clear what’s in it for them. Focus on the value of your product or service, such as how it solves problems, fulfills specific needs, or makes life easier. Translate technical jargon or specs into tangible benefits that speak directly to your customer’s needs. Be relatable! For Facebook ads, for instance, give someone three major benefits, each one building on the last one, and add some sort of performance guarantee to drive the value home, says Jon Benson of Sales Copy Secrets. ### 3. Write Compelling Headlines Headlines should grab attention and keep your audience wanting more. Think of them as the gateway to your ad content. Use strong action verbs, pose questions, include numbers, or create a sense of urgency so consumers engage quickly. Roughly 80% of people will read a headline, but only about 20% will keep reading the rest of your copy, so make your headline do the heavy lifting: “Your first line needs to be a reality pattern interrupt,” Benson explains. ### 4. Use Clear, Concise Language Unlike content marketing, which relies on long-form educational or explanatory content, smart ad copy is straightforward and typically short. Avoid jargon, complex or run-on sentences, and unnecessary words. Aim for simplicity and clarity, especially if you have less space to work with. ### 5. Include a Strong Call-to-Action (CTA) Every piece of ad copy needs a clear call-to-action that tells the audience exactly what to do next. Whether you want them to shop, sign up, read more, or get started, your CTA should be compelling, action-oriented, and create a sense of urgency where it makes sense. “It needs to be a two-step closer,” advises Benson. “First, get them to agree to a simple truth, then give the clear call-to-action. That gives it just a little bit more kick.” ## How to Test Ad Copy ### A/B Testing A/B testing involves creating two or more versions of an ad with some slight tweaks in the copy elements, such as the headline, body text, or CTA. Once you send out the different versions, you can then measure which one performs better. This data-driven testing method helps you improve your messaging based on actual audience responses rather than guessing what will perform well. This is why it's worthwhile to work with ad platforms that offer AI-powered A/B testing, allowing you to continuously test and optimize in real time. ### Multivariate Testing More complex than A/B testing, multivariate testing looks at how multiple factors interact with each other. This approach helps you see the most effective combinations of copy elements that will result in the desired outcomes. ### User Testing and Feedback Getting direct feedback from your target audience can offer valuable insights into how your ad copy lands (or misses) with customers. You can gather this feedback through surveys, focus groups, or one-on-one interviews to gain qualitative data about your messaging’s impact. ### Performance Metrics Analysis Track key performance indicators (KPIs) to gauge how well your copy performs in the wild. This includes measuring click-through rates, conversion rates, search rankings, cost per acquisition, and return on ad spend to understand if your ad copy is moving the needle. These metrics can help you find trends and optimize your messaging backed by data. ## How to Improve Ad Copy Performance ### Use AI and Automation A recent HubSpot survey found that 43% of marketers now use AI to write copy, create images, and generate new ideas, but be aware that while AI tools can help you ideate and fine-tune your copywriting strategy, the copy they generate can often sound generic. AI automation can certainly help you generate variations, though, as well as predict performance and find opportunities for improvement, enabling marketers to create more compelling messages at scale. ### Incorporate More Video and Visual Content Visual formats, including short-form video, images, and live videos, have outsized popularity and preference. They also generate high ROI for marketers when done right. Using strong visual elements with your text helps your brand create more engaging and memorable ads that have the potential to go viral. ### Personalize It Creating personalized ad copy — with a little help from AI to coordinate work across platforms — can lead to a 35% lift in marketing performance. And, according to Attentive’s 2025 Consumer Trends Report, 96% of consumers say they're likely to purchase from brands that send personalized messages. In other words, the more you tailor your message to your audience, the more engagement you’ll see. ### Optimize Copy for Voice and Visual Search AI voice assistance and visual search tech is gaining popularity, so your ad copy needs to be responsive to conversational queries and visual search patterns. Consider how people speak informally when they use voice assistants, and which visuals will light up image recognition. ## How to Match Ad Copy to Audience Funnel Stage There are three main funnel stages, and your ad copy needs to meet your audience where they’re at in each stage of their journey. You should also work audience segmentation into the mix as you write copy for each funnel, considering audience segments based on demographics, interest, behaviors, or previous interactions with your brand. ### Top-of-Funnel (TOFU) Copy Awareness-stage copy should focus on addressing pain points, introducing your brand, and providing information of value. At this stage, it’s not about the hard sell: Aim to educate and engage potential customers who don’t yet know that your product or service can solve a problem they may have. ### Middle-of-Funnel (MOFU) Copy This is called the consideration stage, and copy should detail your unique selling proposition, showcase how your solution outperforms your competitors, and give evidence of your claims through testimonials, case studies, or data points. MOFU copy aims to nurture leads who are actively considering potential products or services. The goal is to keep driving them down the funnel. ### Bottom-of-Funnel (BOFU) Copy Decision-stage copy creates urgency, addresses potential objections, offers incentives, and issues strong calls-to-action that result in immediate purchase decisions. BOFU copy targets prospects who are ready to buy but may need a final nudge to get off the fence, so your word choices matter more than ever at this stage. ## Key Takeaways Memorable ad copy combines persuasive language with clear value propositions to drive specific actions from your audience. Remember, understanding your audience and their pain points is fundamental to creating copy that resonates and converts, and that testing and optimizing your copy across channels is an ongoing process, not a one-time deal. Different funnel stages will require different copywriting approaches and messaging strategies that meet your customers at various stages in their journey. ## Frequently Asked Questions (FAQs) ### What are examples of effective ad copy? Compelling ad copy is clear, benefit-focused messaging that addresses specific customer needs. It doesn’t trumpet features, but emphasizes benefits and how it solves problems. It also needs a strong call-to-action, an authentic brand voice, and concise language that grabs (and keeps) attention. Nike’s “Just Do It” is a prime example of simple yet effective messaging, as is Dollar Shave Club’s “Shave time. Shave money,” which touches on the dual benefit of convenience and saving money. ### What is the PAS formula in ad copy? The PAS (Problem-Agitation-Solution) formula is a copywriting framework that follows a simple three-step sequence: - Problem: Identify a problem or pain point your target audience regularly experiences. - Agitation: Elaborate on the negative consequences of the problem, deepening the emotional connection with the audience. - Solution: Share how your product or service is the ultimate solution to the problem, and why. The formula works because it taps into our psychological triggers. First, it establishes relevance by acknowledging the reader’s challenge, then intensifies the emotional response before finally offering relief through your brand’s solution. ### What are the best tools to optimize ad copy? There are a variety of tools to help marketers create, test, and fine-tune ad copy, including AI writing assistants (Copy.ai, Jasper, and ChatGPT), A/B testing platforms (Optimizely, VWO, and Google Optimize, Realize), SEO tools (SEMrush, Ahrefs, and Moz), emotion analysis tools (IBM Watson Tone Analyzer and Grammarly), which evaluate emotional tone and copy impact, and heatmap/user behavior tools (Hotjar and Crazy Egg), which reveal how users interact with ad copy on landing pages. The best tool for your business to sharpen your ad copy depends on your industry, tech capabilities, and budget. --- ### Conversions: Types, Measurement, Optimization URL: https://www.taboola.com/marketing-hub/conversion/ Last Modified: 2025-05-27 08:20:08 No matter how vital performance metrics like click-through rate are, companies cannot survive on traffic alone. To profit and thrive, businesses must turn clicks into customers: In other words, they need to convert them. “Conversions are the clearest way to measure whether your marketing is actually doing its job,” explains Chris Coussons, founder of Visionary Marketing. “You can have all the clicks in the world, but if no one’s converting, there’s no real value being created.’’ According to Coussons, one of the most common ways brands lose out on conversions is by being pushy, instead of providing a clear path from a relatable problem to an effective solution. However, when brands engage with their audiences and create content that makes them feel understood and confident, “the conversions tend to follow.’’ ## What Is a Conversion? A conversion is when a company convinces a person to take a specific action, making the transition from a lead to a customer. Whether they are buying a product or service, or signing up for a free trial or subscription, a conversion is the ultimate goal of your marketing efforts. ## Types of Conversions Types of conversions vary according to industry, but for e-commerce businesses, the most important conversion is product purchases. For service-based businesses, a conversion may be booking a service or consultation. However, when a person signs up for an email list, downloads a manual, fills out contact forms for a promotional event, or requests a quote, these are important, smaller-scale conversions that can lead to a purchase or booking. Watching videos, clicking through on a post, or spending time on your website also represent essential micro-conversions. ## What Is a Conversion Rate and How Is It Calculated? The easiest way to calculate conversion rate is to take the number of sales, bookings, or other specific desired customer actions, divide it by the total number of visitors, and multiply that number by 100 to get a percentage. “For example, if 10 people buy from a landing page that had 100 visitors, that’s a 10% conversion rate,” says Rodrigo César, co-founder and CEO of SSinvent. “That number tells me if I’m driving quality traffic and if my pages are doing their job.” ## How to Optimize for Conversions ### Speed and UX To figure out where to focus optimization efforts, it’s crucial to determine where traffic is dropping off. For instance, if you have a high click-through rate but a low conversion rate, the problem could be technical. “Fast-loading, mobile-friendly pages lead to more conversions,” notes Jessica Hitchen, SEO content executive at Quirky Digital. ### Compelling Headlines and CTAs Once you’ve cleared up any technical issues, the best way to increase conversions is through content that incites consumer action, whether buying, downloading, or subscribing. Clear headlines, subject lines, and calls-to-action (CTAs) that compel people to keep clicking are the first step toward doing this. Utilizing special offers to create a sense of exclusivity or urgency and personalizing CTAs for specific audience segments can further optimize content for conversions. ### A/B Testing To ensure your content resonates with your intended audience, experts recommend A/B testing, a type of experiment in which two versions of headlines, images, CTAs, forms, or other content are targeted toward similar audiences simultaneously to see which performs better. César has found through A/B testing that “simplifying pages, reducing distractions, and creating a clear, compelling value proposition has the biggest impact.” ## Role of Content in Conversions ### Content Should Address Pain Points The purpose of content is not just to drive traffic, it’s to drive action, and knowing your audience's frustrations is one of the most effective ways to do this. Companies can learn more about people's needs through online surveys and market research or by engaging directly with individuals on social media. ### Content Should Solve a Problem Once companies understand their audience’s pain points, they can address them directly with blog posts, demos, and other content that shows how their products and services will make consumers’ lives easier. Hitchen recommends using testimonials, reviews, and accreditations as evidence or “social proof” of this. “This also helps with EEAT, or expertise, experience, authoritativeness, and trust, which Google loves,” she adds. ### Content Should Be Tailored and Targeted To achieve the highest possible conversion rate, content should be tailored and targeted to segments of your audience. For instance, someone clicking through on your newsletter, spending time on your website, and even putting a product in their cart, may be far enough down the marketing funnel where a special offer may convert them. Conversely, someone just spending a few minutes on your website may benefit from more blog content to make them feel more “warmed up and informed,” Coussons says. ## How to Leverage Different Channels for Conversions ### Identify High-Performing Content Knowing what content engages your audience is crucial for leveraging it across different channels. By using tools like Google Analytics, brands can begin to examine which posts are driving traffic and what people are clicking on, to brainstorm content for other channels. ### Repurpose on a Different Channel Leveraging different marketing channels is essential for maximizing the return on content. César recommends repurposing high-performing blog posts into YouTube videos, Instagram carousels, and lead magnets. This multi-channel approach not only leverages content, but it also helps keep messaging consistent. ### Don’t Forget About Email Although blogs and paid search are crucial content channels, email is an underrated path toward more conversions. “It’s where your most engaged audience lives, and it can do a lot of heavy lifting when it comes to follow-ups, offers, and building trust over time,” Coussons says. Blog posts with many views can be leveraged into newsletter content, ideally leading to a growing email list to increase conversions. ## Measurement and Analysis of Conversions ### Google Analytics 4 César and Coussons recommend using Google Analytics 4 (GA4), the most recent version of Google Analytics. GA4 allows companies to see where their traffic is coming from and which channels perform better than others. It also allows for more detailed tracking by measuring individual user events rather than group sessions. Both versions, GA4 and the older Universal Analytics, give big-picture insight into how traffic aligns with conversions. ### Hotjar Hotjar adds to the insight of Google Analytics by showing businesses what their audience is ignoring, or where they’re dropping off or getting stuck, through heatmaps and session recordings. “That insight is gold when you’re trying to figure out why someone didn’t convert,” Coussons says. Hotjar offers tools that can be helpful for A/B testing, as well. ### HubSpot Once you’ve analyzed where your traffic is coming from and how your audience behaves after they get to your website, tools like HubSpot help companies track leads through the sales funnel and turn them into conversions. The Customer Relationship Management (CRM) system offers a centralized platform to keep track of traffic and customer information, such as emails, while managing blog and newsletter content. Like Hotjar, A/B testing resources are also available. ### Realize Realize stands out as a powerful performance marketing platform because it offers comprehensive tools to create, track, and optimize conversions effectively. Through its integration with the Taboola Pixel, advertisers can define and monitor a wide range of conversion events, from simple page views to dynamic purchases with varying values. The platform allows for precise control over conversion windows and even enables tracking of offline conversions via CRM integration. Crucially, Realize leverages advanced AI and unique first-party data to optimize campaigns in real-time, intelligently identifying high-intent audiences and adjusting bids to maximize return on ad spend (ROAS) and other key performance metrics, ensuring advertisers can truly realize their business objectives. ## Key Takeaways A conversion occurs when a person takes the desired action of marketing efforts, which is often to buy a product or service. For instance, in e-commerce, conversion rate is calculated by dividing the number of online sales by the total number of website visitors and multiplying that number by 100 to get a percentage. Optimizing for an efficient user interface and creating engaging content that compels consumers to take action are effective ways to boost conversion rates. ## Frequently Asked Questions (FAQs) ### Micro vs. macro-conversions: What’s the difference? While macro-conversions refer to the end goal of marketing efforts — i.e., turning a lead into a customer — micro-conversions represent benchmarks for achieving this. Some micro-conversions include how long someone spends on a page, if they watch a video, how far they scroll, signing up for a newsletter, or adding a product to their cart without purchasing. Micro-conversions suggest potential, whereas macro-conversions indicate action. ### What is the impact of trust and credibility on marketing conversions? Building trust and credibility with your audience is important when converting new customers because it removes consumer doubts before they surface. Coussons says that incorporating “trust signals” such as online reviews can go a long way towards instilling confidence. Whether it’s a well-researched blog post, an informative video, or an ad for a helpful product, people are likelier to act on content they believe in and share it with others. ### What is the best practice for optimizing paid advertisement campaigns for conversion rates? Paid ads are a great way to drive traffic, but optimizing these campaigns for higher conversion rates mostly comes down to what consumers see after they click. That is why it’s essential to test headlines, images, and other content, and keep messaging consistent throughout. That way, even when using paid ads to target specific audiences and pain points, the landing pages deliver on the promise to make life easier, without confusing them along the way. ### What are alternatives to social media to improve conversion rates? Social media can help boost micro-conversions, like getting potential customers to watch a video or read a blog post, but it’s not everything. To increase macro-conversions, SEO-driven content for organic and paid search, email marketing, and affiliate or referral partnerships are all great alternatives to social media. --- ### Contact Forms: Definition, Types, How to Create and Automate URL: https://www.taboola.com/marketing-hub/contact-form/ Last Modified: 2025-11-13 10:43:12 Contact forms may seem like small elements of a website, but they’re a highly valuable aspect of your marketing, customer retention, and business reputation. A carefully designed contact form is an essential tool that your business can use to collect feedback, generate leads, provide customer support, and fulfill countless other purposes. ## What Is a Contact Form? A contact form is a form that can be embedded on your business website for communication purposes. Customers can enter information into the form and submit that information to your business without using an email address. Contact forms are usually brief and easy to use, and once a customer has submitted their information, your business will have a way of following up with the customer, such as their email address or phone number. ## Why Is a Contact Form Used? Contact forms allow your business to gather information without publishing your email on your website. This is advantageous, since including your email on your site can lead to spam and phishing attacks: According to Hoxhunt, in 2024, 64% of businesses faced business email compromise attacks, and financial losses from the attacks averaged $150,000. Since 2021 and the implementation of malicious AI, phishing attacks have increased by 49%, highlighting the importance of using contact forms instead of sharing personal contact information on your business website. Using a contact form instead of an email address also makes it easier for your business team to delegate customer service and support. If you have a multi-purpose team, you can set your form plugin to send form responses to certain team members based on the reason the user contacted your business. For example, technical support requests could be routed to your technical support team lead, while inquiries on your services from prospects could be routed to your sales team. This ability can help your business deliver a faster and more accurate response. Contact forms can be used for communication with current customers as well as with leads. The form can allow customers to easily contact your business if they need support, want to provide feedback, or have questions about your services. Similarly, leads can ask questions, allowing you to build a relationship with them which could potentially convert them into customers. ## How a Contact Form Works You can create custom contact forms to include the fields that you need, and you can embed the form in your website. The user will enter data into each required form field and submit that data to your business. Your website form plugin will forward that message to an email address that you’ve assigned during the form setup process. You’ll receive an email message containing the data, and you can then respond to the customer or lead. When the customer or lead submits the form, they’ll receive an automated email confirmation telling them their message has been received. You can customize the text of this email: For example, it’s a good idea to tell them how you’ll reach out to them, including an approximate timeframe in which they can expect to receive a response. You can also include other helpful information, like a link to your FAQs page, a phone number for live customer support, or a call to action (CTA), like a suggestion to follow your business on social media. ## Types of Contact Forms Businesses use many types of contact forms. Depending on your business structure and goals, one or more of these types of forms might be right for you: ### General Inquiry Contact Form A general inquiry contact form serves as a catch-all form for your website. You can use the form to allow site visitors to contact your business with questions. You can then use those form responses to sort the inquiries so the correct departments or teammates respond to each form submission. ### Customer Support Contact Form A customer support contact form serves specifically to collect customer questions and feedback. You can add questions that prompt customers to provide details about the issues they’re experiencing or their feedback on your products or services, so your customer service team has the information needed to respond to each inquiry appropriately. ### Sales Inquiry Contact Form A sales inquiry contact form can help capture lead contact information so you can add prospects into your sales funnel. By allowing leads to specify the product or service they’re interested in, and to add any questions they have to the form, your sales team can follow up with a personal call and the details the lead needs. ### Feedback Collection Contact Form You can also use contact forms to collect customer feedback. Publishing a customer feedback form demonstrates that your business is invested in the customer experience. The feedback you receive can help you identify ways to improve your business and also gives you a chance to reach out to customers to correct any issues they’ve experienced. ## What Makes a Good Contact Form? Just like every other element of your marketing campaign and business materials, your contact form needs to be carefully thought-out and designed. A poor-quality contact form can frustrate website visitors and may cause them to abandon the form altogether, so you could lose out on valuable conversations and feedback. ### User-Friendly Design Good contact forms are user-friendly. They should be easy for website visitors to find and access, and they should have a clear, minimalistic layout. The instructions need to be specific, and the fields should be laid out logically, usually starting with the visitor’s contact information, including their name and email address. Make sure that the fields are all clearly labeled to avoid any confusion for visitors and to ensure you’re collecting accurate data. The form should be accessible on different devices, including mobile. ### Easy to Understand In addition to a user-friendly design, a form needs to be easy for website visitors to understand. It should feature a clear call to action, such as a message encouraging visitors to submit any questions they have. Once the form is submitted, viewers should see a message confirming the submission was received and outlining what will happen next. They should also receive an email with this information. ### Quick and Easy to Complete Designing a great contact form means striking a balance between keeping the form short and easy to complete, while also collecting the information your business needs to provide an appropriate response. Try to include only absolutely necessary fields, keeping the form as short as possible. ### Functionality Most importantly, contact forms need to work well and reliably. Test the forms repeatedly to make sure they’re easy for users to complete and submit. Make sure that any error messages that appear clearly identify any fields that need to be completed, so users can easily correct the errors. Spend plenty of time verifying not only that the form works, but that you can access the submitted information and that you have processes in place to promptly respond to form submissions. ## How to Create a Contact Form Creating a contact form is easy, and you can use one of the many form builders available, like Jotform or Google Forms. If your website is built on WordPress, there are many plugins that simplify the process of creating a form. You can also code a form into your site. No matter which method you choose, make sure that your form is embedded within your site, so visitors don’t have to leave your site to complete the form. Once you’ve determined how you’re going to build the form, you’ll need to plan out the form’s design: ### Determine the Form’s Purpose and Fields Start by identifying the form’s purpose, whether you’re creating a general inquiry form that could serve several different purposes, or a form specifically for a purpose like customer service. Think about the ways that visitors will likely be using the form and the type of information they’ll need to provide. Next, determine the fields you need to include. A field to collect the visitor’s name and email address is essential. You may want to add a field for a phone number, too. You can use multiple-choice fields to let website visitors identify the purpose of their form submission, and a text box where visitors can write comments can help you gather any other important information. ### Include a Call to Action Include a clear call to action that helps website visitors understand why they should complete the form and what will happen once they do. For example, a call to action might be, “We always want to hear from our customers. Please share your recent experience with our business and let us know what we can do better.” ### Craft Form Submission Messaging Craft a confirmation and thank you message that visitors will see once the form is submitted. Set up an auto-response so submitters also receive the information via the email they provided. The message should include clear details about what will happen next and when website visitors should expect to hear from your business. ### Set a Form Submission Email Address Before your form is complete, you’ll need to set up an email address, or multiple email addresses, to receive the form submissions. Make sure you choose emails that you monitor daily so you can promptly respond to submissions. ### Develop a Process for Responding to Submissions Develop a process to respond to submissions before your contact form goes live. If multiple team members will be responding to submissions, determine how the submissions will be divided among them. Set a goal for your response time and make sure that you’ve prepared any resources needed, such as information on return policies or new products. ## Contact Form Conversion Optimization The higher your contact form conversion rates, the more submissions you’ll receive and the more opportunities you have to engage with potential customers or to retain current customers. In addition to the best practices highlighted above, like keeping your forms brief and making sure they’re easy to complete, there are several other ways to optimize your contact form conversion so you get maximum value from your forms. ### Make Phone Numbers Optional Many people are hesitant to give out their phone numbers, so consider making this field optional or not including it at all. Starting with an email address can allow you to make contact with a customer, and you could offer to contact them by phone once you’ve built a relationship, or if they need phone support. ### Offer an Incentive Consider offering visitors an incentive to complete a form, such as a discount or a free consultation. This extra incentive can help motivate individuals to complete a form, and it can be a particularly valuable technique when you’re looking to add leads to your pipeline. ### Include Real-Time Error Validation If you’ve ever completed a form, only to receive a message that there was an error that needs to be corrected, you know how frustrating delayed error validation can be. Some forms even delete the information, leaving individuals to complete the form again from scratch. To avoid this issue, implement real-time error validation on your forms, which will highlight fields that are incorrectly or impartially completed in real time. Users can quickly identify and correct those errors, increasing the chances of a form conversion. ### Experiment With Form Placement Test out different form placements on your website, as you may find that certain placements drive more conversions. For example, putting a form above the fold means users can access it without having to scroll to the bottom of your site. Monitor form performance in different placements and see if you can further optimize your conversions. ### Try Different Calls to Action Your call to action encourages visitors to complete a form, so test out how different CTAs perform. Try including mention of your incentive, experiment with different CTA language, and see which CTAs your audience best responds to. ## How to Ensure Security & Prevent Spam From Contact Forms Contact forms are a frequent target for spam and bot submissions. These spam emails can drain your resources and make it difficult to sort through and find legitimate form submissions. They can also be used to send you malware and can jeopardize your site’s security. There are several ways to ensure your form’s security and prevent spam: ### Use a CAPTCHA To minimize spam, consider adding a CAPTCHA to your form. The CAPTCHA test, such as clicking on photos of a certain type of object, is intended to be easy for humans to complete but difficult for bots to navigate. Some CAPTCHAs can be difficult and frustrating to complete, so consider the type that you choose to use. If you’re using a form plugin, you may be able to design your own custom CAPTCHA, such as asking submitters to perform a simple math problem. ### Add Questions to Your Form You can also add questions to your form that serve to sort human submissions from bots. By requiring that submitters answer a simple question, like which of two letters comes first in the alphabet, you can cut down on spam submissions without making forms overly difficult for humans to complete. ### Implement Double Email Opt-In A double email opt-in can ensure that the form submitter is using an authentic email address. Once they submit the form, they’ll receive an email with a link they must click to verify their email address. This type of spam prevention requires the submitter to complete an extra step, so it’s often best reserved for instances where the submitter stands to receive value from the form, such as if they’re signing up for an incentive like a discount or a free guide. ### Publish a Privacy Policy If your contact form collects personally identifiable information, like customer names and email addresses, you’ll need to publish a privacy policy identifying what type of information you collect, how you use that information, and how the information is shared. There are several laws protecting the privacy of individuals in states and countries. For example, the California Online Privacy Protection Act went into effect in 2004 and requires commercial websites and online services to provide specific privacy disclosures to users. ## How to Automate Contact Forms By automating your contact forms, you can streamline the process of responding to them. Automation saves your team time and means that your website visitors will promptly receive responses. ### Use an Automation Platform You can use automation platforms like Zapier or HubSpot to trigger a series of actions when a form submission is received. ### Set Up Your Automation Trigger You will need to identify a trigger that will begin the automated processes. When working with contact forms, that trigger is usually the submission of a form. ### Identify Your Automated Actions You can automate contact forms in many ways. For example, you might want contact form submissions to be automatically added to your CRM as leads, or for the data to be added to spreadsheets. You can also choose to have team members notified of new form submissions, or you might want to send a series of automated emails to individuals who have submitted forms. ## Key Takeaways Contact forms are valuable additions to your website and to your overall marketing strategy. By carefully designing contact forms to be user-friendly, you can increase your form conversions and use those forms to gather feedback, collect new leads, build relationships with your customers, provide customer support, and more. Adding forms to your website can boost your marketing efforts and customer retention while also giving website visitors an easy way to get in touch with your business. ## Frequently Asked Questions (FAQs) ### What fields should I include in a contact form? It’s best to keep contact forms as short as possible while ensuring they collect the information you need. At a minimum, you’ll need to collect the submitter’s name, email address, and the reason why they’re contacting your business. Depending on the form’s purpose, you may need to add more fields. ### How long should my contact form be? The shorter your contact form, the better, but it’s important that the form collects the necessary information your business needs, too. When designing your contact form, try to keep the fields minimal so it’s quick and easy for users to complete. ### Is it safe to collect data through a contact form? It is safe to collect data through a contact form, but you’ll need to make sure that your website publishes a privacy policy disclosing what personally identifiable information you’re collecting, and how you’re using it. --- ### Ad Units: A Complete Guide for Digital Marketers URL: https://www.taboola.com/marketing-hub/ad-unit/ Last Modified: 2026-06-22 08:39:09 If you’re a marketer wanting to maximize ad performance and ROI, you need to understand how ad units work. Ad units are foundational to online advertising, as they determine where and how your ads will be displayed across websites, apps, and video streaming platforms. This guide explains what an ad unit is and breaks down the different ad types, formats, sizes, and placements. I’ll also show you how to optimize your ad units for better engagement and performance. ## What Is an Ad Unit? An ad unit is a specific space on a web page or app that displays various types of ads, such as banner, native, video, and interstitial ads. It often contains code that tells the ad network the size, format, and placement details. ## Four Common Types of Ad Units Ad units come in a variety of types, and understanding their differences can help you choose the right ad format for your campaign. Here are the most common ad unit types: ### 1. Display Ad Units Display ad units can appear on websites as static images or rich media-based ads and are usually located in a banner or sidebar format or within the web page content. Many display ad units use standardized sizes, like 250 X 250, or 728 X 90. ### 2. Video Ad Units Regular video ads, also known as pre-roll, mid-roll, or post-roll, play before, during, or after video content on a wide range of platforms. They can be very effective for storytelling, as a well-crafted video can evoke emotion from the viewer. However, they can also be disruptive if they delay the content that the viewer is trying to watch. ### 3. Native Ad Units Native ads units are designed to blend into the surrounding content. Marketers can do this in various ways: For example, native ad units can be placed within a news or social media feed, on a search results page (appearing like an organic listing), or as a content recommendation widget at the end of an article. ### 4. Interstitial Ad Units Interstitial ads are full-screen ads that appear during natural breaks in the content, like between articles or sections of an app. They’re very good at grabbing a viewer's attention, but can become overwhelming and disruptive if not used selectively. ## Ad Unit Formats Ad unit types and formats are closely related, but different. An ad unit type refers to the category or placement of an ad unit, e.g., display, video, native, interstitial. An ad unit format describes the specific structure of the ad content. Here are some common ad format examples: ### Static Image Static image ad units are simple, and typically use formats like JPG, PNG, or GIF. They can be easily produced and are quick to load. ### Rich Media Rich media ads offer more advanced features, such as video, audio, or other elements that make it easier for viewers to engage with the content. While rich media ads can boost engagement, they usually take longer to load than static image ads. ### HTML5 HTML5 is the latest version of the HyperText Markup Language (HTML), the standard language used to create and display web content. Because HTML5 is supported by all modern web browsers and is compatible across all device types, the format is very effective for dynamic, interactive, and mobile-friendly advertising. It’s responsive, too, so you can deliver the same ad to users on a laptop, phone, or tablet. HTML5 ads are often lighter than rich media ads, so they can load faster, which improves the user experience. ### Display Ads Display ads constitute broad category that includes many common ad formats like banner ads (rectangular images), skyscraper ads (tall, narrow ads), and medium rectangles. These are classic examples of ad units, defined by their size and placement on a page. ### Carousel Ads While carousel ads are a more dynamic and interactive format of advertising (allowing multiple images/videos to be swiped through), they are still implemented within a specific, predefined ad space or container on a platform. Each "card" within a carousel ad can have its own headline, description, and call to action, all contained within that single ad unit. ### Native Content Native ads can be considered a type of ad unit as well as an ad format. The latter refers to how native ads look and interact. Essentially, they’re designed to blend seamlessly with the surrounding content by matching its look and feel. Because they are less intrusive than traditional online ads, like video, banner, or pop-up ads, native ads often generate higher engagement and better value for marketers. ## Ad Unit Sizes Digital advertising ad unit sizes have been established by the Interactive Advertising Bureau (IAB), which is an industry organization that works with advertisers, publishers, and technology providers to develop technical standards and guidelines for digital advertising. The table below outlines just a few standard fixed ad unit names and dimensions, along with their typical placements: Ad Unit Name Dimensions (px) Common Placement  Billboard 970 X 250 Top of premium pages Half Page 300 X 600 Sidebars or interstitial areas Large Rectangle 336 X 280 Within content or sidebars Medium Rectangle 300 X 250 Within content or sidebars Leaderboard 728 X 90 Top or bottom of web pages Skyscraper 120 X 600 Sidebars While these classic fixed ad unit sizes are still widely used, the industry has moved toward flexible and responsive ad sizing as part of the IAB New Ad Portfolio. These ads are able to adapt their size and specifications to the user’s screen and device. ## Ad Unit Placements Ad placement refers to where an ad appears within a website, app, or video stream. By selecting the best possible placement, you can maximize visibility and user engagement. Below are some common ad placement options: ### Above the Fold Above the fold refers to the portion of a web page or app screen that is visible without scrolling. By placing ad units above the fold, they will typically receive higher impressions and engagement. ### In-Content In-content placement refers to placing ads within articles or other types of content, such as videos. Doing so can lead to higher click-through rates. It also provides marketers with an opportunity to provide contextual relevance. ### Sidebar Sidebar placements are located on the left or right side of a web page, beside the main content. Many desktop environments use sidebar layouts, as there is more horizontal space to work with. When the web page is displayed on a mobile device, the sidebar content usually moves below the main content, or is hidden entirely. ### Footer Ad units in a footer placement are located at the very bottom of a web page or app screen. They usually appear below all of the main content on the page. ## How to Optimize Ad Units Here are some common strategies marketers can use to optimize their ad units and improve ad performance: ### Test Different Ad Formats and Dimensions Consider A/B testing different ad formats to determine which ads resonate best with your audience — never assume that one format works best for all campaigns. By choosing an ad platform that offers AI-powered testing, you can continuously A/B test all your ads, optimizing both targeting and creative in real time. ### Prioritize Viewability Some ad unit types, such as in-content placements and sticky sidebars, tend to remain in view longer. By focusing on these ad units, you may be able to improve performance and boost ROI. ### Use Contextual Targeting Try to run ad units with relevant content in order to improve user engagement. This is where native ad units shine, as they blend into the environment. ### Optimize Ads Units for Mobile According to Google, mobile-friendly websites are prioritized in search results. Also, many advertisers receive the majority of their traffic from people using mobile devices. As such, it’s critical to ensure your ad units are responsive and perform well on mobile. For example, the 320 X 50 ad unit is designed for small screens, as are many native units. ## Key Takeaways Ad units are critical components of digital advertising. They define where and how ads are displayed across websites, apps, and streaming platforms, and come in various types and sizes, based on IAB standards. Remember that placement plays a key role in the performance of ad units. You can also optimize ad units by A/B testing, focusing on contextual relevance, and ensuring they are responsive: Over time, and with the right strategies, you can generate more impressions, boost viewer engagement, and improve your campaign performance. ## Frequently Asked Questions (FAQs) ### What are some common ad unit examples? Popular ad units include the 120 X 600 Skyscraper, often displayed in sidebars, a native video appearing in a social media feed, or a 15-second video ad that plays before a YouTube video. ### How do you create an ad unit? The process for creating an ad unit is fairly straightforward, but will differ slightly based on the ad platform you’re using (e.g., Realize or Google Ad Manager). You will typically start by setting your ad unit size, format, and placement. Once that’s done, you can generate the ad unit code and place the code on your website or app. Note that you don’t always have to manually embed ad code: Many modern ad platforms support dynamic or automated placement methods. ### What is an ad unit code? An ad unit code is an HTML or JavaScript snippet that tells the ad server where to serve an ad on a web page or app, and how to render it. The code contains information such as the ad size, placement ID, and targeting parameters. When the web page loads, the ad unit code communicates with the ad server or exchange to retrieve and display a relevant ad. --- ### Cost-Per-Mille (CPM): Your Go-to Guide for Smart Advertising URL: https://www.taboola.com/marketing-hub/cost-per-mille/ Last Modified: 2025-05-26 07:49:38 If you're familiar with the online advertising scene, you’ve likely heard of cost-per-mille, or CPM. It might sound a bit fancy, but it's a highly useful metric to help you determine how much you're spending to get your ads in front of people. Let’s break what cost-per-mille actually means. ## What Is Cost-Per-Mille (CPM)? Cost-per-mille, or CPM, represents the price of reaching one thousand impressions of an advertisement. It’s an essential metric for advertisers who want to analyze cost-efficiency in their campaigns. "Mille" means "thousand" in Latin, so, in simple terms, it measures the cost of getting your ad in front of a thousand viewers. Understanding this can help you make the most of your advertising budget. ## How Do You Calculate CPM? Calculating CPM is pretty simple: To determine your cost-per-mille, begin by dividing your total campaign expenditure by the total number of impressions. This calculation will give you the cost for a single impression. Then, multiply that figure by 1,000 to find the cost for every 1,000 impressions. For example, if you spent $2,000 on an ad and got 500,000 impressions, your CPM would be $4. ## Limitations of CPM CPM provides businesses with some key benefits, such as aiding in effective budgeting and cost comparison across different advertising platforms. It’s also cost-effective, offering some insights into how efficiently ad dollars are spent, making budgeting straightforward by allowing advertisers to determine how many impressions they can afford. However, CPM has its limitations: It lacks direct action metrics, meaning it doesn’t measure clicks or purchases, which are vital for assessing true success. Unlike with performance advertising, where you only pay for direct results, relying on CPM could see you wasting money, since it’s possible for your ads to attract numerous impressions without actually engaging viewers. Considering if and when your Google and Facebook Ads translate into diminished returns, and recognizing these limitations is essential for the effective use of CPM in advertising strategies. ## CPM vs. Other Pricing Models ### CPM vs. CPC Cost-per-click (CPC) is all about actions, such as clicks, whereas CPM focuses on getting your ad seen by a large audience. For example, in the digital advertising model of CPC, if a business uses Google Ads with a CPC bid of $2, they pay $2 for every click on their ad. This model is effective for driving traffic and generating leads. In contrast, CPM focuses on visibility, with advertisers paying for their ad to be shown a thousand times, regardless of clicks. A CPM rate of $10 means the advertiser pays $10 for every 1,000 impressions, regardless of any click-through actions. This approach is more beneficial for brand awareness campaigns, which can lead to long-term revenue growth, while CPC actively focuses on finding click-worthy customers. ### CPM vs. CPA Cost-per-acquisition (CPA) measures the cost to acquire a customer taking action, such as making a purchase. For instance, spending $500 to gain 50 customers results in a CPA of $10. Conversely, CPM indicates what advertisers pay for 1,000 ad displays, regardless of actions. Understanding both your CPA and CPM helps businesses effectively align their advertising strategies with specific goals. ## Budgeting and Planning CPM When developing your CPM strategy, it's essential to consider tactics such as market analysis, audience segmentation, and competitive positioning, all of which play an important role in maximizing the effectiveness of your advertising efforts. ### Based on the Budget Begin by determining your overall advertising budget, which serves as the foundation for your marketing efforts. Next, use the CPM metric to calculate the potential number of impressions you can acquire. This approach enables you to effectively measure the reach of your campaign and optimize your spending to maximize visibility among your target audience, allowing you to make the most of your budget. ### Based on Goals Aligning your CPM strategy with your overarching advertising goals is essential. Whether you're aiming to build broad brand awareness or drive deeper engagement, your CPM strategy should reflect these objectives. ### Based on Advertising Channel Different advertising channels come with their own unique costs and target audiences, so it’s essential to think carefully about the platforms you choose for your CPM strategy. Analyzing where your ideal customers are most active can help you allocate your resources effectively, ensuring that your advertisements reach the right people at the right time. ## How to Optimize CPM You can optimize your CPM by employing several tactics, such as ad targeting, creating engaging content, and benchmarking your CPM. The following steps can help lower your CPM and enhance the effectiveness of your paid ads: ### Ad Targeting Ensure you’re targeting the right audience with your ads. The more relevant your ad is to them, the better your CPM results will be. When your message resonates with the right audience, it typically yields better results and greater value for your investment. ### Creative Content Captivating and well-crafted ad content has the power to spark curiosity and evoke strong engagement from your target audience. By weaving together vivid visuals and compelling narratives, you not only capture attention, you also encourage meaningful interactions. This heightened level of engagement translates to a more effective use of your advertising budget, allowing you to achieve greater returns on your investment while building a stronger connection with potential customers. ### Benchmarking CPM Understanding where your CPM stands in relation to these standards is important, as it not only indicates whether you're operating within the expected range, but also highlights areas for potential improvement. Monitor your CPM and compare it to the prevailing industry benchmarks. Delve into performance metrics and analytics surrounding your CPM to gain insights into your advertising’s effectiveness and optimize your strategy accordingly. ## Performance and Analytics Around CPM ### Benchmarking Evaluate and analyze your CPM in relation to your other advertising campaigns, as well as the prevailing market rates within your specific niche. Consider the key factors that help determine whether your CPM is excessively high and how it compares to industry standards. ### CPM vs. Other Metrics Combining CPM with other metrics, such as click-through rate (CTR), offers a richer understanding of ad performance. This approach reveals not just how often ads are viewed but also how effectively they engage the audience, providing valuable insights into the overall reach and impact of advertising efforts. ### Understanding How CPM Impacts Your Overall ROI Think about how your CPM fits into your overall return on investment — it’s all about making sure your spending aligns with your goals and that you're getting the most value for your advertising dollars. ## What Alternatives Do Marketers Have Other Than the CPM Model? There are various alternatives and innovations that have distinct advantages over traditional CPM tactics. Cost-per-Click (CPC) is a solid alternative if you want to focus on user actions rather than just impressions. Cost-per-Action (CPA) is another option that looks at how much you spend to get specific conversions or actions. Innovations in performance marketing, such as AI-driven tools and holistic performance platforms, allow for measurable results with precise tracking of conversions and real-time data. This approach targets specific audiences, increasing conversion likelihood, and incorporates continuous optimization for better returns. Performance marketing also offers agility, allowing for rapid testing and adaptation to market changes. ## Key Takeaways CPM helps you gauge the cost of visibility for your ads. It’s great for budgeting, but may not capture all user actions. As a small or medium-sized business, consider exploring various pricing models to determine which one works best for you. ## Frequently Asked Questions (FAQs) ### What is the difference between CPC vs. CPM? CPC is all about action (it stands for cost-per-click), while CPM focuses on visibility. ### What is the difference between eCPM vs. CPM? The difference between Effective CPM (eCPM) and CPM is primarily in their focus. CPM measures the cost advertisers pay for every thousand impressions, while eCPM provides a broader view of earnings across various ad types, including clicks and actions. Essentially, CPM focuses on ad costs, while eCPM offers insight into overall revenue performance from different ad formats. ### Is a $20 CPM good? Ultimately, the effectiveness of a $20 CPM largely hinges on conversion rates and the specific context in which it’s applied, and a lot depends on your industry, product, and campaign goals. In e-commerce, a $20 CPM can be justified if it leads to significant sales, especially when each sale generates a $200 profit. Similarly, in the automotive sector, this CPM is reasonable due to the high commission potential from car sales, which can bring substantial payouts. For non-profits, on the other hand, a $20 CPM may be excessive if it doesn't result in enough donations, as low contributions can outweigh costs. In the technology and SaaS industries, this CPM can be acceptable if it leads to long-term subscriptions, given the potential lifetime value of customers. ### What’s a typical CPM on LinkedIn ads? When considering advertising on LinkedIn, it's important to anticipate higher CPMs, typically falling within the range of $25 to $50. This elevated pricing is largely attributed to the platform's unique professional audience, which consists of decision-makers, industry leaders, and skilled professionals, making it a prime space for targeting a discerning market. ### Can I optimize CPM for TikTok ads? To effectively engage your audience on TikTok, tailor your content to fit the platform's unique style and trends. By understanding what resonates with your target audience, you can create captivating videos that lower your CPM and optimize your advertising investment. ### Should I focus on lowering CPM or increasing CTR? A balanced approach is often the most effective. Achieving a lower CPM alongside a high CTR indicates that your campaigns are performing exceptionally well. When both metrics align favorably, it demonstrates not only cost efficiency, but also strong audience engagement. ### What factors influence CPM in display ads? Audience targeting, ad placements, and competition are crucial factors that influence your CPM. By clearly defining your target audience, you can craft impactful messages that resonate effectively. Strategic ad placements boost visibility and interaction, while the competitive landscape can drive CPM variations as advertisers compete for the same audience. These elements work together to shape your advertising costs and effectiveness. ### What’s a good CPM for B2B vs. B2C? B2B generally sees higher CPM due to niche targeting, while B2C could have lower costs. ### When is CPM better than CPC? CPM works best for campaigns purely focused on raising brand awareness rather than immediate actions. --- ### Full-Funnel Marketing: From Fantasy to Real World Results URL: https://www.taboola.com/marketing-hub/full-funnel-marketing/ Last Modified: 2025-07-22 13:22:58 Full-funnel marketing refers to the practice of targeting customers at every stage of their buying journey — from awareness to consideration to conversion — using tailored content and tactics to guide them toward a purchase. The concept of the marketing funnel has been around since long before the internet, originally being developed by E. St. Elmo Lewis in 1898. Addressing the different parts of the funnel with bespoke tools and tactics has effectively served generations of businesses advertising and selling to consumers. These days, though, this straightforward concept of a user journey has stopped applying to many buying cycles. There’s a lot more complexity in marketing now, so a linear, staggered funnel doesn’t necessarily ring true for sophisticated digital marketers. User journeys today vary widely depending on the person, the product, the channels they use, the transaction type, and more: Sure, it could look like a funnel, but it could equally look like a web of interactions on the way to a purchase. Actually being effective across the entire range of customers’ actions has become much harder in this busy digital landscape, where users are constantly bombarded with messages and businesses have so many channel and tactic options to choose from. I’ve talked with advertisers across industries and verticals about where they feel full-funnel marketing is still relevant, and where it’s struggling to continue being effective. I’ve seen firsthand how appealing the promise of full-funnel marketing is, but I’ve also seen how rarely it actually delivers without the right structure, partners, and expectations to support it. In this post, I’ll dig into what makes full-funnel marketing so challenging these days, and how savvy marketers can approach the buyer’s journey more effectively. ## What Is Full-Funnel Marketing? At its core, full-funnel marketing is a framework that recognizes that consumers don’t make purchase decisions instantly. Instead, they move through stages, first becoming aware of a need or a product, considering various options, and eventually making a decision to buy. Full-funnel marketing aims to tailor content, media tactics, and messaging to all of these distinct phases, guiding users gradually down the funnel toward a conversion. What this means in practice is that marketers use different creative strategies, tactics, and platform tools for each stage of the funnel, with different KPIs for each: - Upper Funnel (Awareness): Reaches broad audiences with brand storytelling, video, and display ads to build recognition and interest. - Mid-Funnel (Consideration): Uses more targeted messaging, retargeting, and educational content to inform and nurture prospects. - Lower Funnel (Conversion): Uses performance-focused tactics like search ads, product retargeting, or direct response campaigns to drive action. ## Why Full-Funnel Marketing Is So Hard to Get Right The funnel may have been simple and straightforward in the days before online marketing, but for advertisers and marketing teams in the digital era, the funnel has become ever more fragmented. The new reality is that most advertisers struggle to execute full-funnel marketing successfully, especially in digital environments where consumers are continually moving across devices and channels. There are a few key reasons why: ### Fragmented Platforms and Data Upper-funnel campaigns might live in a brand-focused DSP, mid-funnel engagement might happen via social media or email platforms, and lower-funnel action might take place in ecommerce or search tools. Stitching these efforts together across platforms is technically and operationally complex. ### Creative and Measurement Misalignment Each stage of the funnel requires a different type of creative and a different KPI, yet many advertisers use a one-size-fits-all approach to creative, or they measure all campaigns with the same success metric (like ROAS), regardless of where they fall in the funnel. ### Budget Allocation Challenges It’s often difficult to justify upper- and mid-funnel investments because they don’t drive immediate conversions. This leads to over-indexing on lower funnel tactics, which can deliver short-term gains at the expense of long-term growth. ### Vendor Over-Promise Many platforms and partners promise full-funnel solutions, but fall short on delivery. Generalist platforms keep businesses from truly addressing the entire funnel appropriately. Which brings me to the most important part of this topic… ## Generalist, Full-Funnel Platforms Can’t Get the Job Done Anymore In recent years, demand-side platforms (DSPs) and some ad tech vendors have positioned themselves as full-funnel solutions. Indeed, many of them offer features for campaign deployment right across the awareness, consideration, and conversion stages. But the problem is that, due to the complex nature of full-funnel marketing, addressing all stages of the funnel is in no way the same as being excellent at each of them. To be truly effective at any single stage of the funnel requires: - Deep expertise in the media channels that work best at that stage. - Purpose-built technology to optimize for the relevant KPIs. - Creative tools to match the mindset and intent of the audience. Achieving this across all funnel stages demands significant investment, specialization, and focus — something few generalist vendors can realistically offer. What you end up with instead are platforms that do a passable job at everything, but don’t drive exceptional results anywhere. What’s really problematic is that these platforms lack the technology to actually move a consumer from one stage of the funnel to the next. They may serve ads at multiple stages, but can’t orchestrate a sequence that actively progresses a user from passive awareness, to engaged consideration, to being ready to buy. ## Why Best-Of-Breed Can Be a More Suitable Approach As I’ve talked with advertisers and explored the various technology options available, I’ve come to realize that businesses trying to address the whole funnel would be better served by assembling a best-of-breed stack — that is, a curated set of specialized partners, each optimized for a specific stage of the funnel, united by a shared understanding of how to pass the baton downstream. As an example, a company advertising throughout the entire funnel might assemble a stack that looks like this: - A brand awareness vendor specializing in CTV or premium video, with advanced reach and frequency tools. - A mid-funnel vendor focusing on content marketing or engagement of target audiences, prospecting based on purchase intent signals, and retargeting capabilities. - A conversion vendor offering granular performance optimization, real-time bidding for e-commerce, or advanced attribution modeling. No less important to the specialization in each stage or type of interaction, is the ability to connect between them, and actually drive potential customers towards a sale. Even within some of the larger platforms, these stages remain siloed and disconnected. When savvy marketers are able to make these best in class vendors and platforms work together, the business achieves: - Superior performance at each stage, designed by experts who live and breathe that segment. - More agile measurement and optimization strategies that respect the nuance of each funnel stage. - Greater transparency and control over spend, targeting, and outcomes. Naturally, this approach requires more coordination and strategy, but for brands committed to long-term growth, the payoff is well worth it. In Taboola’s experience with thousands of clients, the advertisers who take the time to map their funnel, define stage-specific goals, and match them with the right tools, consistently outperform those who lean on a single, full-service vendor. ## Tips for a Successful Full-Funnel Experience If you’re an advertiser or agency leader trying to make full-funnel marketing work, here are a few guiding principles: ### Start With the Customer Journey Map out how your target customers move from awareness to action. Which channels do they use? What content do they engage with? What motivates them to convert? ### Set Clear, Stage-Specific KPIs Don’t measure awareness campaigns by conversions: Be sure to use appropriate metrics like brand lift, viewability, or engagement rate. Then, ladder these into mid- and lower-funnel KPIs like CTR, CPA, and ROAS. ### Pick Partners With Purpose Choose vendors who specialize in a particular stage, and who have proven results in your industry. Ask how their offering integrates into the broader journey. ### Make the Connection Make sure that the platforms chosen allow you to connect the dots and actually shift people "down the funnel.” Also, your CRM, analytics tools, and ad platforms should all talk to each other: This is key to understanding user progression and making real-time optimizations. ### Test, Learn, and Iterate Full-funnel marketing will always require tweaking and optimizing as you go, so build a feedback loop where performance data informs creative, targeting, and media spend decisions. ## Key Takeaways More than a century after its invention, full-funnel marketing still definitely plays a role for advertisers, but there’s an awful lot more complexity to contend with now. It’s not just checking boxes: It takes real understanding of the way consumer behavior flows from one moment to the next — how real people move, in real time, down a complex, nonlinear path to purchase. Success in full-funnel marketing, then, depends not just on addressing each stage, but on creating momentum between them, so consider your audience, what their path looks like, and craft a journey with the right tools and KPIs to build your funnel from top to bottom. The future of marketing isn't about mastering the funnel in segments. It's about engineering meaningful movement through it, turning fragmented touches into a cohesive, compelling journey that truly converts. --- ### Retargeting: How To Hook Your Audience URL: https://www.taboola.com/marketing-hub/retargeting/ Last Modified: 2025-05-25 10:29:40 Retargeting is the delivery of ads, emails, or other media to remind a user of a product or service in which they had demonstrated an interest. Here's how it works. Imagine someone stepping into a convenience store on a hot and sunny day, grabbing a soft drink from a cooler, holding it for a moment, then changing their mind and stepping back outside. As this person sits on a bench on this muggy midsummer day, a bus happens to pass by, on the side of which is emblazoned a huge ad for the very beverage the person just passed on. The idea of the refreshment is refreshed in their mind, they march right back into the store and buy the soda. In the physical world, that’s an example of retargeting. Of course, you could also call it a stroke of luck, at least for the marketers who placed the ad and for the beverage brand. In the digital world, however, marketers don’t have to hope a thirsty person will be curbside as a bus passes by. Online, marketers can reach their intended audience much more directly. ## What Is Retargeting? Allan Hou, sales director of TSL Australia, explains it like this: “Retargeting is showing ads to people who’ve already visited your website or app but didn’t buy, sign up, or take whatever action you wanted them to. It’s like when you look at a product online, leave without purchasing, and then keep seeing ads for it later while scrolling through social media or reading articles.” Long story short, when you have a lead on the line but they don’t end up taking the proverbial bait (i.e., making a purchase online), retargeting is the sending down of another hook that meets them in a different part of the pond, but has similar bait to the one they almost went for earlier. “Let’s say someone visits your site, checks out your freight services, clicks on the customs clearance page, stays there for a few minutes but leaves without sending a quote request,” says Hou. “With retargeting, that same person could later see your ad on LinkedIn, YouTube, or while browsing the news. The ad could show a testimonial from a similar business or offer a limited time quote. The goal is to remind them of what they already saw and give them a reason to come back and finish the process.” ## How Does Retargeting Work? Retargeting works by gently nudging people back in a direction they were already considering going. “The whole point of retargeting is reminding people about your brand, specifically those who have already shown interest by visiting your website, but haven’t converted yet,” says Andrius Surdokas, digital advertising specialist at Omnisend. “The success of such campaigns mostly depends on how relevant the offer is. The thing is, all users behave differently — some may have spent only a few seconds on your site, others may have already filled their carts. Some may be ready to convert, others may need more nurturing. Treating all of them the same way won’t lead anywhere.” “That’s why you need to track user behaviors,” Surdokas continues. “Marketers typically use cookies or pixels for this. For example, the Facebook Pixel or Google Ads tag can be useful for tagging visitors for future targeting. You can then segment your audience and plan campaigns accordingly.” ## Types of Retargeting ### Site Retargeting This common approach uses what is called a pixel on your website to track visitors and retarget them with ads on other websites and platforms. Also known as a tracking pixel or web beacon, a pixel is a small, usually invisible image or piece of code that website owners embed in their website's HTML to monitor user activity and track website performance. These pixels help businesses understand user behavior, measure campaign effectiveness, and build audiences for targeted advertising. ### Social Media Retargeting This approach targets people who have engaged with your social media content online or in an app, such as by liking a post or following your brand. You can have them served curated ads that feel more like native posts (in the form of pictures, videos, carousels, and more) that appear in the course of their natural scrolling. ### Email Retargeting Commonly used after someone abandons a digital shopping cart, but also useful in other instances, email retargeting is one of the most direct approaches, and also one of the most likely ways to actually reach a user. There’s no guarantee they open the email and even less guarantee they convert, but they’re certainly likely to see it. ### Impression Retargeting There’s never a guarantee that a person saw your online ad even if it was served to them, but when an ad is served, it’s still referred to as an impression either way. This type of retargeting is a way to reach out to leads that have been served impressions, and try to ensure they come into more direct contact with the brand. ## Benefits of Retargeting ### Increased Conversions When a customer returns to a site or app on which they were already close to taking an action the marketer had hoped to trigger, the party will be that much more likely to convert, be that making a purchase, signing up for a newsletter, or downloading a brochure. ### Enhanced Brand Awareness By its nature, retargeting puts a brand back in front of someone who already had at least some touchpoint with it. Retargeting and finding the person in another channel only increases their brand awareness, making conversion more likely, even if it takes time. ### Better ROAS Customer retargeting often means a better ROAS (return on ad spend) because you’re reaching out to parties that have already demonstrated at least some interest in your products or services, so the ads can be more focused and delivered to a smaller audience, which saves you money. ## Considerations When Retargeting ### Privacy Concerns Retargeting in marketing raises privacy concerns due to the collection and use of personal data for targeted advertising. Some people don’t want their online activities tracked, even if your intention is simply to connect them with a brand they will love. You also need to make sure you don’t fall foul of GDPR or CCPA regulations. ### Ad Fatigue Everyone has a point at which ad fatigue kicks in, and once someone has simply seen enough of your brand’s ads, further exposure will have the inverse effect of what you want, annoying them and driving them away instead of to you. ### Wasted Ad Spend Done well, retargeting saves you money. Done wrong, it can be a total waste, with ads delivered to people who have either already converted or who are tired of your brand and are never going to take action. ## How Can I Set Up a Retargeting Campaign? ### 1. Choose Your Platform There are many different services you can use when you undertake ad retargeting, including Google Ads, LinkedIn Ads, AdRoll, Facebook Ads, and Realize. Each has its benefits, depending on where your audience tends to spend the most time. ### 2. Establish Your Goals Determine what you want users to do after seeing your retargeted ads. Do you want them to make a purchase, download a resource, sign up for a newsletter, or something else? Knowing this will guide your ad creative and retargeting strategies. ### 3. Create a Retargeting List Use tools like tags, pixels, cookies, and email lists to create a master list of as many potential customers as you can. ### 4. Establish Tracking Mechanisms Now is also the time to embed the tags or pixels onto your site, so you can actively track more users as they navigate around. ### 5. Launch Your Retargeting Campaign Select the people on your list you want to target and start sending them curated ads. Make sure to establish your ad spend budget and calendar before launching. ## Best Practices for Effective Retargeting ### Segmentation Retargeting segmentation involves dividing your retargeting audience into different groups based on their interactions with your website or brand. This allows you to create more personalized and effective retargeting campaigns by tailoring your messaging and offers to specific user behaviors and needs. ### Avoiding Fatigue Avoiding ad fatigue involves using strategies to prevent users from becoming bored or annoyed with the same ads, which can lead to a decline in engagement and a negative brand perception. This is achieved by balancing retargeting efforts with prospecting, diversifying ad placements and creatives, and implementing frequency caps. ### Ad Creatives Ad creatives are the advertisements you design and display to people who have previously interacted with your website or app. These ads aim to remind them of your brand, product, or service, and encourage them to re-engage and potentially convert. ### Personalization Retargeting personalization is a marketing technique where ads are displayed to users who have previously interacted with a website or brand, but are tailored to their specific behavior and interests. The goal is to reignite their interest in a particular product or service by showing them relevant ads and offers. ## Effective Retargeting Strategies in 2025 “Retargeting is marketing to people who have already interacted with you in some way,” reiterates digital marketing expert Lucas Lee-Tyson. “It’s much easier to market to someone who has already transacted with you or interacted with your business in some way, rather than a brand new customer.” ### Dynamic Product Ads Dynamic ads are display ads that update in real time based on user behavior, such as showing related products or matching accessories to items that are actively being viewed. ### AI-Powered Personalization Many marketers these days turn to AI to help the rapid generation of ads that will be highly relevant to a given population, or even to an individual user. The more data gathered about the party, the more effective these can be. ### Multi-Channel Approach Marketers who encounter a user in one realm can reapproach them in another with potentially greater success. Say you get a lead on LinkedIn, but that person actually spends more time on TikTok. Messaging there might be better received, despite it being about the same brand. ## Measuring the Success of Retargeting Campaigns There are a few KPIs (key performance indicators) to track when you’re conducting a retargeting campaign: ### Return on Ad Spend Calculating your ROAS alone can tell you so much about your ad campaign’s success. How much did you spend, and how much did you make? If the latter is a lot bigger than the former, then that was successful retargeting. ### Conversion Rate Because conversions are not always sales that can be measured in dollars, calculate your total conversion rate, including things like sign-ups, downloads, likes, follows, and more. ### Cost Per Acquisition Abbreviated as CPA, this is the amount of marketing cash that was needed to be spent to acquire a new customer through a retargeting campaign. The CPA is usually lower with retargeting than it is with an acquisition of a fully new user. ## Optimizing Retargeting for Best Performance ### Personalize the Ad Experience Don't treat all website visitors or app users the same way. Segment based on actions like page views, time on site, and if they've previously converted. Use dynamic and curated ads that speak to the individual. ### Optimize Timing To avoid ad fatigue, use frequency caps to limit the number of times a user sees your retargeting ads, and determine the ideal length of your campaign before launching it. This can require some research, but it’s worth it. ### Test and Reassess As with any marketing campaign, with retargeting, it’s essential that you execute A/B testing so you can see what’s working and what’s not (or what’s working better, at any rate). ## Common Mistakes to Avoid in Retargeting ### Ineffective CTAs The CTA, or call to action, is the prime directive of an ad. If you get an ad in front of someone who is already familiar with your brand and your CTA is weak, confusing, off-message, or otherwise poor, you have just squandered an opportunity. ### Failing to Run A/B Tests If you don’t run at least two different versions of retargeted ads, you won’t be able to collect data that lets you see what’s working best, and you won’t be able to iterate. Remember, tools that allow you to continuously A/B test in real time are a huge help. ### Overly Broad Retargeting Retargeting is where you narrow things down and hone in on your audience with laser focus, customizing ads to suit each user to the highest degree possible. Go too general with things, and people may have less sense of what you do and offer, rather than more. ## Best Platforms for Retargeting ### Google Ads Google Ads provides comprehensive remarketing capabilities across the Google Display Network, YouTube, and search results. You can create remarketing lists for search ads and leverage dynamic remarketing to show ads based on specific products or items viewed. ### AdRoll Known for its social retargeting capabilities, AdRoll allows you to reach visitors across platforms like Facebook, X, and other publisher websites. It also offers features like dynamic ads, email marketing automation, and cross-channel attribution. ### LinkedIn Ads Ideal for B2B marketers, LinkedIn Ads offer precise targeting based on job titles, industries, and company sizes. They also have a “Matched Audiences” feature that allows you to retarget users who visited your website or engaged with your LinkedIn profile. ### Perfect Audience Perfect Audience offers a multi-channel approach to retargeting, integrating with HubSpot for seamless integration with your marketing activities. It allows for easy segmentation of visitors based on various user-behavior rules. ### Realize Realize is a dedicated platform for managing and running performance advertising campaigns, powered by over 17 years of proprietary data. It uses a pixel that tracks website visitors for retargeting campaigns within the Realize platform, and it can be deployed for Meta, Google, and beyond. ## Key Takeaways Retargeting is reaching audiences who have already shown either interest in a particular product or even intent to buy; it can also refer to reaching people who have previously made a purchase (or taken another desired action) and may be compelled to do so again. “Retargeting is a way to show your adverts to people who have shown interest in your business in the past, but haven't taken the next step,” says Iqbal Ahmad, founder and CEO of the Britannia School of Academics. “You need to monitor user behaviour closely and adjust the campaigns accordingly. Ensure you understand your audience’s mindset and design content and campaigns that address their concerns. Using discounts and offers works wonders for my business, and I believe it will work for everyone if done correctly.” “Retargeting is a very thoughtful form of digital marketing meant to reach out to users that already interacted with a brand but haven’t ever converted,” adds Mike Szczesny, owner and vice president of EDCO Awards & Specialties. “Businesses can serve specific advertisements to these users while they navigate browsing pages thanks to cookies or tracking pixel advertising systems. Hence, brand recall is kept alive to stimulate return interactions. This effort is effective as it concentrates on known interested users and their chances of converting are significantly higher than new users.” “Retargeting isn’t about bombarding someone until they click,” concludes Mary Sahagun, founder and PR strategist at TargetLink. “It’s about sequencing a conversation they already started. You’re not reminding them that you exist, you’re reminding them why you matter. The most effective retargeting isn’t repetitive — it’s progressive. Each touchpoint should deepen context, answer unspoken objections, or move the user closer to trust.” ## Frequently Asked Questions (FAQs) ### What is cross-device retargeting? Cross-device retargeting is a digital marketing strategy that allows advertisers to target the same user across multiple devices, such as phones, tablets, and desktops, with consistent messaging and offers. It leverages cross-device tracking to identify a user's online activity across different devices and serves relevant ads to them based on their behavior. ### How is AI being used in retargeting? AI is revolutionizing retargeting by enabling more precise audience segmentation, personalized ad delivery, and continuous optimization. AI algorithms analyze user behavior, browsing history, and other data points to create detailed customer profiles, identify patterns, and predict future actions. This allows for the delivery of targeted ads that are more likely to resonate with individual users and drive conversions. ### What are the implications of privacy changes for retargeting in 2025? Privacy changes, particularly the decline of third-party cookies and the rise of privacy-focused technologies, significantly impact retargeting, leading to reduced effectiveness and a need for new strategies. Advertisers face challenges in tracking users across the web and showing personalized ads, forcing them to adapt to new approaches like contextual advertising and first-party data. ### Cookies vs pixels: What are the key differences? Cookies and pixels are both used in tracking people’s activity on a site, but they serve different purposes. Cookies are small text files stored on a user's browser, while pixels are small, invisible images used to track user activity on a website. Cookies store information about a user's browsing history and preferences, while pixels are primarily used for tracking and analysis. --- ### Touchpoints: Optimizing Each Lead Interaction URL: https://www.taboola.com/marketing-hub/touchpoint/ Last Modified: 2025-07-15 09:58:59 Every touchpoint with your potential customers matters, as each makes up part of their journey. These interactions with the company — whether in person, through direct messaging, on social media, or other ways — are essential for leads learning more about your brand, and moving further through the marketing funnel. But, touchpoints don’t happen on their own — they’re part of an intentional marketing strategy. ## What Are Marketing Touchpoints? Marketing touchpoints are interactions with potential customers, from the first time they visit the website and download a free resource, through calling them to answer questions before a final sale. “The most important touchpoint is their first, wherever that comes in,” says Justin Barlow, marketing director at Nigel Wright Group. “This is because how we treat that lead will influence the potential customer’s positive, indifferent, or negative perception of us.” He adds that other key touchpoints are especially noteworthy when the customer takes action to engage with them. “That may be by providing their contact details when they download our content, or sending us an email, or completing a survey, or anything that wasn’t in response to someone calling them to begin a conversation.” He adds that these can be digital or offline touchpoints, including: ### Digital Marketing Touchpoints - How a customer interacts with your product online. - Can occur at any point during the purchase process. ### Offline Marketing Touchpoints - Occur outside of the digital space. - Includes in-person events, trade shows, radio or television advertising, newspaper advertising. ## Different Types of Marketing Touchpoints Just as each of your leads have specific and unique needs, they will interact with your company in their own individual way, too. This means that diversifying your marketing touchpoints to include multiple of the following types is important: ### Website If someone needs something, often they still just “Google it.” That’s where your website comes in as a first essential touchpoint. “Website interactions tell you what people are interested in and how ready they are,” says Colleen Barry, head of marketing at Ketch. “If someone is looking at product features or pricing, that’s a big buying signal. If they’re just reading blogs, they might be early-stage. We track what pages are viewed, time on site, and what CTAs are clicked to score their interest level.” ### Email “Emails are like quiet reminders that you exist and are ready to help,” Barry says. “They can guide leads through the journey without being too pushy. We use emails to educate first, which includes sending useful guides or industry news, and then offer demos or trials once people are warmed up.” Dan Salganik, managing partner and CEO of VisualFizz, calls email “an incredibly powerful touchpoint.” He says it allows for direct and personalized communication at all stages of the journey. Those touchpoints through email might be to introduce the company, nurture leads with valuable content, share relevant updates, offer target promotions, or provide ongoing support, he shares. “For many of our B2B clients, email remains a fundamental tool for building relationships and driving conversions.” ### Online Advertising Barry calls ads a “first handshake.” “They introduce people to your brand, but they have to feel connected to the next step. If someone clicks an ad about ‘how to prepare for new privacy laws,’ and then lands on a random homepage, you lose trust,” she says. “The ad and the landing experience must fit together like puzzle pieces.” Your ads should also be seriously attention grabbing, and should guide leads to their next step in the journey, whether that’s a landing page, website, or other spot, Salganik says. “The effectiveness really hinges on relevant targeting and a clear message.” ## Additional Examples of Marketing Touchpoints If you want to go a bit beyond the basics with touchpoints, diversify your marketing strategy by integrating multiple of these options: - Blog posts: Highly researched blog posts, including subject matter experts and up to date resources, can earn trust and help position you as a thought leader and reputable provider in your industry. - Social media: People are scrolling through their feeds — are you there? Chip away at building a following by posting helpful content on Instagram, X, Meta, and other places your audience hangs out. - Referrals: Sometimes happy fans of your business will just provide these, and sometimes you have to ask — don’t be shy! Some of the best business comes from referrals. - Events: Not everything can be virtual. Plan an in-person event to get people out and experiencing your brand, then talking about it. - Email campaigns: These are ideal for increasing the number of touchpoints a future customer experiences, and solving problems where there are gaps in the journey. ## Marketing Touchpoints and the Customer Journey ### Key Touchpoints in a Customer Journey “Key touchpoints are all the moments a customer interacts with your brand, ads they see, emails they get, your website, webinars, live chat, even reviews they read online,” says Barry. “Every small interaction either builds trust or pushes them away. For us, the first download (like a checklist) is huge, and so is the first time they attend a webinar or open a pricing page.” Some touchpoints matter more than others, and therefore should have more time, money, and resources dedicated to them. Salganik says that for a smaller business, a “compelling social media presence might be a critical early touchpoint.” On the other hand, for a larger enterprise client, he suggests that a crucial touch point would be a “well-received presentation from our team.” ### How to Explore the Practical Application of Mapping Mapping the customer journey involves visualizing, ideating, and formalizing the touchpoints involved in how the lead moves from initial touchpoints through conversion, and ultimately to brand loyalty, referrals, and long-term engagement. This is often done through research on how your audience currently moves through your funnel, then ideating with your team to turn that data into next steps. “We start by sketching out a simple flow, from first contact, like an ad or a blog, to becoming a paying customer,” says Barry. “We map what they see, where they click, what emails they get, and when sales should step in. We also use journey analytics tools in HubSpot to see the actual paths people take. Sometimes they’re very different from what we expect.” Salganik adds that this is where you get into the nitty-gritty of the customer journey. “We often facilitate workshops with our clients where we literally map out every conceivable step a customer might take,” he says. “For a mid-sized growth company, this could mean diving deep into their website analytics to see where users are engaging and where they're dropping off. We might also survey their existing customers to understand their initial discovery process.” He shares that the “real application" comes when they then overlay their current marketing efforts onto this map. “It helps us identify gaps, areas of strength, and opportunities to better align our messaging and activities with the customer's needs at each specific stage.” ### Identify the Measurable Touchpoints Measurable touchpoints are those that a tech tool or your team can evaluate and analyze through data and analytics reports, audits, and other processes. Barlow says to first identify the touchpoints you want to measure. “These are ones that we believe we should be influencing and focusing on to generate more responses. There’s no point analysing touchpoints if they’re not important enough — from an actionable, client-winning perspective — for us to drive activity toward,” he says. From there, Barlow says to ensure all contacts and leads from those touchpoints are actually making it to the marketing team’s tools. “They cannot be left with our salesforce because we won’t be made aware of them all, so our measurement will be wrong. So, we record the lead and its source to enable us to accurately track and understand the volume, fees generated, and ROI,” he says. “Our best converting leads are from SEO, where under one in 10 become a new client within three months. Our slowest converting touchpoints, such as downloaded web content, can take many months and require hundreds of different companies downloading them before any are converted.” ## How to Measure the Impact of Your Marketing Touchpoints ### Get Specific About Your Goals Are you trying to increase your leads within a specific segment of your audience? Are you trying to identify areas of drop-offs in the conversion process before and after your final touchpoint? Your specific goals will help you identify which measurements you need to gather data around. Once you have your goals in place, you can start to ensure that your marketing team is able to answer data questions around those goals. “By doing this, we can analyse their progress every month and calculate resulting fees from new clients won. This enables us to understand the different values at each stage in the leads funnel,” Barlow says. ### Determine Which Metrics Best Reflect Those Goals Once you know what you want to measure, you need to implement tools and experts to gather that data. Metrics might include bounce rates, impressions, clicks, time on your web page or landing page, scroll depth, email open rates, click to conversion rates, form opens/completions/abandonment rates, rate sheet downloads, and signups for newsletters, offers, and other items. Other touchpoint metrics further down the funnel include cart abandonment rates and checkout abandonment rates. Finally, after they’ve converted, there are additional metrics to consider if your goal is to improve customer satisfaction, brand loyalty, and referral rates from satisfied customers. ### Consider Single-Touch Versus Multi-Touch Attribution Models Should every touch get credit for the final conversion? That’s the question at the heart of the single versus multi-touch attribution model discussion. “A crucial best practice is really understanding attribution — which touchpoints are truly driving those conversions. It's not always straightforward, and we often explore different attribution models to get a clearer picture,” Salganik says. Multi-touch attribution models divide “credit” for a conversion amongst each of the touchpoints in the funnel, which is something to note as you consider metrics and make decisions from them. There are pros and cons to using those models, depending on your goals. “We always look at multi-touch attribution, not just ‘last touch,’ because the first helpful guide someone downloaded is just as important as the last demo they booked,” Barry says. ## How to Optimize Touchpoints to Improve User Experience ### Personalize Your Touchpoints “Make every touchpoint feel personal and valuable. We constantly ask: ‘Is this helpful, or is it just noise?’” Barry says. You can personalize through adding names, segmenting your audience to have various touchpoints speak to specific industries or roles, and other strategies. ### Don’t Make Them Wait — They’ll Go Elsewhere “We also optimize speed, fast-loading pages, quick replies to chats, and clear emails. And we always test,” Salganik adds. “For example, just by shortening our webinar registration form, we doubled signups,” Barry says. “For a smaller client, ensure their website is lightning-fast and easy to navigate on any device.” ### Use A/B Testing You aren’t supposed to just guess which touchpoint will work better — instead, try them both. “The key is to remove friction points and make every interaction as valuable and seamless as possible. We often encourage A/B testing and actively seeking customer feedback to continually refine these experiences,” Salganik says. This is why it’s important to choose an ad platform that offers A/B testing powered by AI, which can test constantly and help you optimize on the go. ### Add Actual Value Are you just blasting out ads without any real purpose or greater value for the customer? Flip the script by putting yourself in their place and thinking through what you would want to see. “To me, optimizing touchpoints for a better user experience is about empathy. It's about putting ourselves in the customer's shoes at each interaction,” Salganik says. ### Prioritize the Big Picture Sometimes you need to zoom out from specific touchpoint data to look at the larger picture. “You can over-email your database to generate increased responses in the short term, but this could be harming the user experience, resulting in increased requests to unsubscribe from your database,” Barlow says. “So, marketing teams should keep an eye on wider unintended consequences resulting from their activities.” ### Ask It Again — What Do Your Customers Want? “In general, look at ways people want to engage with businesses and consider how you can provide these touchpoints and make the experience seamless,” Barlow says. “So, for example, do all your “contact us” pages on a website provide a range of ways you can be contacted — by phone, email, form, LiveChat, etc.? And is this within working hours, or have you tested demand outside working hours that may or may not justify investment to extend your service?” ## Best Practices to Manage Marketing Touchpoints ### Lead to Action “You have to tie each touchpoint to a real action. Did they book a demo after the webinar? Did they open five emails, but never click anything? We track micro-conversions like guide downloads or event registrations,” Barry says. ### Keep Tabs on Your Data Your marketing team should know how each of your touchpoints is performing at any given time, but that shouldn’t be the only thing they consider. Instead, balance group brainstorming, expert advice, and metrics to achieve your goals. If you aren’t sure how to find specific types of data, bring in an expert marketing consultant, a service, or a technology tool that can dig out that data and get it in front of your team. ### Don’t Be Afraid to Try New Things A/B testing, catchy advertisements, and other out-of-the-box strategies have their place. You don’t know what types of touchpoints will work with your audience unless you try, so consider playing it safe in some respects with predictability and reliability, while branching out to explore one new idea at a time for specific touchpoints you want to improve. ## Key Takeaways Touchpoints are interactions between your company and your leads across all platforms and outreach on either side. Together, touchpoints represent a customer’s journey through the marketing funnel. Through goal setting, data gathering, concept testing, and expert input, you can improve the efficacy of your touchpoints for more conversions and more loyal lifetime customers. ## Frequently Asked Questions (FAQs) ### What role do emails play as marketing touchpoints? Emails are essential to understanding and gauging interest through open rates, subscribe rates, unsubscribe rates, and other metrics. Targeted email campaigns can be an essential part of personalizing touchpoints for leads, and ultimately better understanding the customer journey through the funnel. ### How do online advertisements act as marketing touchpoints? Online ads build awareness, help leads through the consideration and decision/action phases, and can ultimately be the reason a lead converts. Examples include display ads, Instagram ads, video commercials, native content advertisements, and others. ### How do website interactions serve as marketing touchpoints? Website interactions might seem small or insignificant, but they are far from it. In fact, website interactions — including chatbot discussions, contact form messages, downloads, and email list subscriptions — matter. Each serves as a marketing touchpoint showing that an interested lead is ready for the next steps. --- ### Push Notifications: The Small But Mighty Messages That Matter URL: https://www.taboola.com/marketing-hub/push-notifications/ Last Modified: 2026-06-22 08:32:56 With the growing amount of apps, websites, and online services we use daily, staying informed without being overwhelmed is a challenge that can quickly become disorienting. Push notifications offer a direct and efficient way to get timely updates, without needing to constantly keep an eye on everything. These short and simple messages often have a well-planned strategy behind them. As a marketer, that can make them a challenge to craft. Sure, they’re short, but there’s an interactive component you don’t otherwise have when writing for a TV spot or print ad. Even a banner ad, which is still asking for the user’s click, it’s not as personal as a push. These notifications get you right in front of a user’s eyes, on their device, so they’ve got to be quick, concise, and informative. This article will take a look at the whole process of push notifications, from understanding their fundamental purpose and inner workings, to mastering their setup, implementation, and optimization for a better, more targeted user experience. ## What Are Push Notifications? Starting with the basics: Push notifications are a way for applications and websites to send users updates, alerts, or reminders in real time. Instead of people needing to always check for new information, with push notifications, the information finds them. These notifications aren’t limited to any one type of app or service, either. Push notifications can cover a wide range, from a breaking news alert, food delivery status, an appointment reminder, or a direct deposit confirmation. They’re quick and practical, designed to be immediate and attention-grabbing, with the goal of ensuring people don't miss the important updates that matter to them. ## How Do Push Notifications Work? The process behind push notifications involves several key aspects and moving parts. It all starts with the marketers behind an application or website who want to send the notification. They craft the message and prepare it to send out. The mode of delivery is typically a Push Notification Service (PNS) specific to your operating system (like Apple's APNs for iOS or Google's FCM for Android) or web browser. When you install that app, or the website you’re visiting asks for permission, your device registers with that relevant PNS. The application or website then sends the notification data to the PNS, which in turn routes it to your specific device. Finally, your device receives the notification and displays it to you, often with a sound or vibration (or recognizable haptic). ## Different Types of Push Notifications Push notifications are a simplified, streamlined, and stripped way down to receive important information on a device without opening up the app or site. They appear in various forms in order to meet different needs. Here are the ones you’ll most often see: ### Alert Notifications These are by far the most common type, delivering immediate information like news headlines, social media updates, or direct messages, and usually appear as banners or pop-ups on your screen. ### Badge Notifications Badge notifications are small icons in bubbles that appear on your main screen’s app icons. They indicate the number of unread messages or pending actions within the app they’re on. ### Sound Notifications Sound notifications accompany visual notifications with a specific sound to capture your attention and get you to take action. Much like the haptics, these can be useful to let users know which application is notifying them, without having to look at their device. ### Rich Media Notifications Going beyond simple text, rich media notifications can include images, videos, or interactive buttons, providing more context and allowing for quick clicks directly from the notification itself. ### Location-Based Notifications These are triggered by your device's location, and offer up relevant information or reminders when you’re in a specific geographic area, like a nearby attraction or travel tips. ## Benefits of Push Notifications ### Staying Informed This is the area where push notifications truly shine. They’re one of the best ways to provide timely updates on important events, news, or information you want your users to know. ### Increased Engagement Apps and websites can re-engage users who haven’t signed in for a while by reminding them of new content, features, or unfinished tasks through push notifications. ### Personalized Experiences Push notifications can easily be tailored to a user’s preferences and behavior, delivering relevant and useful information that they want to see, not just filling their alerts with generic junk. ### Improved Efficiency Before push notifications became commonplace, the need to constantly check a device for updates was annoying and tedious. These notifications save users time and effort, offering up a few short sentences with everything they need. ## How to Set Up Push Notifications Setting up push notifications can vary slightly, depending on whether someone is using a mobile device or a desktop browser, and what the app’s settings offer, but it’s generally a similar process: ### Mobile When users install a new app, one of the first things it’ll ask is for permission to send them notifications. They can usually choose to allow or deny, and also manage those preferences later in their device's settings menu if they become too frequent or invasive. Users can also customize which apps can send them notifications, as well as the style of those notifications, like banners, sounds, and badges. ### Desktop When users visit a website that supports push notifications, the browser will usually ask for their permission to send them updates, and they can grant or deny that request. If they want to go back later and manage that website’s push notifications, they can find the settings within the browser's preferences, often under "Privacy and Security" or "Notifications." This is where they can see a list of websites that have asked for permission, and manage what they’re allowed to send them. ## Push Notification Strategy Tips With push notifications, you’ve only got a small window and few words to get your message across. To make the most of them, consider these strategies: ### Personalization Generic notifications are less likely to be effective, so customize yours to individual user preferences and typical behaviors, making the user feel seen and cared about. ### Segmentation Much like personalization, but on a larger scale, group your users based on their interests, demographics, or past actions. By doing this, you can send more relevant notifications to specific audience segments. ### Timeliness The right timing is a big part of push notifications. Send them out at the moment when users are most likely to engage with them, and don’t forget to consider time zones and user activity patterns, too. ### Value Proposition Make sure your notifications provide genuine value to the user and not just useless attempts to keep their attention. That may work in the short term, but vital updates, exclusive offers, or helpful reminders are information people generally welcome if they want your services. ## Push Notifications: Best Practices ### Obtain Clear Consent Always ask for explicit permission before sending push notifications, and clearly explain the type of notifications users can expect. Skipping that step can be seen as invasive and untrustworthy, and create a negative association with your brand and product. ### Don't Overdo It Be mindful of frequency and avoid sending too many notifications, as this can lead to user fatigue and opt-outs. ### Make Them Actionable When possible, include clear calls to action within your notifications. If not, you may get a user who wants to take that next step, but isn’t sure how, and gives up soon afterwards. ### Test and Iterate Don’t be afraid to experiment with different types of notifications, messaging, and timing to see what resonates best with your audience. ## How to Measure Push Notification Performance Tracking core metrics is crucial for understanding the effectiveness of your push notification strategy, but to truly grasp the impact they’re having, it's best to go beyond the basics. Realize, for example, offers in-depth features like comprehensive analytics dashboards, which can provide you with deeper insights into metrics like user engagement and conversion paths that were initiated by your notifications. Taboola Push takes it a step further, allowing you to send notifications to both browsers and mobile devices, at the frequency you choose, to grow your audience and gain their trust. This multi-pronged plan helps you develop and refine your approach for better results. Here are some key performance indicators (KPIs) to track over time to gauge the effectiveness of your push notifications: ### Open Rate Open rate is one of the most important metrics to see how your notifications are performing. This shows you the percentage of users who open or interact with your push notifications, and is key to evolving your approach to what’s working and what isn’t. ### Click-through Rate (CTR) CTR tracks the percentage of users who click on a link or call to action (CTA) within your notification. ### Conversion Rate A conversion rate indicates how many users completed a desired action after receiving a push notification. This could be anything from making a purchase to signing up for a service, or whatever your end goal is. ### Opt-Out Rate Just like keeping track of how many people click through and take action, the opposite is true, too. Opt-out rate monitors the number of users who unsubscribe from your push notifications, and is equally important for identifying the potential issues within your strategy (and working to fix them). ## How to Optimize Push Notifications When you have accurate and complete performance data to guide your efforts, you can increase the efficiency of your push notifications for better results. Here are just a few ways: ### A/B Testing This lets you experiment with different content, timing, and calls to action to see which variations perform best. This method takes a lot of the pressure off choosing just one final option, too, as if it were a billboard or something more permanent. ### Refine Segmentation Don’t just send your ad out into the world and hope for the best: Continuously analyze your user data to create more targeted and effective audience segments, and keep on refining based on what you learn. ### Improve Personalization Users might be more likely to click if they know the notification was tailored to them. Use the insights you’ve learned from the data about your users’ behavior and preferences to deliver more relevant and engaging content that makes them want to click through. ### Analyze Delivery Times Even the most meticulously crafted messages can flop if the timing isn’t right, so the time you send out your push notifications is a major piece of the puzzle. When you take a good look at your users’ activity patterns and engagement metrics, you can better determine the most effective times for grabbing their attention. ## Key Takeaways Push notifications are a small but powerful tool in your marketing arsenal. They’re excellent for delivering timely information and engaging with users, and by gaining a more nuanced understanding of how they work, the different types, and best practices for implementation, you can keep on perfecting them. ## Frequently Asked Questions (FAQs) ### How can I ensure my push notifications are personalized and engaging? It all comes down to data. Start by taking the data you have about your users, like their names, past interactions, and stated preferences. This information is a goldmine when it comes to customizing the content of your messages, and shows users that you’ve taken the time to create the best experience possible. Next, instead of generic blasts, go for notifications that feel directly relevant to the person receiving them. Craft your notifications with compelling content, incorporating rich media like images or GIFs that can capture attention more effectively than plain text alone. Keep them short: Remember that messages should be concise and benefit-driven, clearly communicate value to the user, and always include a strong CTA that prompts a specific action. Finally, don’t settle: Always be analyzing the performance of your notifications, and evolve it based on what’s working best with your audience. ### What are the best practices for obtaining user consent for push notifications? Your first instinct might be to get the consent page in front of users’ faces as soon as they visit your site or download the app. But, take a step back for a minute, and know that it’s entirely possible to get users on board with push notifications in a way that's both respectful and doesn’t disrupt their privacy. Not only that, but it’ll boost your brand’s reputation with them if you’re transparent and up-front with everything. The opt-in process should be straightforward, with soft prompts that blend in naturally, rather than intrusive pop-ups. Be sure to clearly explain the perks of opting in, letting users know what kind of notifications they’ll be getting. Think about how these alerts will improve their experience and lives, like timely updates, exclusive deals, and personalized content. Once they sign up, it's also important to provide users with quick access to manage their notification settings when they want to change them, including options to completely opt out or adjust the types and frequency of the notifications they’re getting. ### How can I avoid push-notification fatigue and keep users engaged? Push-notification fatigue is really common due to the number of apps people use, but also probably the easiest to avoid in the first place. Don’t just bombard users with notifications, and remember that less is more. Really get to know your audience by segmenting them, so your notifications hit home with what they care about, rather than clogging up their notifications with junk, focusing on fewer but more impactful alerts. Let users tweak their notification settings, too, so they can choose what they want to see and when. Make sure every notification is going to be worth their time, whether it's sharing important updates, giving them special perks, or encouraging them to interact. Once you send those messages out into the world, keep a close eye on how your notifications are performing to see what works best, always aiming for quality over quantity. ### What are some effective strategies for segmenting my push-notification audience? Luckily, there’s lots of options for this one. You can segment users based on data like demographics, behavior (such as app usage and purchase history), interests, and what content they engage with. Think about lifecycle segmentation, too, like sending onboarding messages to new users and special deals to your most loyal customers. All this helps you make your push notifications even more targeted, boosting engagement and lowering the chances of users opting out. ### How can push notifications be used to improve customer support and satisfaction? When you picture push notifications, the first thing that pops up in your mind might be messages and alerts, but don’t overlook the power that they have as a tool for boosting customer support and satisfaction, too. These are ideal for providing timely and proactive assistance, and if you’ve ever been eagerly waiting for a package or needed assistance with something immediately, you know how helpful they can be. Push notifications are an ultra-convenient way to keep your users updated on their support requests. Instead of them having to refresh their screens or wait on hold, you can send a quick update straight to their phone or device. It’s also a smart way to stay ahead of any potential issues, and can reduce the strain on your customer service reps, too. For example, if there’s a service interruption or a recurring problem, a fast and friendly notification with a heads-up or a useful FAQ can avoid a lot of frustration, both for the user and for you. Notifications can also check in on specific questions they’ve asked or provide personalized solutions just for them. It’s definitely more efficient, but goes beyond that — it shows users that you’re attentive, you care about their experience, and in the end, makes them want to use your services more. --- ### YMYL (Your Money Your Life): What It Is, Why It Matters URL: https://www.taboola.com/marketing-hub/ymyl/ Last Modified: 2026-03-10 11:50:07 Not all content is created equal — at least in the eyes of Google. When it comes to a searcher’s health or financial well-being, Google’s algorithms take things seriously. In the interest of consumer safety, the top search engine puts certain content in a category called “Your Money or Your Life,” abbreviated as YMYL. Whether you’re a content creator, marketing professional, or business owner, understanding YMYL isn’t just important — it’s essential. This guide will help you understand what YMYL is, why it matters, and how you can make it a part of your content creation strategy. ## What Is YMYL? Your Money or Your Life (YMYL) is a content guideline that Google uses to protect its users. Google defines it as content that could significantly impact the health, financial stability, or safety of individual users, or the well-being of society as a whole. Since the nature of this content has potential to create harm, Google puts safeguards in place that prioritize expertise and trustworthiness. “The internet is an incredible source for learning and exploring, but since anyone can publish anything, not everything you find is accurate, helpful, or even true,” says Kelley Muhsemann, marketing manager at R.W. Rogé & Company, Inc. “If content has the potential to impact someone’s finances, health, safety, or general well-being, Google wants to ensure it’s accurate, trustworthy, and created by experts. The goal is to protect the reader and ensure they’re getting reliable, helpful information.” ## What Does YMYL Mean in SEO? Search engine optimization (SEO) is all about creating content that ranks. For SEO purposes, YMYL means meeting Google’s stricter standards in order to rank well. If your content falls into one of these subject matter areas, you’ll need to demonstrate trustworthiness, credibility, and relevance. “YMYL is about trust, plain and simple,” says Paul DeMott, chief technology officer of Helium SEO. “Trust takes more than SEO tricks: It takes real credentials, real transparency, and regular upkeep. You cannot treat it like any other content category.” ## Why Is YMYL Important for Google? YMYL is central to Google’s mission, which is “to organize the world's information and make it universally accessible and useful.” Google knows the repercussions that can come from misinformation in certain areas, and for that reason, the algorithms are trained on rigorous standards for those same areas. “Google is essentially trying to keep the user safe from seeing any content that can mislead or harm users,” says Liam Quirk, founder of Quirky Digital. “For example, let’s say you’re a company providing this type of information and are potentially misleading or harming your readers. You’re then going to be damaging your trust and reputation as a business, ultimately leading to no one trusting your business for serious advice and your rankings dropping.” ## How Does Google Treat YMYL Content Differently? Google uses a combination of algorithms and human reviewers to ensure YMYL content meets its standards. Content is evaluated for factors like: - The author’s credentials. - Transparency about the site’s purpose. - User trust signals like reviews. - Factual accuracy. Poor-quality YMYL content can not only fail to earn top rankings, it may even be penalized or de-indexed. ## In 2025, What Are the Latest Google Updates Related to YMYL? Over the past couple of years, Google has leaned heavily into a framework known as E-E-A-T. Sites that demonstrate a high level of one or more of these qualities will rank higher. E-E-A-T stands for: - Experience. - Expertise. - Authoritativeness. - Trustworthiness. In 2025, Google has placed even greater emphasis on these factors. Content authored by those with lived experience or by verified experts takes priority when Google chooses which sites rank. ## What Industries/Categories/Topics Are Considered YMYL? ### Health and Medical Information One of the biggest areas impacted by YMYL is health content. Google expects to see expertise when a site is recommending remedies and describing diseases. This applies to both physical and mental health, and as Eunice Arauz, founder of Pets Avenue, learned, human health isn’t the only area impacted. “We lost traffic and engagement on one of our most trafficked pages after Google's update in March,” Arauz says. “This page recommended natural remedies but did not include sources or help from an expert. The content was technically accurate, but that didn’t mean it was worthy health-related guidance for Google to share.” Pets Avenue added their veterinary consultant’s name, and within two months, impressions improved by 42% and click-throughs had nearly returned to previous levels. ### Financial Advice Whether it’s tax deductions or credit card interest rates, financial content always falls under YMYL. Google sees misleading financial content as harmful and looks for signals that content is factual, trustworthy, and backed by experience. ### Legal and Government Information Legal topics always see plenty of search activity. Law firms need to be aware of YMYL, as do site owners with content related to topics like civil rights and estate planning. ### Safety and Emergency Preparedness Content relating to natural disasters and crime prevention needs to meet YMYL guidelines to rank well. The same goes for content relating to online safety. ### News and Current Events Misinformation runs rampant online, and for that reason, Google does consider some news-related content YMYL. Not all news content is YMYL, though. More lighthearted news items like entertainment won’t have to meet these stricter guidelines. ## E-E-A-T and YMYL Content As mentioned above, E-E-A-T has become more important than ever in Google’s ranking factors. Let’s go a little deeper, though: - Experience means that the author of the content has firsthand familiarity with the topic being discussed. - Expertise refers to an author’s experience, education, or formal qualifications. - Authoritativeness describes how the brand is seen by those who view the content. - Trustworthiness requires that a site offer transparency and accuracy. The more factors your content has, the better your chance of ranking well within YMYL categories. ## YMYL and SEO If your content falls within Google’s YMYL parameters, standard SEO strategies won’t cut it. Here’s what you should consider: ### Considerations in Terms of SEO Strategy Keywords aren’t completely out of the equation, but a few things are more important. Fresh, authoritative content is more valuable in 2025 than the number of times you’ve used a specific phrase. One of the best things you can do to improve your chances of ranking is to prioritize credibility. Consider how the typical site visitor would view your content: Does it come across as authoritative and trustworthy? This will help you maximize your efforts. ### What Factors Are Critical? Here are a few of the most important factors in ranking for YMYL: - Ensure your content has bylines featuring clear experts in the field. - Include citations to authoritative sources. - Ensure your site uses the secure protocol (HTTPS). - Maintain clear and up-to-date About and Contact pages. - Collect genuine user reviews and feedback. ### How to Improve Rankings If your content is struggling, here are some things you can do to give your site a boost: - Build topic clusters around core YMYL themes. - Update content frequently, adding any new facts and/or data. - Use Schema.org to highlight credentials. - Leverage high-quality visual formats like carousels and videos to boost engagement. Fortunately, it’s easier than ever to create content that responds to modern ranking signals. Using tools like predictive targeting and social asset importers, you can make the most of every piece of content you create. ### Backlinks With YMYL, one thing hasn’t changed, and that’s the importance of backlinks. But, in 2025, Google pays close attention to the quality of backlinks. Links from respected sites in your genre (universities, medical facilities, and financial sites, for instance) carry more weight than mentions on lesser-known sites. ## Best Practices for Creating YMYL Content ### Build Trust Google prefers sites that are as transparent as possible, seeing them as trustworthy. Make sure every article you post includes: - Clear author bios. - Current contact information. - Verifiable facts. ### Emphasize User-Generated Content Nothing builds trust like statements from current and former customers. Your site should include testimonials and honest reviews, especially if you’re promoting products or services. Reviews serve as social proof, demonstrating that your site is trustworthy. ### Embrace Multimedia for Clarity Part of building user trust is ensuring people can easily comprehend the information they’re reading. For that reason, if you’re describing complex concepts on your website, it’s important to clarify them for visitors. This not only helps with ranking, it also improves the user experience. Creative formats like videos and charts can help clarify things, too. You can also use interactive tools to fully engage customers and drive your points home. ### Automate, but Don’t Cut Corners AI-powered tools have made it easier than ever to reach consumers. You can use these tools for everything from building ads to testing copy, saving time and gaining a competitive edge. However, it’s important to ensure you’re still meeting all YMYL requirements as you’re creating content. ## Risks of Inaccurate YMYL Content and Compliance ### Legal and Financial Liabilities If you’re posting erroneous content — unwittingly or not — you could find yourself on the wrong side of a lawsuit. Say someone reads a piece of unvetted financial advice on your site and acts on that advice. That person could later take any losses to the legal system, costing your business serious money. ### Loss of Organic Visibility Google can delist or penalize sites that it sees as misleading or dangerous. This may mean your site doesn’t show up on the first few pages of search results for related queries, or it could mean it doesn’t show up in search at all. With each year, Google’s algorithms seem to get better at sensing and taking action against misinformation. ### Reputation Damage If your site isn’t meeting Google’s requirements, chances are it isn’t up to par with the public. If your website visitors see your content as misleading or questionable, you may find they leave and don’t come back. ## Troubleshooting YMYL Ranking Issues ### Ranking Drop If your YMYL content suddenly plummets in rankings, start by looking at the following: - Is your content accurate, with trusted sources cited? - Is your content outdated or stale? - Does the bylined author of the content have credentials? - Does your site seem trustworthy, with updated About and Contact pages? - Are all backlinks to your site coming from respected sites in your industry? ### How to Recover From a YMYL-Related Ranking Decline Losing rank isn’t the end of the world: It might not seem like it, but you can come back. Go through each of your pages and optimize them for YMYL, including having experts either update or review articles so you can add a byline. Make sure you’ve cited any sources for facts that are included in your pieces. You can also get a hand from technology. Tools that predict audience intent and streamline asset repurposing can help you fill your site with YMYL-friendly content. ### How Often Does Google Update Its YMYL Guidelines? Google might require transparency from the sites it ranks, but that same transparency doesn’t apply to its search rankings. For that reason, there’s no publicly posted schedule for algorithm updates. Like other ranking factors, Google updates its YMYL guidelines when changes become necessary. Typically, updates are made in response to changes in trends, user behaviors, or issues that need to be addressed. ## Key Takeaways YMYL, which stands for Your Money or Your Life, is Google’s designation for content that impacts a user’s financial stability, physical and mental health, or safety. Search algorithms apply stricter standards to content that falls within its YMYL parameters, which include ensuring content demonstrates experience, expertise, authoritativeness, and trustworthiness. High-quality content and AI-powered tools can improve visibility when used to enhance, but not replace, human expertise. ## Frequently Asked Questions (FAQs) ### Is finance considered YMYL? Yes, financial topics are considered YMYL content to Google’s algorithms. Content relating to financial topics should be written or reviewed by qualified professionals. “Anything that affects a person’s future or well-being in a serious way usually falls under YMYL,” DeMott says. “That includes taxes, retirement planning, credit repair, mental health, medical advice, and even parenting in some cases. We have had to flag this with clients more than once. Just because something seems like just another blog post does not mean it is low risk.” ### Why did my YMYL website's rankings drop? Google is constantly tweaking its algorithms, so what ranks well today might be on Page 20 tomorrow. As Milda Darulienė, senior SEO specialist at Omnisend, explains, YMYL has to meet E-E-A-T standards to rank well and hold that ranking. “If a YMYL website’s rankings drop, the content probably lacks signals of trust and authority,” Darulienė says. “It might be outdated, missing sources, not reviewed by experts, or written by someone without the right background. Sometimes, the website itself may lack clear information about the business, such as About or Contact pages. Page design can also make it feel untrustworthy.” ### Why do YMYL pages face stricter quality checks? YMYL pages face stricter quality checks because the nature of the content is so sensitive. Inaccurate information in those subject matter areas could harm consumers’ health, safety, or finances. By working to protect its users, Google can build and maintain trust. “If someone follows bad advice from a lifestyle blog, maybe they waste a Sunday,” explains Sasha Berson, co-founder and chief growth officer at Grow Law Firm. “But if they follow bad advice from a YMYL site, they might lose their life savings or make a decision that messes up their health. So, Google brings the heat.” ### Should I hire experts to write YMYL content? Absolutely. Expert-penned content not only performs better in search, but it also helps build credibility with readers. You don’t have to hire full-time help to take your content to the next level: Freelancers and co-authors can take content creation off your hands. You could also consider paying an expert to review your content to share a byline. ### How often should I update my YMYL content? Stale content can kill visibility, so it’s important to regularly refresh any YMYL content you’ve produced. Experts seem to agree that you should update YMYL content twice a year at the very least. “This isn’t evergreen content you can set and forget,” says Ivan Vislavskiy, CEO and co-founder at Comrade Digital Marketing Agency. “If your law blog is quoting laws that changed last year, you're losing credibility with readers and with Google.” --- ### Ad Creative: What It Is, What Makes It Effective URL: https://www.taboola.com/marketing-hub/ad-creative/ Last Modified: 2026-03-24 08:30:19 In an increasingly saturated digital landscape, audience attention is harder to capture than ever. While media buying strategies and audience targeting are critical parts of any marketing plan, even the most precise placement can be for naught if paired with lackluster creative. In this guide, I’ll break down what ad creative is, its standard formats, and some best practices to ensure your campaign hits the mark. ## What is Ad Creative? Ad creative refers to the visual and textual elements used in advertising. Its primary goal is to communicate a message to your target audience and persuade viewers to take action, such as following a social page, subscribing to a newsletter, or making a purchase. It can even convey information to entice users to take action outside the digital space. A single unit of ad creative can include an array of elements, like headlines and copy, video, call-to-action buttons, or enticing, inspiring, or otherwise captivating images or photos. While being visually appealing is a hallmark of this marketing asset, strong creative doesn’t just look good — it aligns with your audience’s intent, causing them to pause and act. In short, successful ads disrupt our scrolling habits by raising brand awareness. ## What Are the Different Types of Ad Creative? ### Static Images Static images include display banners or social ads with a single image, headline, and supporting text. They’re simple and quick to produce, making them ideal for quick-turnaround efforts or testing messaging. ### Video Ads Video is one of the most engaging formats available today: Some studies indicate that videos can increase conversions by as much as 80%. Videos allow brands to tell a more robust story and are particularly effective for top- and middle-funnel campaigns. Some ad creative platforms take this a step further by offering in-feed, native video placements that mimic editorial content. ### Interactive Ads Interactive creative includes elements like polls, sliders, or quizzes. These can significantly boost the time users spend with your ad and improve engagement metrics. ### Display Ads Display ads use eye-catching graphics and multimedia elements to grab the user's attention including images, animations, videos, and often combine these with text and a call to action, leading the users to specific landing pages. They appear in designated ad spaces on websites, mobile apps, and social media feeds. ### Native Ads Native ads blend seamlessly with the content surrounding them. They don’t disrupt the user experience and often outperform traditional ads. ### Carousel Ads Common on social platforms or websites, carousel ads allow you to showcase multiple products, features, or benefits in a swipeable format. They can also be used to tell a story or otherwise engage with users over a series of slides. ## How Do Marketers Use Ad Creative? ### Increasing Brand Awareness In a saturated market, brands must fight to stand apart from others in their vertical, and ad creative is frequently their first opportunity to do that. Strong visuals and copy work together to communicate your brand’s voice, mission, and why users should care. The relationship between ad creative and brand awareness is particularly important when potential customers are at the top and middle of the funnel. ### Cultivating Funnel-Specific Content Ad creative can also be a valuable tool for moving potential customers further into the funnel. For example, creative designed to raise brand awareness on social media can welcome customers to the top of the funnel, and once there, retargeting campaigns leveraging tweaked ad creative can help usher customers further along, eventually providing them with the information, trust, and sense of urgency to purchase or otherwise complete a conversion activity. ### Increasing Engagement Well designed ad creative can hook a user, grabbing their attention as they scroll through social media, read a publication, or visit a landing page. Great ads push users towards completing a desired action, whether it’s clicking through to your website, following your page, signing up for an email, or making a purchase. ## What Are Some of the Biggest Challenges Marketers Face with Ad Creative? ### Ad Fatigue Ads can become less effective the more they're shown. When audiences are exposed to repetitive ads, they can tune them out and scroll past without a second thought. As such, stale ads can be ineffective and costly, even if they were high performers at launch. Tip: Set a cadence for rotating creative every 2–4 weeks. Use performance signals like declining click-through rate (CTR) or engagement to trigger refreshes. ### Testing Failures As your team designs creative, it’s easy for even the most experienced marketers and innovative individuals to overlook elements that may not resonate with the target audience. Without testing creative, your team may leave dollars on the table and audiences unengaged. Tip: Create a testing matrix that outlines variables (headline, image, call to action ) and tracks performance. A/B test everything — if you have access to an ad creative platform that utilizes AI for A/B testing, this will allow you to test continuously and optimize in real time. ### Inconsistencies Across Channels Ad campaigns should be unified, telling the story of your brand across multiple channels. However, maintaining visual and messaging consistency across channels can be more challenging than it sounds. Tip: Develop a creative style guide for your brand and use platform-specific checklists to ensure each version meets channel requirements without losing cohesion. Then, ensure that the creative is cohesive while also meeting the needs of the target audience. ## Ad Creative: Best Practices ### Ideation Every ad campaign should start with ideation, allowing you to establish your target audience and intended outcome. The best way to begin is by clearly understanding your audience, including pain points, desires, and behavior. Use insights from previous campaigns, competitor research, or keyword intent to brainstorm relevant concepts. ### Copywriting Copywriting is often undervalued, but it’s a key part of effective marketing campaigns. Ad creative should have clear and concise content that resonates with your audience and your intent. Efforts should focus not only on the primary message but also any CTAs. ### Design A good design is more than just aesthetically pleasing — it guides the eye, reinforces the message, and keeps the experience intuitive. That means every ad should be built using fundamental design best practices, emphasizing elements like contrast, hierarchy, accessibility, and clarity. Further, design needs will change from channel to channel. Always consider these changes when designing ad campaigns to ensure they are scalable and meet the standard of the intended channel. ### Funnel Stage Alignment and Goals Your creative should align with campaign objectives at each level of the funnel. Below is a look at the role ad creative plays in each level of the funnel: - Top-of-funnel creative should educate or entertain, serving as your audience’s introduction to the brand. - Middle-of-funnel creative should educate further while differentiating your brand, giving your audience the information they need while deciding on a purchase or other conversion. - Bottom-of-funnel creative should reduce friction, build trust, and encourage conversion. ## How to Test Ad Creative ### A/B Testing A/B testing involves running two versions of an ad with one variable changed — like the headline or image — while keeping everything else constant. This allows marketers to determine which specific element is driving performance. It’s a reliable, low-risk method for learning what resonates with audiences and what doesn’t. Tip: Run A/B tests on high-traffic placements first so you can gather statistically significant data faster and repeat them more confidently. ### Multivariate Testing Multivariate testing examines the performance of multiple variable combinations at once — like headline + image + CTA. This method gives deeper insight into how creative components interact, helping marketers build higher-performing combinations. Tip: Use this method when you have the budget and traffic volume to support it. Otherwise, it can take too long to reach reliable conclusions. ### Segmented Audience Testing Audiences don’t respond to creative in a monolithic way. A segmented testing approach breaks your audience into groups — by behavior, interest, or demographics — and tests creatives within each segment. Tip: Use platform data to build meaningful segments, then compare performance to identify which creative works best for each group. ## How To Optimize Ad Creative ### Rotate Ad Creative To combat ad fatigue and keep your campaigns performing, it’s essential to rotate creative regularly. When users see the same ad multiple times, they may begin to ignore it, which can drive down your engagement and click-through rates. Build a creative calendar that outlines rotation timelines, allowing for fresh visuals and messaging every few weeks. ### Make Data-Driven Tweaks Optimization doesn’t mean reinventing the wheel — it’s often about making smart tweaks based on real data. Pay close attention to performance metrics like scroll depth, bounce rate, and conversion rates. Analyze which parts of your creative are performing and double down on those elements. ### Personalize Creative Generic creative may drive awareness, but personalized creative drives action. Tailor messaging, visuals, and CTAs based on user behavior, geographic location, or browsing history. ### Create Cohesive Campaigns Your ad may be great, but if it leads to a mismatched landing page, conversions will suffer. Ensure your visual language, tone, and offer stay consistent from the ad to the landing page. This reduces friction and builds user trust. ## Ad Creative Terms to Know ### Click-Through Rate (CTR) The percentage of users who click on your ad after seeing it. High CTR usually indicates relevant and engaging creative. You can track CTR directly through your ad platform’s analytics dashboard, and use it as a quick barometer for how well your creative grabs attention. If CTR drops, it’s often a sign that your creative needs refreshing or better targeting. ### Frequency Capping Frequency capping is the practice of limiting how often a user sees the same ad to avoid fatigue. This setting helps prevent oversaturation, which can lead to declining engagement or even negative brand perception. Most ad platforms let you set frequency caps per user per day or week. Use performance trends to find your optimal cap threshold. ### Return on Ad Spend (ROAS) ROAS is a performance metric that shows how much revenue you earn for every dollar spent on advertising. A high ROAS indicates that your creative is not only attracting clicks but also driving conversions. Track ROAS alongside other funnel metrics to understand whether your creative drives qualified traffic. ### Dynamic Creative Optimization (DCO) DCO is a technology that automatically tailors ad creative based on audience data. You can monitor DCO performance by tracking segmented engagement metrics — like CTR by audience type or conversion rate by location — within your reporting tools. ## Key Takeaways Ad creative is a critical factor in campaign performance. Different formats work better for different objectives and funnel stages, and consistent testing, optimization, and personalization are essential to stay competitive. ## Frequently Asked Questions (FAQs) ### What does Adcreative.ai do? Adcreative.ai is a tool that uses artificial intelligence to quickly generate and test ad creative. It automates the creation of visual and copy assets based on your campaign goals, helping marketers save time and reduce creative bottlenecks. The platform also provides predictive insights to guide decisions before launching paid campaigns. It’s especially useful for scaling creative output without relying heavily on design teams. ### What are the components of ad creative? Ad creative includes several essential elements that work together to grab attention and encourage action. These typically include visuals like images or video, a headline that captures interest, supporting body copy, and a clear call to action (CTA). Some creative may also feature interactive components, such as buttons or sliders. Each element plays a distinct role: While the image stops the scroll, the headline builds intrigue, and the CTA drives clicks. Thoughtful composition and alignment with campaign goals are key. ### How do I make ad creative? Start with a clear goal and audience in mind. Choose visuals and messaging that align with the campaign objective and platform. Use tools to create variations, then test and refine performance by identifying your objective, audience, and platform. Then develop visuals and copy tailored to that context, and test variations. ### What tools exist for ad creatives? Platforms like Canva, Adobe Express, and Realize help streamline the process of designing, testing, and optimizing ad creative across formats. These tools enable marketers to move faster while maintaining creative quality. ### What ad creative should be used to increase app installs? To increase app installs, use native or video creative that demonstrates key features and benefits of your app. Highlight social proof, smooth functionality, and a simple CTA that directs users to download. Focus on mobile-first formats and keep the messaging short and compelling to drive clicks and conversions efficiently. ### How do I write creative that builds trust for BOFU leads? To build trust with BOFU (bottom-of-funnel) leads, focus on clear messaging, social proof, and low-risk offers. Highlight customer testimonials, satisfaction guarantees, security certifications, or free trials. Avoid hype or exaggeration and use straightforward, confident language that reinforces credibility. Your creative should assure the user they’re making a smart, safe decision with minimal risk or hassle. ### How can I improve my CTR with better creative? Focus on using attention-grabbing visuals, benefit-led headlines, and compelling CTAs. Align the creative with audience intent and test different formats regularly to find what resonates most. ### What are common reasons for ad creative fatigue? Ad creative fatigue occurs when audiences repeatedly see the same visuals and messaging, reducing engagement and lowering performance. Common causes include overexposure, stale design, and lack of variety across campaigns. When ads feel repetitive or irrelevant, users tune them out. To avoid this, marketers should regularly refresh creative assets, test new formats, and personalize content for different segments. Regular updates help maintain relevance and improve long-term campaign efficiency and ROI. --- ### 8 Travel Ads That Show How To Grab the Attention of Future Tourists URL: https://www.taboola.com/marketing-hub/travel-ads/ Last Modified: 2026-03-30 07:09:49 Travel and tourism advertising in 2025 uses a range of formats and channels to engage and nurture prospective tourists. Destination marketing these days includes influencers, AR or VR technology, social media, user-generated content, and more. Many typical performance marketing tactics can be successful in travel and tourism advertising, but personalization, great quality creative, and storytelling are especially important. When you’re working in destination marketing, consider ways to reach goals by using audience data wisely, understanding your target markets, telling great stories, building amazing visuals, and enticing your prospects with hidden deals, insider tips, and one-of-a-kind experiences. Plus, consider all the potential channels and formats to use to reach your desired targets. ## Which Tourism Ad Features Best Capture the Attention of Potential Customers? Modern travel and tourism ads have gotten more creative and interesting in showcasing destinations — both well-known and off the beaten path. Consider the following tips when you’re building these ad campaigns: ### Use Amazing Visuals You should always use the best possible visuals in creative ad campaigns. Nowhere will poor quality imagery stand out more than with travel and tourism audiences, so it’s worth putting your budget toward photography and videography of the destination you’re promoting. Recognizable landmarks or other sights can help anchor your campaign visually. ### Tell a Story Today’s travelers often have goals in their travel: Checking off a bucket-list item, volunteering locally, or exploring a particular food scene. When you’re marketing a destination, consider creating ads around various points of view or angles, and show real travelers to increase connection and relatability. This will also make personalization tactics easier. ### Create or Highlight Experiences Travelers today are interested in unique, less-well-known destinations or special experiences in popular locations. Make sure to match your audience segments with experiences, or create targeted copy or offers to show off experiences to prospects. These unique features might invoke emotion, spark daydreams, or align with values, such as sustainability. ## 8 Best Travel Ads to Get Inspired in 2025 (and Why) There are both classic and newer travel and tourism ads you can draw inspiration from. Take a look at a few below: ### 1. Stopover for Free With Icelandair Icelandair’s “Stopover Buddy” ad campaign caught lots of travelers’ attention in 2016, when they made it easy for travelers to add a few days in Iceland to their trips with no added cost. The airline saw a 31% increase in stopover bookings, and the word of mouth and social buzz was significant. As Iceland’s national airline, Icelandair is marketing its services, but also marketing travel to the entire country. Icelandair has continued its marketing strategy more recently with ads focused on authenticity and natural beauty. Its “Easy to Stop, Hard to Leave” video picked up the stopover story and showcases friendly Icelanders and an array of gorgeous scenery, too. https://www.youtube.com/watch?v=0DmHhCXVeYc&t=69s ### 2. Live There With Airbnb Airbnb, arguably the inventor of direct home and room rentals, grew organically and has created many memorable ad campaigns since its founding in 2007. Their “Live There” campaign brought a new angle to travel, showing users experiencing destinations immersively, rather than as tourists. With a rise in remote work, they targeted users who are able to extend visits to weeks, months, or more, living in an Airbnb like it’s home. They focused on authenticity, human connection, and showed off an interesting, diverse range of destinations. https://www.youtube.com/watch?v=ddRBr2It00k ### 3. Australia — Nothing Like It Tourism Australia launched its “There’s Nothing Like Australia” advertising campaign in 2010 to show off stunning visuals of its country, home to landscapes and wildlife found nowhere else on earth. The campaign built an online platform for Australians to share travel stories and photos, then added TV spots and more. After massive brushfires dominated global news about Australia 10 years later, the board revived the campaign. “There’s Still Nothing Like Australia” built on the branding of the first campaign and reminded visitors of rebuilding and the tourism opportunities that could help do good as part of that rebuilding effort. ### 4. What Happens in Vegas When Las Vegas leaned into its “What Happens Here, Stays Here” marketing slogan, they coined a term for the city that’s used in casual conversation to this day. Previously, the city had been trying to rebrand itself as a family-friendly destination, but with this campaign, marketers leaned into Vegas’ reputation as a place for grown-up fun beyond simply gambling. They emphasized fun, freedom, and escape with the “What happens in Vegas, stays in Vegas” phrase across multiple formats and channels. The city has continued to build on that initial campaign, billing itself as a destination for dining, sports, and more. Recent numbers show an 89% favorability rating of the city among Las Vegas ad viewers and increases in visitors to more than 42 million annually. ### 5. Emirates Offers Luxury There’s a place for budget travel, but Emirates Airline launched its “Fly Better” ad campaign to show off luxury and comfort to prospects. Giving audience members permission to treat themselves can be a great piece of a personalization strategy, too — aspirational, “bucket-list” angles can spark daydreaming and long-term planning, as well as luxury add-ons to enjoy the travel itself. Emirates used TV ads to show off its in-flight services, beautiful details, and creative editing to spark imagination for prospective flyers. The campaign also served as a brand-builder for the airline. https://www.youtube.com/watch?v=BYD_zn0DGk8 ### 6. New Angle for Northern Ireland Northern Ireland, with its turbulent history, found a fresh angle for tourists with a Game of Thrones tie-in. After a 2016 storm destroyed many of the trees used in filming the HBO show, workers carved the fallen trees into doors to create a new “Journey of the Doors” experience across the country, enticing Game of Thrones fans to visit and creating a checklist-type event in the process. Like with Thomas Dambo’s Trollmap, this kind of public art can bring visitors to lesser-known or out-of-the-way destinations to boost local tourism. The photography also uses a lot of close-up imagery, which Taboola data found increases CVR by up to 74%. https://www.youtube.com/watch?v=95nrOw_mbno&t=6s ### 7. Users Generate for GoPro Along with destination-specific marketing, travel and tourism advertising also includes products and services closely tied to travel. That includes credit cards, airlines, accessories, and more, such as selfie sticks and action cameras like the GoPro. The GoPro is tailor-made for user-generated content advertising, since its audience is generally adventurous and looking for places to capture amazing photos and videos. Users were already posting heavily on social media, so GoPro tapped into that trend in various ways, encouraging users to post with their #GoPro hashtag, and offered other ways for them to submit content, including contests. They’ve created and maintained a strong brand identity along the way. ### 8. Inspiring Travelers With the World Wildlife Fund (WWF) Prospective travelers these days are often interested in sustainable travel, staying in lodging that’s carbon-neutral or offers other eco-friendly options. The World Wildlife Fund launched its recent “Adopt a Reef” campaign to educate visitors and offer a way for them to contribute to the places they’re visiting. Using high-quality imagery and gamified ways to get involved, like adopting a reef, the WWF also shows tourists the cost of tourism on coral reefs and the ways they can be involved in environmental protection. It’s a good lesson in brainstorming ways to take travel deeper for visitors and help align with values. ## Advertising Strategies for Successful Travel Marketing ### Target Intelligently Travel and tourism advertising lends itself well to targeting and personalization. Use the data you have to segment your audience by demographics, spending ability, and interests, such as whether they’re looking for honeymoon trips, family vacations, or an adult getaway. Then, create top-notch imagery and copy to get your destination in front of them on the appropriate channels, and personalize messaging accordingly. ### Use Channels Wisely Travel and tourism marketing is a key channel for influencers. Travel bloggers covering niche areas have been around for many years, and social media influencers can play a major role in creating buzz around a destination. Targeting and audience segmentation is also essential when you’re using influencers: Depending on your goals, you could engage budget-conscious travelers, seniors, parents, wine-drinkers, or many other segments. This also lends authenticity and local flavor, both important to many travelers today. ### Explore Personalization Travel and tourism offer lots of ways to reach your prospects and then tailor to them. Depending on your industry, try engaging with micro-influencers in a particular segment for higher engagement rates, or focus on a particular experience at the destination you’re promoting, then use the resulting creative accordingly. Think about geo-targeting and retargeting your ads, and get the most out of local SEO and high-quality backlinks. ## Key Takeaways Travel and tourism marketing relies heavily on high-quality imagery and audience personalization. Use the best practices you already know, but put heavy emphasis on photography and video that tells a story, evokes emotions, aligns with values, or offers a must-see or bucket-list experience. Successful creative ad campaigns can draw travelers in to picture themselves in a destination, then book the trip to reach their goals and yours. ## Frequently Asked Questions (FAQs) ### What are the key performance indicators (KPIs) you should track to measure the success of your tourism advertising campaigns? You can generally use the same KPIs for tourism and travel advertising campaigns as other marketing campaigns. Track website traffic, online reviews, revenue generation, and whatever you’ve set for conversions, such as bookings. Use ROI, CPA, CTR, CAC, and other typical marketing metrics. In addition, track the success of influencer marketing campaigns to see if those are a good fit for your overall strategy. ### What are the most effective digital marketing channels for reaching travelers? The destination you’re promoting may have different effective digital marketing channels from another destination, depending on its benefits and particular audience. Disneyland and Las Vegas, for example, likely have different segments and thus different useful tactics. Understand your audience first, then build platform-specific strategies for each of them with the right tailored copy, creative, ads, and landing pages. ### What types of visuals (photos, videos) are most engaging and inspiring for travel advertisements? While photos and videos vary greatly across destinations, consider your audience when you’re building creative. Whatever the destination, use high-quality images with close-up details that show off what users want, whether it’s time at the beach or a great craft beer scene. Tailor copy to match, and see how you might use polls, quizzes, or interactive maps to connect with prospects. If possible, offer a live Q&A session or VR opportunities. ### What are the current travel behaviors and preferences of different target audiences in 2025? There’s more emphasis than ever on experience in travel these days, which varies greatly for target audiences — a 40th-birthday girls’ trip is quite different from a family reunion to introduce grandparents to grandkids! Whatever the case, use the data you’ve gathered up front to understand who you’re marketing to and what they want. Elicit excitement, joy, nostalgia, relaxation or whatever other emotion might be tied to the trip or destination. Measure your results to further refine messaging and target prospects. --- ### Mobile Attribution: Tracking Your Audience As They Interact with Apps URL: https://www.taboola.com/marketing-hub/mobile-attribution/ Last Modified: 2025-06-30 09:19:44 You know what they say about not seeing the forest for the trees? Well, imagine a forest made up of computers, tablets, and smartphones. If you don’t see those as a larger whole, all of which a potential lead might be using, you’re really not getting the bigger picture and you might be missing out on opportunities. For marketers, tracking the activity of members of their audience is critical, and that means not just seeing what they’re doing on one device, but tracking their entire engagement across multiple devices. If you only pay attention to a lead as they browse and shop on their phone, for example, you might get an entirely wrong impression of their actual engagement and behavior, since that person might be converting left and right on a computer, with you thinking they’re not a loyal customer based on your limited tracking. Here, I’ll take a close look at mobile attribution, also known as cross-device attribution, so you can get a better picture of how you should be tracking your people. ## What Is Mobile Attribution? “Effective marketing in our mobile-first world relies on mobile attribution, which is the process of identifying what campaigns, channels, or touchpoints facilitate target user actions such as installs, purchases, or app engagements,” says Joe Giranda, director of sales and marketing for CFR Classic. “By establishing the connection and order of outcomes against their sources, marketers gain useful insights into what works and what doesn’t, which enables them to allocate budgets effectively and optimize overall spending.” Mobile attribution is the way marketers understand the journey a user takes to arrive on their site or in their app, and to follow what they do once they've landed there. Web marketing uses the well-known cookie to track a user’s journey, but the same is not possible for mobile apps, thus cross-device attribution’s importance. ## Why Is Mobile Attribution Important? “Without mobile attribution, you're wasting resources, as without it, you cannot understand what's working and what isn't,” says Mirna Huhoja-Dóczy, founder of Bits of Brand. “Whether you're driving installs, in-app actions, or purchases, attribution helps you understand how users interact with your marketing, which helps refine your strategy for better engagement and higher lifetime value." The more holistic sense of audience activity that cross-device tracking allows for lets you see what’s working best and which marketing strategies need to change. It also helps you learn more about how your audience operates, along with catching any potentially misleading markers, like a cart abandoned on mobile only to see a sale made on a computer, or an email unopened on a computer that nonetheless inspired a buy via mobile just by adding some brand awareness through the subject line. ## How Does Mobile Attribution Work? Mobile attribution works by connecting app installs and subsequent user activities within the app to the specific marketing campaign, ad, email, or other touchpoint that drove those actions. This involves tracking user interactions, such as ad clicks, likes, views, and more, and associating them with subsequent app installs or in-app events. It's like tracing a person’s journey from seeing an ad to taking action within the app. ## Types of Mobile Attribution ### First-Touch Attribution First-touch is the very first time a user clicks (or taps) on an advertising material is considered the first moment of attribution. It is essentially a proof-of-concept moment. ### Last-Click Attribution This is the assigning of significance to the last action a user completes before making the marketer’s desired conversion — a “Learn More” tap before a purchase, for example. ### Probabilistic Attribution This refers to statistical methods used to estimate attribution when actual deterministic matches are not possible. It often relies on factors like IP addresses and device types. ### Linear Attribution Linear mobile attribution is a method of assigning credit for conversions in a marketing campaign where each touchpoint along the customer journey is given equal credit. It's a multi-touch attribution model that doesn't favor any particular interaction, but rather treats all steps in the user's journey equally. ### Time-Decay Attribution This approach assigns more importance to touchpoints closer to a conversion. It assumes more recent interactions are more influential. ## Challenges with Mobile Attribution Models ### Myriad Touchpoints People often interact with a brand across various channels, platforms, and devices before making a purchase or completing a wanted conversion. Tracking these interactions and accurately attributing credit can be a challenge. ### Fraudulent Activity Malicious actors can generate fake app installs or other in-app activities that distort and inflate metrics, leading to wasted ad spend and inaccurate attribution data. ### Inconsistent Data Reporting Different attribution providers often use different methodologies and algorithms, leading to discrepancies in reported results. ## How to Choose the Best Mobile Attribution Model Choosing the best mobile attribution model depends on your specific business goals, marketing strategies, and the nature of your customer journey. Different models offer unique insights into which touchpoints drive conversions. Consider your priorities, such as brand awareness, conversions, or sales. ### Consider the Complexity of the Customer Journey More complex journeys often benefit from models that distribute credit across multiple touchpoints, like linear attribution. ### Count Your Marketing Channels If you have a wide range of channels, a model that considers all touchpoints is advised, like a position-based attribution model, which assigns credit for conversions to the first and last touchpoints, with the remaining credit distributed evenly across middle touchpoints. It's a way to understand how different marketing interactions influence a customer's journey toward a conversion. ### Measure the Length of the Sale Cycle Assuming your sales cycle is long, a model like time-decay attribution might be more suitable, giving more weight to recent touchpoints. ## What Are Common Mistakes to Avoid with Mobile Attribution? ### Relying on a Single Attribution Model Attribution models like last-click, first-click, or linear attribution, while providing insights, don't paint the whole picture of the customer journey. Using only one model can lead to misallocation of credit and a distorted view of which marketing efforts are truly driving conversions. A multifaceted approach, testing different models and combining their insights, is crucial for a more accurate understanding of the customer journey. ### Failing to Account for Web or Offline Conversions Not all conversions happen in the apps you’re tracking, or even online at all. If a business has a retail presence or deals with offline interactions, you need to track those conversions as well. Failing to do so leads to an incomplete picture of your attribution strategy. ## How to Interpret Mobile Attribution Data and Reports To effectively interpret mobile attribution data and reports, focus on understanding how user journeys are traced from ad impressions to in-app actions, like installs or purchases, and how different ad channels and creatives contribute to these conversions. ### Assisted Conversions This metric shows how often ads on one device helped drive conversions on another device. For example, a tablet assist ratio of two means that for every conversion on a tablet, two conversions on other devices were assisted by a tablet ad. ### Attribution Windows These define the time frame within which a user's actions (like an ad click) can be attributed to a specific ad campaign or channel. ### Touchpoints These are the individual points of interaction a user has with your advertising, such as seeing an ad, clicking on it, installing the app, and navigation within it. ### Channel Performance This is identifying which ad channels are driving the most installs, conversions, or revenue. ## Mobile Attribution Terminology ### Mobile Measurement Partner (MMP) A mobile measurement partner is a third-party service that helps mobile app developers and marketers track and analyze their app's performance, particularly in terms of user acquisition and engagement. They act as a central hub for data, enabling marketers to understand which marketing channels are most effective and to optimize their campaigns accordingly. ### Attribution Window An attribution window is the timeframe within which a user's interaction with an ad, like a click or view, can be linked to a conversion, like an app install or a purchase. It's the period during which an advertiser can claim credit for a conversion. This window helps determine which marketing touchpoint drove the desired action, even if there's a delay between seeing an ad and taking action. ### Click-Through Attribution (CTA) Click-through attribution refers to the tracking of direct clicks on an advertisement that lead to a conversion, such as an app install or purchase. It focuses on the specific interaction between a user clicking an ad and then taking the desired action ### View-Through Attribution View-through attribution refers to the process of attributing conversions or actions to an ad that was viewed, even if the user didn't click on it. This is important because users may see an ad, remember it later, and then take a desired action like installing an app without clicking on the ad directly. ### Deep Linking Deep linking is the practice of directing users to a specific screen or page within a mobile app, instead of the app's homepage, after clicking a link. It's essentially a hyperlink that takes users directly to desired content within the app. This can enhance user experience and engagement, improve marketing campaign performance, and allow for more accurate attribution. ### Deferred Deep Linking Deferred deep linking is a mobile attribution technique where a deep link directs users to specific in-app content, even if they haven't installed the app yet. If the app is not installed, the link redirects the user to the app store for download. Once installed, the user is automatically redirected to the designated in-app location. ### Install Referrers (on Android) In mobile attribution, install referrers (specifically, the Play Install Referrer API on Android) are a mechanism for tracking and attributing app installs to specific sources. This helps advertisers and app developers understand which campaigns or channels are driving user acquisition. ### SKAdNetwork (on iOS) SKAdNetwork (StoreKit Ad Network, or SKAN), a framework from Apple, is used for privacy-focused attribution on iOS devices. It helps advertisers measure the effectiveness of their ad campaigns without compromising user privacy. Instead of individual user data, SKAN focuses on aggregate attribution data, such as click-through attribution, conversion values, and campaign IDs. ### Privacy Manifests Privacy manifests are standardized files that app developers and software development kit (SDK) vendors use to clearly outline the data practices of an app or third-party SDK. They essentially act as a "privacy nutrition label" for an app, detailing the types of data collected, tracking domains, and APIs accessed. This information helps developers and users understand how an app handles their data, promoting transparency and trust. Privacy manifests significantly impact mobile attribution on iOS by restricting probabilistic attribution and ending fingerprinting, two common methods used for tracking users across apps. ### Fingerprinting Fingerprinting is a method of identifying users or devices without relying on unique device identifiers. It involves collecting various device characteristics (like IP address, OS type, or screen size) to create a unique profile or "fingerprint" for each device. This fingerprint is then used to match ad interactions to subsequent app installs or conversions, allowing advertisers to measure campaign performance. ## About App Install Attribution (a Significant Subset of Mobile Attribution) ### Definition of App Install Attribution and How It Works App install attribution is the process of tracking which marketing touchpoints (like ads or campaigns) led a user to install an app. It helps determine the effectiveness of different marketing channels in driving app installations, allowing advertisers to optimize their campaigns and understand which strategies are most effective. It involves tracking user journeys, properly assigning credit for actions and conversions, and analyzing data and optimizing. ### How Do You Track App Installs? To track app installs, you can use a variety of methods, including third-party attribution services, Google Play Install Referrer, Firebase, and other analytics platforms. These tools help you understand where users are coming from to download your app and how to optimize your marketing efforts. ### What Are the Best MMPs for App Install Attribution? Some of the best mobile measurement partners (MMPs) include Firebase, AppsFlyer, Adjust, Kochava, and Branch. ### How Do I Set Up App Install Attribution for My Campaigns? To set up app install attribution, enable auto-tagging in your Google Ads account and ensure your Google Play Store links include campaign details in the referrer. Also, make sure your app supports deep linking with proper UTM parameters for effective tracking and analysis. ### How Do I Measure the Success of My App Install Campaigns? To measure the success of your app install campaigns, track key metrics like the number of installs, conversion rates, user retention, and Return on Ad Spend (ROAS). Also, monitor lifetime value (LTV) and in-app events. ### What Are the Challenges of iOS 14+ (and Later) for App Install Attribution? iOS 14 and later versions of this Apple OS present several challenges for app install attribution, due to several privacy-focused changes, particularly the introduction of App Tracking Transparency (ATT). These changes restrict marketers' ability to track user activity across apps and websites, making it harder to accurately determine the source of app installs and in-app events. ### How Does SKAdNetwork Work for App Install Attribution? SKAdNetwork is Apple's privacy-centric framework for attributing app installs to specific ad campaigns. It works by using a postback system where Apple notifies ad networks and, optionally, advertisers, when an install is attributed to a specific ad campaign. This attribution happens without revealing any user-level data, ensuring user privacy. ## Key Takeaways “Mobile attribution is about figuring out how users get to your app or site and what they do once they're there,” says Huhoja-Dóczy. “Whether it's an app install, a purchase, or a simple click, it's essential for tracking the true ROI of your mobile campaigns.” Start the process by picking the right attribution model or, in most cases, the right few models. Are you looking at first-click attribution, last-click attribution, or multi-touch attribution? The right model for you will depend on your business and the specific insights you're looking for. “If you're only using one touchpoint, such as the first or last touch, you're missing out on valuable insights into the entire user journey,” Huhoja-Dóczy warns. “This is why it's important to be clear on your goals, whether you’re focusing on brand awareness, lead optimization, or something else." Choose the right tools, study the data you collect closely, and remember that many actions actually prompted in one place (in an app, for example), might take place elsewhere, such as on a website or even in a physical store, so don’t forget to consider that potential data as well. ## Frequently Asked Questions (FAQs) ### What is cohort analysis and how is it used in mobile attribution?? Cohort analysis is a method of grouping users based on shared characteristics or actions and then tracking their behavior over time to understand how different groups engage with a product or service. In mobile attribution, it's used to analyze how users acquired through specific marketing campaigns or channels perform differently over time, allowing for optimization of user acquisition efforts and retention strategies. ### What is re-engagement attribution? Re-engagement attribution is a process that identifies and attributes the source of a user's return to an app or website after a period of inactivity. It's essentially about figuring out which marketing efforts (like retargeting ads or push notifications, to name two examples) led a user back to an app or website. ### How do you track in-app events after an install? To track in-app events after an app install, you'll need to integrate an analytics platform's SDK into your app and then define the specific events you want to track. Once the SDK is integrated, you'll configure your app to log these events whenever they occur. --- ### De Minimis Crackdown: A Wake-Up Call for D2C Marketers? URL: https://www.taboola.com/marketing-hub/de-minimis/ Last Modified: 2025-05-15 05:48:23 For years, the “de minimis exemption,” which allows low value goods to enter the U.S. duty free, has made life easier for direct-to-consumer (D2C) businesses. However, with the exemption no longer applying to goods from China and Hong Kong as of May 2, 2025, the global e-commerce landscape is shifting. How will the move affect D2C brands? Let’s explore the possibilities. ## What is the De Minimis Exemption, and What Does It Mean in e-Comm? The de minimis exemption is a rule that allows low-value imported goods to enter a country without incurring customs duties or taxes. It’s become a critical component of e-commerce and cross-border trade. Since 2016, the U.S. has had a high de minimis threshold of $800. This means that shipments of $800 or less (per person, per day) are exempt from customs duties and most taxes. This exemption applies to most goods, though there are some exceptions. The de minimis exemption is based on the idea that low-value goods are usually not worth the cost or effort of processing. This is significant for the e-commerce industry, as it has enabled international sellers to keep prices low for consumers and speed up delivery times. Take D2C giant Temu, for example: Per Statista, its gross revenue grew by 239% between 2023 and 2024, with the U.S. far and away its biggest market. Much of this success can be attributed to the de minimis exemption, which has acted as a loophole, allowing Temu to bypass costly duties and other fees on most of its products, which are shipped directly to individual consumers and usually fall below the $800 de minimis threshold. ## What Changes Were Made to the de Minimis Exemption? On April 2nd, 2025, President Trump signed an executive order ending the de minimis exemption on packages originating from China and Hong Kong, effective Friday, May 2nd. According to an official White House statement, the move was made in part to “target deceptive shipping practices by Chinese-based shippers, many of whom hide illicit substances, including synthetic opioids, in low-value packages to exploit the de minimis exemption.” The move is expected to directly impact D2C businesses that import goods from China. This includes large marketplaces like Shein and Temu, whose generally lower-cost products are typically shipped to individual consumers, rather than in large quantities inside shipping containers. Packages that were previously included in the exemption will now be subject to duties of 120% or a flat fee of $100, increasing to $200 in June. This is in addition to tariffs of 145% that have already been placed on Chinese goods. It remains to be seen how Temu and Shein will respond. According to a recent CNBC report, for example, Chinese online retailer Temu had already “started adding 'import charges' of about 145% in response to President Donald Trump’s tariffs.” ## How Might This Change Affect D2C Businesses? D2C is a business model in which a company sells its product directly to consumers without going through wholesalers, retailers, or distributors. Popular U.S. D2C companies include Dollar Shave Club and eyewear brand Warby Parker. Globally, online marketplaces like Shein and Temu also operate on a D2C model, by connecting consumers worldwide directly with manufacturers, mostly based in China. The elimination of the de minimis exemption for China and Hong Kong could potentially raise costs for D2C businesses that rely on low-cost imports from those places. ## Can D2Cs Marketers Take Measures To Prepare for This Economic Shift? The elimination of the de minimis exemption may present major challenges for D2C businesses that have relied on low-cost, cross-border shipping of Chinese goods. This means marketers have a critical role to play in preparing their brands for this economic shift. This includes: ### Communicating Clearly With Customers Your messaging strategy matters more than ever. If delays or price increases are on the horizon, proactive, transparent communication can help maintain customer trust. If you communicate clearly, customers are more likely to be understanding, so marketers should work closely with their customer experience team to create messaging that explains the “why” behind any changes, and act quickly before customers feel the impact. If you deliver clear updates via email, social media, and your website, with empathy, you can preserve customer loyalty. ### Positioning Supply Chain Changes as Brand Value If your company decides to diversify its supply chain due to the de minimis changes, try to position this as a brand strength. For example, you could highlight a shift from China to Vietnam, India, or Mexico as a commitment to resilience, quality, or sustainable sourcing. Internally, it might feel like a simple logistical change, but you can make it part of your brand story with customers. ### Adjusting Pricing if Necessary Some businesses may temporarily absorb the additional import costs, but your company may eventually need to raise its prices. If so, you can soften the impact by framing the change as a value-add. Let customers know that they’re gaining, whether it’s through additional services, loyalty programs, or product bundles. ## Other De Minimis Questions Answered ### Is the $800 Threshold for Shipments From China and Hong Kong Completely Gone, or Are There Any Exceptions? Yes, as of May 2, 2025, the de minimis exemption of $800 has ended for all goods coming into the U.S. from China and Hong Kong. The exemption remains in place for other countries. ### How Will the Increased Customs Scrutiny — And Potential Delays as a Result — Affect My Shipping Times and Customer Satisfaction? According to FreightAmigo, the new U.S. Customs regulations “may lead to longer transit times for some shipments. This could impact delivery promises and customer satisfaction for e-commerce businesses.” As such, it’s critical to look for workarounds for your business. This includes resetting expectations with customers and identifying alternative supply chains. ### Will Shipping Carriers Increase Their Fees in Response to the New Tariffs and Increased Processing? Yes, several major shipping companies have begun increasing their fees in response to the tariffs and changes to de minimis rules. Per software company Alavara, FedEx increased its disbursement and duty and tax forwarding fees on May 2nd, 2025, on all packages with a customs value of $800 or less. Meanwhile, UPS has implemented a $0.29 per-pound surge fee. ### Should I Absorb the Tariff Costs To Maintain My Current Pricing? If So, How Will It Impact My Profit Margins? Individual businesses must assess their capacity to absorb increased costs from tariffs and de minimis rule changes. Without increasing prices, profit margins could tighten unless they can create other efficiencies or source lower-priced products elsewhere. ### How Will This Affect My Competitiveness Against Domestic Sellers, or Those Sourcing From Countries Still Under the de Minimis Exemption? While many variables are at play, D2Cs that can still operate under the de minimis exemption may gain a competitive advantage. This is something you’ll want to consider as you assess your supply chain options. ### Should I Consider Shifting My Sourcing or Manufacturing to Countries Unaffected by These Changes? If you foresee an increase in your cost of goods, slower delivery times, or other supply chain disruptions, you should look for alternative sourcing from countries still under the de minimis exemption. ### Is Bringing Fulfillment Operations to the U.S. Viable To Reduce Exposure to Tariffs? Moving your product fulfillment to the U.S. may or may not be viable, depending on your situation. For example, you may save money by importing in bulk and shipping domestically to U.S. customers. Once the goods are in the U.S., you can likely ensure faster delivery. However, you may face increased storage and labor costs, inventory forecasting challenges, and more complex logistics. You’ll have to consider the pros and cons and determine what’s best for your business. ## Key Takeaways The end of the de minimis exemption for Chinese and Hong Kong imports ushers in a new era of global trade regulation, specifically in the e-commerce sector. D2C businesses that rely on low-cost international shipping from Chinese manufacturers are likely to face higher duties, longer shipping times, and increased border scrutiny. To remain competitive, marketing teams must respond quickly by creating clear messaging around pricing and delivery changes, and reinforcing brand value through compelling stories. Whether your company is considering new manufacturing centers, changing fulfillment processes, or adjusting prices, your efforts will influence how customers perceive and respond to the changes. Ultimately, brands that act quickly and communicate clearly won’t just survive the shift — they’ll build strong customer loyalty and become more resilient. ## Frequently asked questions (FAQs) ### What is a "de minimis threshold"? A de minimis threshold refers to the minimum value below which a company can import goods into a country without triggering customs duties or taxes. In the U.S., the de minimis threshold has been $800 (per person, per day) since 2016, when the U.S. Congress increased it from $200. ### What is de minimis in tax reporting? When filing taxes, de minimis refers to small amounts of income or benefits that are considered too small to report or tax. Examples include small employee perks, such as holiday gifts under a specific dollar amount. Small mistakes in tax forms, such as rounding errors, may also be considered de minimis and not penalized. ### What is Section 321? Section 321 is a specific U.S. Law (Tariff Act of 1930) provision that authorizes the de minimis exemption. The two terms are related, but de minimis is used globally, while Section 321 is U.S.-specific. --- ### How Trump's Tariffs May Affect Digital Advertisers URL: https://www.taboola.com/marketing-hub/trumps-tariffs-for-digital-advertisers/ Last Modified: 2025-05-15 08:53:52 Trump’s new tariffs could hit digital advertisers hard in the short term, but will it be short-term pain for long-term gain? Here's what you can expect to happen. On April 2, 2025, following months of warnings and widespread speculation, the Trump Administration announced a 10% baseline tariff on imported goods from all countries. It also introduced additional tariffs on dozens of individual countries, but the administration placed a three-month pause on these higher rates for everyone except China a week later, in a reversal that surprised many. While it’s possible that the tariffs may have a positive effect on businesses over time, the general consensus is that very few will be left untouched in the short term, even with the baseline 10% rate. Part of the issue is that the concern around tariffs could cause changes to consumer behavior. In particular, digital advertisers could suffer from increased costs, softening consumer demand, and smaller ad budgets. ## Trump’s Tariffs Explained There is much speculation about the reasoning behind the Trump administration’s tariffs, but here’s what we know. Generally, tariffs are placed on imported goods to protect domestic industries and boost revenue. The Trump administration has suggested that tariffs are needed to address existing trade imbalances between the U.S. and dozens of countries worldwide. It is currently too early to say whether such a move will prove to be beneficial or not. ## Hardest Hit US Trading Partners Under Trump’s Tariffs While virtually all countries may feel the effects of Trump’s trade war, China has received the highest tariff rates, largely in a case of tit for tat. On April 9th, when the administration announced the 90-day pause on additional tariffs for other countries, the US instituted a 74% tariff on most Chinese imports. China retaliated almost immediately, with an 84% tariff. Things continued to escalate, and US tariffs on most Chinese goods currently sit at 145%, while China has set its reciprocal tariff rate to 125%. China’s export-dependent economy is already feeling the impact, but some pain could also be felt by US consumers soon, with various product shortages expected. According to a May 6 report from CNN, the first boats impacted by the 145% tariff rate are arriving in US ports, and “many of them are half full.” ## Which products will be affected by Trump’s Tariffs? Due to the universal 10% tariff applied to most goods, consumers may see increased prices on some products imported to the U.S. in the coming weeks. While it remains to be seen, you could, for example, find yourself paying more for items such as chocolate, coffee, an Apple iPhone, a new bike, or an imported car. ## How Digital Advertisers May Be Affected By Trump’s Tariffs (and Why) As noted earlier, the uncertainty around Trump’s newly announced tariffs may impact various industries, including digital advertising. In the short term, if the cost of imported goods rises, consumers could reduce spending. A pullback could hurt ad performance, with fewer people looking to make purchases. At the same time, a weakening global economy could force many SMBs to cut costs. With marketing and advertising budgets often among the first to be scaled back, fewer dollars will flow into digital ads. AdTech vendors could also be impacted. Many of these platforms rely on infrastructure that could become more expensive due to tariffs on countries such as China, Japan, and Taiwan. Vendors may have no choice but to pass these additional costs to advertisers, making it more expensive to run cost-efficient ad campaigns. ### SMBs SMBs rely on digital advertising to reach their customers, but many will find themselves having to pull back on already limited marketing budgets if tariffs reduce their profit margins. For example, an SMB selling imported goods is expected to see an increase in its cost of goods sold. SMBs could also see their ad performance drop if retail prices rise, as consumers will likely reduce their spending, especially on discretionary items. ### Search and Social Tariffs will directly impact the digital advertising business of tech giants such as Meta and Alphabet (Google's parent.) Both companies bring in billions from the ad space annually from global companies selling products to U.S. customers. As new tariffs increase the price of these products, consumer demand is likely to decline, leading to fewer purchases being made and less money being spent on ads. ### e-Commerce The e-commerce industry, which relies heavily on digital advertising, is highly exposed to the global tariff fallout. Thousands of online D2C businesses sell imported goods to U.S. customers, and on marketplaces like Shopify and Amazon. These sellers, along with major brands, spend millions on ads through Meta, Google, and other ad networks. This revenue could be severely impacted if e-commerce sales plummet. ### AdTech vendors It’s not just advertisers who could potentially feel the pinch from tariffs — the platforms that facilitate digital advertising may also suffer. This includes demand-side platforms (DSPs) and ad networks that sell inventory across multiple marketplaces. If the tariffs remain in place for an extended period, these companies may face higher infrastructure costs, which could ultimately lead to increased advertising prices as they pass those costs on to clients. ## How Can Businesses Prepare for These Eventualities? If faced with a weakening market and/or growing economic uncertainty, brands should focus on flexibility, efficiency, and diversification. Proactive brand responses may involve, for example, adjusting supply chains, monitoring expenses more closely, and shifting marketing budgets toward performance-driven channels which can more accurately measure return on investment. ## What Does the 90-Day Tariff Pause Mean for the Economy? Despite the 90-day reprieve on additional tariffs announced on April 9th, many economists believe a certain amount of damage has already been done, and that the global economy will continue to suffer, at least in the short term. This, they explain, is due to the U.S. signaling a significant and surprising shift in trade policy, resulting in global uncertainty. The escalation of the U.S.-China trade war only compounds the problem. If the world’s two largest economies continue to battle it out, the fallout will be felt globally, regardless of the pause. ## What Will Happen When the 90-Day Pause Ends? According to President Trump, if countries fail to make a deal with the U.S. during the pause, “then we go back to where we were … We’ll have to see what happens at that time.” In other words, without any further policy changes or exceptions, the tariffs would be expected to resume at the end of the 90-day period. On May 12, 2025, the United States and China reached a temporary trade agreement in Geneva, Switzerland, agreeing to significantly reduce tariffs for 90 days to allow further negotiations. The US reduced tariffs on Chinese goods from 145% to 30%, while China lowered its rate on US goods from 125% to 10%. While the tone from officials on both sides was positive, there is still much work to do. Mark Williams, chief Asia economist at Capital Economics, called it “a substantial de-escalation” but added that “there is no guarantee that the 90-day truce will give way to a lasting ceasefire.” For digital advertisers, the 90-day pause provides a temporary reprieve from the threat of escalating costs on imported goods, which, if it came to pass, could lead to a reduction in consumer spending, potentially impacting companies' ad budgets and ad performance. That said, it’s important to note that tariffs remain elevated over previous levels and that the exact impact of the temporary US-China agreement remains uncertain. ## Key Takeaways How long the Trump Administration’s tariffs will stand remains to be seen. Regardless, the sudden economic shift demonstrates how major geopolitical decisions can disrupt the digital advertising industry, from ad performance to budget spend. ## Frequently asked questions (FAQs) ### What is the difference between a tariff and a trade sanction? A tariff is a tax placed on imported goods, and is primarily an economic measure aimed at raising revenue or protecting domestic industries. A sanction, on the other hand, is a penalty or restriction, usually imposed on a country to achieve a political or diplomatic goal. For example, the U.S., along with many other countries, imposed economic sanctions on Russia following its invasion of Ukraine in 2022. ### Who pays for tariffs — the exporting country or U.S. consumers? Tariffs are paid by the importer when they bring a foreign good into the country. The tariff is meant to incentivize companies to seek out domestic manufacturers and suppliers rather than paying higher prices on imported products or materials. However, this is not always possible, and the tariff cost is often passed along to U.S. consumers through higher prices. ### Will the Trump tariffs hurt small businesses? The long-term impact on small businesses remains to be seen. However, in the short term, tariffs can hurt small businesses by disrupting supply chains and increasing the cost of goods, leading to reduced sales, smaller profit margins, and cash flow problems. --- ### Conversion Funnel: Importance, Stages, Benefits URL: https://www.taboola.com/marketing-hub/conversion-funnel/ Last Modified: 2025-07-22 13:24:09 Learn how conversion funnels guide customers from awareness to loyalty, and how the latest AI-powered tools can optimize each step for greater return on investment. You may have heard marketers talk about something being “top of the funnel” (TOFU) or “bottom of the funnel” (BOFU) and wondered what they were talking about. Or maybe someone told you that your team was MOFU and you weren’t sure whether you were being insulted or not. Sometimes called a conversion funnel, this term is a way of visualizing the process of bringing in potential customers and turning them into actual customers. The term has been around for ages, but it’s only very recently that AI-driven performance marketing tools became available to flesh out accurately what the funnel for a given business looks like — or should look like. The average consumer encounters a whopping 6,000-10,000 ads a day, if you combine print, digital, television, radio, physical billboards, etc.,according to 2023 data from the bestselling book, Badvertising: An Expose of Insipid, Insufferable, Ineffective Advertising, by Andrew Simms and Leo Murray. That said, how many actual products and services do you really purchase in a day? A good deal fewer than 6,000-10,000, one would hope. The marketer’s goal is to make sure that you at least notice their existence. Then, they want to retain your interest until you finally make a purchase. You might be thinking, “Well, a customer either buys the product or they don’t. That seems pretty binary. Why the need for a silly metaphor?” It’s because the decision to purchase isn’t always a straightforward one! The funnel helps marketers understand, map, and optimize the buyer's journey from discovery to decision. In this article, I’ll break down what a conversion funnel is, why it matters, and how to fine-tune every stage for maximum return on investment. ## What Is a Conversion Funnel? A conversion funnel represents the stages a potential customer goes through before completing a desired action. Just like an actual funnel, which is wide at the top and ever-narrowing toward the bottom, customer conversions start with casting a wide net — an ad placement in a YouTube video, for example — knowing that some of those customers will drop off at each step before making a purchase. The marketer’s goal is to retain as many of those wide-net leads as possible through the funnel until they make a purchase, sign up for a newsletter, or whatever the campaign’s end-goal is. On a macro level, the idea is simple: attract attention at the top, engage and educate in the middle, and convert at the bottom. ## Why Is a Conversion Funnel Important for Marketing? Are you losing potential customers five seconds into your ad? After the first email? After the second newsletter? After the customer puts the item in the cart? A well-optimized conversion funnel is the backbone of effective marketing — it transforms vague strategies into actionable insights. With today’s AI-driven technology, funnels leverage sophisticated tools like predictive audience targeting to adapt to user behavior in real time. Platforms specializing in performance marketing (such as those using proprietary first-party data) can automatically adjust bids, creatives, and targeting to maximize conversions at each stage in real time. It also helps you separate the wheat from the chaff in terms of quality leads. Here’s how: ### 1. Eliminates Guesswork with Data-Backed Insights Conversion funnels provide visibility into exactly where prospective customers lose interest. For example, your brand might discover that 75% of users abandon their cart after shipping costs are calculated for their postal code. This would signal to you that shipping is a dealbreaker for many, and that you should experiment with reduced or free shipping. Without funnel analysis, they'd be left guessing as to why sales were declining. As brand-new as AI-based marketing tools are, there is already abundant evidence of their value in using funnel analysis to make quantifiably impressive improvements in conversions, cart abandonment rate, email open rate, and any other metric or KPI ### 2. Maximizes Marketing Efficiency With digital ad costs skyrocketing — the average Facebook Ads CPM fluctuated from around $5 in 2020 to around $9 in 2024, which amounts to an 80% increase — every marketing dollar must work harder. Funnels help allocate resources strategically in the following ways: - Top-funnel: Broad awareness campaigns. - Mid-funnel: Retargeting engaged users. - Bottom-funnel: High-intent conversion pushes. Performance-focused platforms take this further by automatically shifting budgets to the highest-performing stages using real-time AI optimization. ### 3. Aligns Messaging with Buyer Intent According to a Salesforce study, 73% of consumers expect companies to understand their needs and expectations. Modern funnel tools track micro-conversions (like video watches or PDF downloads) to serve perfectly timed messaging. Here are some examples of messaging you can offer customers, based on their behavior. - TOFU: Educational blog posts, such as “10 Mistakes New Pet Owners Make.” - MOFU: Product comparison guides, such as “The Best Humane Mouse Traps, According to Our Testers.” - BOFU: Examples include limited-time offers, “Free sampling of brandy-filled chocolates with purchase,” buy-one-get-one free. This level of personalization at scale is possible with advanced AI platforms that leverage machine learning to interpret user behaviors. ## Levels of a Conversion Funnel ### Top of Funnel (TOFU): Creating Awareness The goal at the top of funnel is to cast a wide net to attract potential customers. Proven tactics include: - SEO-optimized content in response to user search prompts (for example, "eco-friendly disposable picnic utensils"). - Native advertising that blends into publisher content. - Video ads on platforms like YouTube, where completion rates for bumper videos (non-skippable) can be as high as 70%, per a Mutesix study. ### Middle of Funnel (MOFU): The Consideration Stage At this stage, your goal is to maintain user interest, and make the case for your product or service’s value add. Here are some possible strategies: - Targeted email campaigns: Personalized emails deliver 6x higher transaction rates, 29% higher open rates, and 41% higher click-through rates. - Webinars: Three quarters of surveyed marketers say webinars are among their chief sources of qualified leads. - Retargeting ads: Achieved with dynamic creative optimization, using the latest AI marketing tools. - Optimize for engagement: Some platforms now offer "Optimize for Engagement" settings where AI automatically finds users most likely to take mid-funnel actions (like watching 90% of a demo video). ### Bottom of Funnel (BOFU): The Decision Stage This is your last chance to convert users and overcome whatever final objections they have to entering their credit card number. AI-powered platforms can identify quality leads (such as users who viewed a certain item more than three times) and automatically increase bid weights for these high-intent users. Some strategies you can try include: - Language suggesting urgency or time-sensitivity: Banners like "Only 3 left at this price!" can increase conversions. - Customer reviews: Studies show that nowadays, users trust product reviews from strangers. According to a study by Social Pilot, close to 85% of consumers trust online business reviews as much as personal recommendations from other people. TrustedSite, a company that provides trust badges, claims that their trust badges increase sales by up to 30%. - One-click Checkout: According to research published by Cornell University in 2023, after signing up for an online retailer’s “one-click” checkout service, customers over time increased their spending by an average of 28.5% from previous buying levels. As Forbes points out, the cart abandon rate for an item purchased via one-click is 0%—because there is no cart to begin with. ## Four Stages of a Conversion Funnel It depends on who you ask, but a typical funnel contains the following elements: ### 1. Raising Awareness Typically, this is done by means of social ads, organic search, and PR. ### 2. Intent Effective methods include newsletter signups and email campaigns. ### 3. Conversion This is the Last Stand, and includes checkout pages and customer rep sales calls. ### 4. Loyalty Surely you’ve heard that it’s much more expensive to gain a new customer than keep an old one! ## Three Benefits of a Conversion Funnel ### 1. Higher Conversion Rates Businesses optimizing funnels see more conversions. A survey by Iterable showed that AI-driven marketing tools, which allow at-a-glance funnel analysis, led to a 15% lift in converting buyers to an active state and a 72% lift in converting sellers to an active state. A popular tool for funnel analysis is Google Analytics, which has abundant success stories and case studies. a Chilean grocery shopping app called Líder, used the GA platform for funnel analysis. The conversion rate for the “Likely 7-day Purchasers” audience also increased from 0.3% to 5.4%—that’s an 18-fold increase. ### 2. Reduced Customer Acquisition Costs (CAC) By fixing funnel “leaks,” brands decrease CAC, the amount of money it takes to acquire a new customer. ### 3. Improved Customer Lifetime Value (CLV) Post-purchase emails and rewards programs can significantly increase CLV. According to Klaviyo data, post-purchase emails see open rates that are almost 17% higher than the average email automation. For example, Starbucks' loyalty program is a textbook success case. According to an analyst call in 2022, Starbucks reward members account for 53% of their U.S. revenue. With Sephora, that figure is 80%. ## Challenges Advertisers Face with Conversion Funnel Marketing ### Fragmented Data Only 31% of marketers are fully satisfied with their ability to unify customer data sources across channels, per Salesforce’s 2025 ninth State of Marketing Report. That means 69% of marketers find fragmented data to be a problem. The good news: What was once a massive headache is now much easier with AI-driven platforms, many of which offer end-to-end analytics to consolidate data you get from all your sources. ### Creative Fatigue According to numerous studies, advertising “irritation” reduces the overall effectiveness of ads. AI tools can automatically refresh creatives and make A/B testing a breeze. ### Platform Limitations Given significant recent changes in some of the most widely used social media platforms, some companies are reporting reduced return on investment (ROI) on their social media campaigns. Diversifying with high-intent channels powered by performance AI can help you identify more effective platforms, or simply optimize the ones you are currently using. AI tools are nothing if not adaptable! ### Challenges With Optimization #### a) Prioritizing When KPIs Clash Sometimes, in improving one aspect of the funnel, you jeopardize a different aspect of the funnel. For example, one of your mid-funnel KPIs might be session duration — the amount of time a user spends on a page. But, sometimes, a user might stop scrolling down after a few seconds simply because they got the information they wanted and are ready to make a purchase. How can you tell what the user's intent is? No, you can’t read their minds, but you can come pretty close with the latest AI-driven marketing tools. These will analyze multiple goals and multiple decision trees simultaneously in a way that a human cannot, and suggest a course of action. It can give quantitative and qualitative recommendations for where a long session duration is a good thing from a conversion point of view, where it’s bad, and where it’s irrelevant. #### b) Time and Money Until just a few years ago, only very large companies could afford an entire data science team. This put small and medium enterprises at a huge disadvantage in terms of funnel analytics. Happily for small businesses, the latest developments in AI have made this disparity a thing of the past. As is the trend with all technology, the price point for AI-driven marketing tools has gone down while functionality has gone up, leading to the democratization of conversion funnel optimization. #### c) Nimbleness Let’s say you’ve confidently identified weak areas in your funnel that could stand improvement. Now what? Theory is easy, practice is harder: How do you implement changes across your company’s entire marketing ecosystem, including social, email campaigns, CTA buttons, etc.? Here again, AI tools can respond to new data in real time, whether it’s data you input or information it gleaned from customer behavior. It can automatically adapt your campaigns without constant human input. ## How to Measure and Analyze Your Conversion Funnel’s Performance A conversion funnel is only as effective as your ability to track it. To truly understand where users are dropping off — or where they’re converting like clockwork — you need to monitor performance at every stage with both quantitative and behavioral data. ### Define Clear Funnel Stages and KPIs Start by breaking down your funnel into its major stages: Awareness, Consideration, Conversion, and Retention. Assign key performance indicators (KPIs) to each stage. For example: - Awareness: Measure in terms of impressions, click-through rate (CTR). - Consideration: Measure in terms of time on page, return visits. - Conversion: Measure in terms of purchase rate, form fills. - Retention: Measure in terms of repeat purchase rate, email engagement. This will give you a benchmark against which to measure growth or friction. ### Segment Your Audience Performance often varies by audience segment. Analyze behavior by demographics, device type, referral source, or campaign to uncover hidden trends. For instance, mobile users may abandon more quickly if your site loads too slowly or works well on Chrome but not Safari. Conversion Funnel Optimization Once you’ve measured your funnel, the next step is to refine it, eliminating friction and doubling down on what works. Optimization isn’t a one-time fix; it’s a continual process of testing, learning, and evolving. In recent years, marketers have begun to say “it’s not a funnel, it’s a pretzel,” to indicate the need to constantly find re-entry points for customers to stay in your site’s ecosystem. ### Identify Drop-Off Points and Reduce Friction Use heatmaps, session recordings, and A/B testing to uncover usability issues. Are users hesitating at a form field? Are calls to action (CTAs) below the fold? Seemingly minor changes — like reducing the number of required form fields — can make a big difference. For example, personalized CTA performs a whopping 202% better than generic CTAs, according to a HubSpot study, and a red CTA button outperforms a green one. Go figure. ### Align Messaging Across Funnel Stages Ensure that your ad creatives, landing pages, and email follow-ups all reflect the same messaging and value proposition. Disjointed messaging can erode trust and drive drop-offs, especially during the consideration phase. ## The Best Tools for Conversion Funnel Optimization With so many analytics and marketing tools available, it’s critical to build a tech stack that not only tracks performance but actively improves it. ### Analytics and Funnel Tracking Google Analytics 4, Mixpanel, and Heap offer powerful funnel tracking features with user segmentation and behavioral flow visualization. These tools are essential for spotting leaks in your funnel and identifying your top conversion paths. ### A/B Testing and Personalization Platforms like Optimizely and Convert allow marketers to run controlled experiments to test headlines, layouts, CTAs, and more. ### Predictive Insights and Campaign Optimization For advertisers running display ads, realize is a standout. It uses machine learning to identify which creative elements, headlines, and audience segments are performing best across the entire conversion funnel. Realize doesn’t just track engagement — it helps predict which assets will drive deeper funnel activity, enabling smarter budget allocation and faster optimization cycles. ## Key Takeaways By understanding and optimizing your conversion funnel, you can significantly improve marketing ROI, customer satisfaction, and long-term brand loyalty. Each stage of the funnel requires tailored messaging and strategies. Measuring drop-off points and conversion rates helps optimize performance. The latest AI-driven tools can provide actionable insights across the entire funnel. ## Frequently Asked Questions (FAQs) ### Do e-commerce sales have a different conversion funnel optimization? Yes, e-commerce funnels often involve product pages, cart interactions, and checkout flows that require optimization at each step. ### What are the best ways to nurture leads in MOFU? Email campaigns, webinars, and targeted content are all effective MOFU strategies. ### How do you reduce drop-off in the final funnel stage? Use “trust signals” (testimonials, guarantees), simplify checkout flows, and offer limited-time deals. ### What are some A/B testing ideas for funnel optimization? Some effective A/B testing ideas include testing landing page layouts, button colors, CTA wording, and headlines. ### What are some UX mistakes that break conversion funnels? Confusing navigation, slow load times, cluttered design, and inconsistent messaging can all increase bounce and abandonment rates. --- ### The Direct-to-consumer (D2C) Business Model: What It Is, How It Works URL: https://www.taboola.com/marketing-hub/d2c/ Last Modified: 2026-04-12 07:29:12 A D2C business model enables brands to sell their products and services directly to consumers without middlemen. Here’s how the model works and why it matters. Direct-to-consumer (D2C) is a business model where manufacturers and brands sell their wares directly to consumers. The model cuts out the usual middlemen such as distributors, wholesalers, and retailers, allowing companies to maintain lower prices than traditional consumer brands while giving them complete control over product development, marketing, and distribution. ## What Is the D2C Business Model? While traditional retailers follow a set model for distribution, D2C brands can experiment with diverse distribution methods. From direct shipping to strategic collaborations with brick-and-mortar retailers and pop-up shops, D2C companies have full control over the entire customer journey. Most D2C brands leverage digital channels, including owned websites, social platforms, and mobile apps, to directly reach and engage their target audience. Increasingly, brands are partnering with popular social influencers to spread the word. Here’s a look at some successful D2C brands: - Warby Parker. - Dollar Shave Club. - Hims & Hers. - BarkBox. - Away. - Glossier. Large traditional retailers such as Nike, Adidas, and Lululemon are embracing D2C marketing strategies to build better customer relationships and reduce their dependence on third-party retailers. In 2023, established brands drove nearly $135 billion in D2C sales, and then another $160 billion in 2024, according to data from Statista. This is forecasted to jump to $186 billion in 2025. ## Benefits of D2C ### Direct Customer Relationships Perhaps the best thing about skipping the intermediaries in a D2C model is forging direct relationships with consumers. Research shows that 82% of manufacturers report D2C sales improving customer relationships and experiences. This helps establish meaningful connections with customers, gather real-time feedback about your product, and create personalized experiences that resonate with your audience. ### Higher Profit Margins As Salesforce points out, selling products through a D2C channel allows brands to take the entire amount paid by the customer, so the profits end up being higher than with wholesale distribution. Without those retailer markups, D2C brands can offer competitive pricing and stay profitable — or reinvest those savings into product quality, customer experience, or marketing initiatives. ### Total Brand Control Running a direct-to-consumer business gives brands complete control over their messaging, visual identity, and overall customer experience. This control extends to product presentation, packaging, customer service interactions, and post-purchase engagement. With this level of oversight, a brand directly manages all critical touchpoints that influence brand perception and customer loyalty in ways that aren’t possible when you work with third-party retailers. ### Valuable First-Party Data D2C operations provide unfettered access to first-party customer data across the entire purchase journey. Having this enviable data access gives brands the upper hand, allowing them to stay nimble, adapt to market and customer trends, and to improve their business model. It also enables them to create customized experiences, optimize marketing campaigns, inform product development decisions, and build stronger customer relationships based on real, unfiltered insights — a key competitive advantage. ## Drawbacks of D2C ### Higher Customer Acquisition Costs D2C brands face a major challenge: Higher costs to acquire new customers. Quora business research reveals that D2C advertisers are spending increasing shares of their marketing on social media platforms in order to reach potential customers directly. In fact, approximately 70% of client spend goes toward Meta platforms (Facebook and Instagram) and 30% to Google. Without established distribution networks and retail foot traffic to drive transactions, D2C companies have no choice but to invest heavily in digital marketing, content creation, and brand awareness campaigns — costs that continue to rise as competition heats up in saturated markets to vie for consumers’ short attention spans. ### Limited Physical Presence Despite explosive growth in online shopping, many consumers still value in-person shopping experiences, putting some D2C brands at a disadvantage. According to HubSpot, D2C brands that operate in digital-first spaces may struggle to provide tangible, tactile experiences that customers prefer. To address this limitation, many successful D2C brands look to omnichannel strategies that blend online and offline touchpoints, including pop-up shops, brand showrooms, and strategic retail partnerships, especially for products where physical interaction influences purchase decisions. ### Supply Chain and Fulfillment Barriers Managing end-to-end supply chain operations can be tough for D2C brands, and a brand that previously had a distribution model and switched to D2C might struggle to sell directly to its target customer. Unlike wholesale models that ship in bulk to retailers, D2C operations require managing individual orders, inventory forecasting, warehousing logistics, and last-mile delivery. These challenges can impact delivery times, customer satisfaction, operational efficiency, and operational costs. ### Customer Service Demands The direct relationship D2C brands have with consumers means the brand must handle customer service inquiries and issues directly. This responsibility can strain resources during periods of high demand or when trying to scale the business rapidly. Unlike B2C companies that can leverage a marketplace’s customer support infrastructure during demand spikes, D2C brands must build, maintain, and scale their own customer service operations. This effort requires significant investment in both technology and staffing. ## What Is D2C Marketing? D2C marketing involves the strategies and tactics brands use to connect directly with consumers, build brand awareness, and drive sales without traditional retail support. Unlike conventional marketing approaches that focus on trade marketing or retailer relationships, D2C marketing prioritizes direct consumer outreach, primarily through digital channels. D2C marketing tactics include: - Digital-first owned channels. - Content marketing. - Community building. - Performance marketing. - Personalization. ## Successful D2C Marketing Strategies ### Influencer Marketing and User-Generated Content Partnering with the right influencers is like finding gold for D2C brands — they build instant credibility and expand your reach to engaged audiences who actually trust recommendations. Just look at Glossier, which transformed from Emily Weiss' beauty blog into a global powerhouse makeup brand, largely through strategic influencer relationships and encouraging customers to share their experiences. User-generated content creates authentic social proof that's far more convincing than any polished marketing campaign could ever be. It’s also more cost-effective, but the bigger the influencer, the more you’ll pay. ### Omnichannel Experience The most successful D2C brands don't just live online, they create seamless experiences across digital and physical touchpoints to meet customers wherever they prefer to shop. Warby Parker brilliantly evolved from online-only eyewear to operating hundreds of physical showrooms, complemented by innovative digital tools like virtual try-on features. As well as increasing convenience, this omnichannel approach creates consistent brand interactions that make customers feel understood, regardless of how they choose to engage. ### Data-Driven Personalization When you have direct access to customer data, you'd be crazy not to use it to create hyper-personalized experiences that delight customers. Stitch Fix took this approach, analyzing customer preferences, purchase history, and body measurements to create personalized styling recommendations that feel like having a personal shopper in the comfort of your home. The more relevant your interactions, the stronger your customer relationships become — it's that simple. ### Context Relevance Another way to drive relevance is by matching the advertising format to the user’s environment and device. As Realize advertising sales manager Ari Del Rosario points out, high-end electronics brands are finding success by moving away from the "category clutter" of social media. By using vertical video ads on the open web, these brands reach consumers while they are engaged with premium editorial content—like a tech review or news article. This strategy creates a "mobile awareness, desktop conversion" funnel: capturing interest with immersive vertical video on mobile, then retargeting those high-intent users on desktop when they are ready to make a high-ticket purchase. By meeting the customer where they are with the right content for the right device, brands can break through ad fatigue and build the technical trust necessary for a sale. ### Subscription Models and Loyalty Programs Implementing subscription options is basically getting customers to sign up for a steady relationship with your brand, instead of constantly chasing new first dates. Companies like Dollar Shave Club cracked this code by making recurring purchases seamless and beneficial for both parties — customers get convenience, while businesses enjoy stable revenue. Complementing this approach with thoughtful loyalty programs and helpful, engaging content that actually rewards repeat business keeps customers satisfied and entices them to come back for more. ## D2C Marketing Trends to Watch in 2025 ### Social Commerce Integration Social commerce is where consumers buy from D2C brands directly off social platforms. According to eMarketer, TikTok Shop’s gross sales have crossed a whopping $1 billion monthly (yes, monthly) since July 2024. What’s more, about half of social shoppers on the platform buy something there at least once a month. That’s more frequent than Facebook, Instagram, or Pinterest, eMarketer found. ### Focus on Sustainability and Values Modern consumers are more intentional about making purchase decisions that align with their personal values, particularly when it comes to sustainability and social responsibility. D2C brands that authentically demonstrate commitment to these values can differentiate themselves in competitive markets and build deeper connections with like-minded consumers, who increasingly expect brands to take meaningful stands on social and environmental issues. ### Flexible Payment Options Digital payment services like Buy Now, Pay Later are gaining popularity, with the market expected to reach $39 billion by 2030, growing annually by a rate of 26% until then. Meanwhile, mobile wallet transactions are expected to surpass $10 trillion globally this year, up from $5.5 trillion in 2020, according to a Juniper Research study. These payment advancements are critical for D2C brands selling higher-priced products. By offering flexible, diverse payment options that accommodate different consumer preferences, D2C brands can boost their conversion rates and expand their customer base. ### AI and Automation in Customer Experience Artificial intelligence tools and automation technologies are changing how D2C brands engage with customers across the entire purchase journey. From predictive recommendations to smart warehouse solutions, D2C brands are seeing positive returns in cost optimization and creating positive customer experiences. ## Key Takeaways The D2C model provides brands with direct control over customer relationships, data collection, marketing, and the overall brand experience, with 82% of manufacturers reporting improved customer relationships through direct selling. Successful D2C strategies leverage first-party data for hyper-personalization, with effective personalization reducing marketing costs by 10-20% while boosting conversion rates by 10-15%, according to MoEngage. ## Frequently Asked Questions (FAQs) ### What is D2C vs. B2C? In a D2C (direct-to-consumer) model, manufacturers and brands sell products directly to consumers through owned channels without any third-party middlemen. In contrast, the B2C (business-to-consumer) model relies on selling to consumers through intermediaries such as retailers, marketplaces, or other third-party channels. ### Is Amazon a D2C or B2C? Amazon operates as both a B2C and a marketplace platform, but isn’t strictly a D2C business. As a B2C platform, Amazon is an intermediary marketplace where thousands of brands sell to consumers through the Amazon website. However, Amazon has its own private label brands (like Amazon Basics) that it also sells directly to consumers on its platform. It directly develops, markets, and sells its products to consumers. Many D2C brands also use Amazon as an additional sales channel, creating a hybrid approach that blurs the usual boundaries between these two business models. ### Is D2C the same as dropshipping? No, D2C and dropshipping are completely different business models with separate approaches to inventory and brand control. While D2C brands design, manufacture, and sell their own unique products while keeping control over the entire supply chain, dropshipping businesses don’t create or hold inventory, but instead serve as marketing intermediaries who forward customer orders to third-party suppliers. --- ### Lead Nurturing: How It Works, Benefits, Measurements URL: https://www.taboola.com/marketing-hub/lead-nurturing/ Last Modified: 2025-06-03 12:12:32 Lead nurturing builds relationships with potential customers by sharing relevant content and guiding them through the sales funnel. Here's how it's done. Leads are potential customers that could become customers or clients down the road. Lead nurturing involves building relationships with these potential customers, spending time, energy, and resources to increase trust so when the time is right, they turn to you for your product or service. Getting someone’s contact is “just the beginning,” says Lindsay Marty, founder and CEO of Above the Bar Marketing. “Lead nurturing is what really drives conversions. The challenge is helping that person feel understood, supported, and ready to take action. Every follow-up should make things clearer for the lead and move them closer to becoming a client.” ## What Is Lead Nurturing? Lead nurturing is the process of developing and maintaining relationships with potential customers, or leads, at every stage of the sales funnel, with the goal of guiding them toward making a purchase decision. Instead of pushing for an immediate sale, lead nurturing focuses on providing relevant information, maintaining visibility, and identifying when they're ready to buy. To cultivate those leads, it’s important for marketing teams to develop a concrete strategy for each stage. Here are the steps for doing it effectively. ## How Lead Nurturing Works: Four Stages From the first minute your business connects with a lead, to the moment they become a loyal customer, lead nurturing involves specific steps for success. By completing each of these, you will have a better conversion rate in time, and customers that trust you even more when they do convert. “Acquiring leads is one thing,” confirms Nicolas De Resbecq, marketing expert and CRO specialist at Oppizi. “Engaging them and leading them to conversion is where the real battle lies. In B2B, the lead-nurturing process tends to be longer: Individuals are making a business decision, so there is more information and assurance that is needed.” Here are the steps to convert your leads into loyal customers: ### Step 1: Build Awareness of Your Brand People can’t choose your business if they don’t know you exist. So, the first stage of lead nurturing is getting on their screens through ads, in their inboxes with newsletters or outreach emails, or physically in front of them. Only once a lead becomes aware of your business can they move to the next stage of becoming more interested in what you have to offer. ### Step 2: Pique Interest Now, it’s time to stand out against the competition by using multiple touchpoints to show leads just how beneficial your products or services can be to their business or life. This outreach doesn’t have to be direct or intense, but rather a process of trust-building as you pique the interest of your leads, nudging them towards a more serious consideration phase. You might achieve this through contests, engaging images or posts, ads that grab their attention, and other touchpoints that make them look twice and reconsider what you’re about, and what you have to offer. ### Step 3: Be Valuable In the Consideration and Evaluation Phase Leads may well look to you for advice in this stage, presenting you with an opportunity to add value, serving as a wayfinder or teacher as they make their decision. During the evaluation stage, you can provide helpful and in-depth information in the form of blog posts, whitepapers, video explainers, and more. ### Step 4: Time to Convert Make your final outreach to your leads by proposing a clear and irresistible offer that leads to conversion. During the conversion stage, you can be helpful by ensuring there are no hiccups that cause potential customers to reconsider as they’re onboarding or purchasing, e.g., due to technical or logistical issues with checkout. “A good plan has to span the entire path from first contact to purchase,” says De Resbecq. “That involves defining what content or message aligns with each step, determining what group controls each piece, and having discernible measures to monitor effectiveness. If conversions are the desired end result, don’t monitor just opens or clicks, measure how each touch contributes to the point of purchase.” So, instead of guessing what you need to further develop at each stage, look to the data — or hire additional resources and tools to generate the data — to figure out what touchpoints are working and which need to be improved. ## Lead Nurturing Benefits ### Higher Conversion Rates You will see more leads converting with much less effort on your part if you have an effective lead-nurturing system and strategy. This ultimately leads to more profit and a more streamlined marketing system. ### Increased Brand Loyalty If you’re taking your time on initial touchpoints, you will have customers that trust you long before they even sign their first contract or make their first purchase. This pays off when they are still loyal customers a decade later because of the work you did up front. ### More Referrals Those loyal customers are also more likely to tell their friends about your product or service. They’ve learned to trust you in the initial stages of lead nurturing, and are more likely to mention you to others sooner. ### Fewer Lost Leads Early In The Marketing Cycle Some marketing teams depend too much on expensive touchpoints, and might lose their leads early in the process. Also, those who skip steps might lose potential customers before they have a chance to learn the potential benefits of your brand. ## What Do You Need to Begin Nurturing Leads? To begin nurturing leads, you need a system for finding and identifying the most optimal leads for your offers. This can be a technology tool or service that does it for you (more on that below), a skilled marketing team, or a DIY effort as a marketing manager. You will also likely need a budget for building meaningful touchpoints, whether it’s for a social media manager or new campaign, or to purchase new advertising. ## What are Effective Lead Nurturing Strategies for B2B? ### Education Sometimes, leads don’t even know why they need your product or service. You can answer this for them through education. “Within the B2B space, I always aim to create value from the onset by providing educational materials, presenting case studies, and delivering a personal outreach for every lead,” says Kyle Sobko, CEO of SonderCare. “This helps create trust gradually as we guide our leads through the sales funnel.” B2B marketing might involve more specialized educational materials than B2C, so investing in content strategy and high-quality content such as whitepapers helps build trust in the lead-nurturing stages. ### Email Sequencing A well-timed email campaign can convert leads at multiple points in the process. “Email sequences based upon where a person is in the funnel are effective,” says De Resbecq. “If a prospect viewed a whitepaper, the subsequent follow-ups should dive deeper into that same subject, not a generic product promotion. Timing also works in your favor: You don’t want to send too much, but you also don’t want to let the lead drop cold. A CRM (customer relationship management system) that includes automating based upon lead action will make this work.” ### Personalization Nobody wants to feel like they’re getting pushed through a generic process just for their money. So, don’t rush it, and personalize the touchpoints wherever possible. “In B2B, the time frame is longer, but the quantity is larger. The technique that works here is personalization at scale,” De Resbecq says. “Product suggestions based on browsing or retargeting with ads that feature abandoned items can push a person to do something. The key is to make them think that the message is personal.” ### Lead Scoring Don’t spend all of your time and resources on a lead that might not convert. De Resbecq calls this an “essential element” of lead nurturing. “You need to direct effort where there is actual potential,” he says. “A basic model would have points awarded based on email opens, clicks, visits to the site, and filled-in forms. You'll have to refine it over time, based upon what actions actually result in conversions within your business.” Marty adds that, “Clear actions are more important than flashy words. We use lead scoring like this: If they open emails, download a guide, or attend a webinar, they earn points. Once a lead hits a certain score, our intake team gets a message to call them. This system helps us focus on the most interested leads without wasting anyone’s time.” ### Timing Just like timing that perfect text to follow up after a first date, you have to pay attention to the ideal timing to reach out to leads for the next step. Keep in mind that B2B marketing can sometimes take months, or even years, compared to B2C, which can be days or even minutes in some cases. So, consider those longer sales cycles when you make plans. “While many marketers rely on automated workflows, we have found that a more tailored approach to timing, based on lead behavior, is far more effective,” says Sean Clancy, managing director of SEO Gold Coast. “If a lead has interacted with specific content, we delay the follow-up and send them a more advanced piece of content, like a webinar, rather than an immediate sales offer. This strategy goes beyond typical lead scoring and dives deeper into understanding the mental state and readiness of each individual lead.” ## What are Effective Lead Nurturing Strategies for B2C? Unlike B2B marketing, lead nurturing in B2C can mean quicker cycles, more impulsive decision-making, and behavior-driven strategy. However, it differs based on industry and company. Here’s what to consider if you’re focusing on B2C: ### Clue In to Shopping Behaviors B2C consumers often respond to quicker, emotional, or entertainment-based marketing efforts to make their decisions, as opposed to some of the longer and more analytical aspects of B2B marketing. So, research and lean into those B2C consumer shopping behaviors. Maggie Swift, co-founder and CEO of Unframed Digital, says that, “For B2C, we focus more on visual retargeting, interactive lookbooks, and personalized sequences tied to shopping behavior — like a cart view that leads to, ‘Which Pull Style Matches Your Cabinet Finish?’” ### Use Your Graphics It’s time to get visual. B2C consumers need something that will make them stop scrolling and look again. “It's all about eye-catching visuals, just-right SMS timing, and tapping into emotion,” says Matt Bowman, CEO and founder of Thrive Local. ### Tell the Story Clue your leads into the inner workings and results of your company, by getting better at storytelling. “In B2C, I've noticed trust taking center stage,” says Tomas Melian, SVP of marketing at DiabetesTeam. “People managing health conditions don’t respond to generic funnels. When we slowed things down and used email storytelling — patient voices, actionable advice, and light CTAs (calls to action) — we saw a 2.5x increase in sign-ups over our old, ‘faster’ funnel.” ## Advanced Lead Nurturing Practices ### Build a Genuine Relationship Advanced B2B marketers know how to build genuine relationships, where you care about the client’s success even if, in the short-term, it doesn’t mean more money for you. “One area of lead nurturing that’s often overlooked is the emotional connection you build with leads,” Clancy says. “B2B sales are often viewed as purely transactional, but creating a genuine relationship through storytelling can be a powerful tool. We’ve found success by focusing on the journey behind our clients' success stories.” ### Focus on Long-Term Outcomes Another advanced nurturing practice is to analyze data not just in the immediate months, but years down the road, to see how outcomes over time show your ability to play the long game. “Long-term, we stay in touch with simple monthly newsletters. We include helpful tips, short updates, and reminders for common legal questions,” says Marty. “One client sends a quarterly checklist — like what to update after a life event — and it keeps past customers engaged and even brings in referrals.” ## Automating Lead-Nurturing Processes ### Drive Cold Audiences to Discover Your Brand On Their Own First, to automate lead generation, you can use performance software to get specific content in front of the right decision-makers, utilizing a performance AI based on years of first party data to target potential customers and optimize creative in real time. ### Email Sequencing Use email trigger tools to send emails at various cadences, depending on the client and based on the specific stage they’re at. Track your data — subscribes and unsubscribes, open rates, click through rates, conversion rates, etc. — and make adjustments from there. ### Lead Segmentation You want to stay organized with your cold and warm leads before starting the lead-nurturing process. You can choose an automated lead segmentation tool or do it yourself. Organize the leads by their size, potential budgets, behavior, touchpoints they’ve interacted with, job titles, industries, and other notable interests. Some tools allow you to automate assigning leads to these various categories for easier organization and engagement. ### Retargeting Once your leads have interacted with some of your content, you retarget warm leads that started to interact with a touchpoint but didn’t convert yet. Helping them through the consideration phase with retargeting, such as sharing lead magnets like webinars, case studies, eBooks, articles, and other content, can give them more information to make their decision. ### Help Them Make It Over The Line In the final stages of lead nurturing, you can help close the deal through automated SMS/email reminders for booking calls or making purchases. You can use calendar-based tools like Calendly to automate the process of reminding them and getting them booked, or ActiveCampaign for finalizing those last steps. You can also consider sending personalized materials such as a case study with clear results, to help make the decision. Finally, you can automate next-step triggers if someone downloads a rate sheet, pricing guide, or offer. You can also reengage them if they stall in the process through reminder emails or texts after a delay. ## Personalization and Customization Tactics ### Analyze the Behavior of B2B Leads in Your Industry When and where are your clients most likely to engage with your material? What motivates them? If you aren’t sure, do some research and surveying before making expensive guesses. Consider what’s next with each missed touchpoint attempt — if they didn’t open an email, consider moving to SMS, changing the subject line, or going another route. If they check out your pricing information, send a follow-up, personalized proposal, quote, or email to see what questions they have or to share a competitor analysis to show your own competitive pricing. ### Mix Behavior with Industry “Mix what they do online (clicking links, visiting pages) with who they are (their job, their industry),” Bowman says. “Smart marketing includes sending messages when people take specific actions, using several communication channels together, asking for information bit by bit, and making different content for different audience groups.” ### Use First-Party Data to Focus on Their Needs You can use first-party data responsibly to help inform the lead-nurturing process. Integrate this data into your CRM for easier access and management. Ensure leads are seeing buyer stories and case studies that accurately show similar narratives to their own problem, and demonstrate the solution you provide. ## How to Measure and Optimize Your Lead-Nurturing Process So, what data should you be watching? To truly measure and optimize your lead-nurturing process, you need a strategy built from that data. Here are some common concerns to address: ### How Do I Measure the ROI of My Lead Nurturing Efforts? First, determine what types of leads you will be measuring. For example, you may want to distinguish between your ROI for marketing qualified leads (MQL) versus sales qualified leads (SQL) rather than mixing results together. You also might consider measurements around the average time to conversion rate, the size and impact of the closed deals, and the resources necessary to complete the strategy versus what you are getting in return. ### What Are Common KPIs for Lead Nurturing Campaigns? The KPIs you need to measure are your click rates, email unsubscribe and subscription rates, total costs with each stage of the nurturing process, advertisement costs versus return on investment, and resources spent on lead nurturing content, such as eBooks or newsletters, compared to their engagement rates. You can look at other common KPIs, too, such as bounce rates, to determine where you might be losing opportunities with leads. ### How Can I A/B Test My Lead Nurturing Strategies? A/B strategies involve testing two different processes against each other to determine which leads to a higher conversion rate. Don’t introduce too many different variables in each test so you can more easily understand what’s working and what’s not. Also, conduct A/B tests over a reasonable period of time. ## Lead-Nurturing Tools and Technologies You don’t have to go through the lead-nurturing process alone — smart marketing teams rely on tools and technology to make it easier. Consider these: ### HubSpot Swift uses Hubspot for lead segmentation and scoring. Sobko also likes how the platform helps him watch the lead journey while he navigates a variety of conversations: “I can simply automate an email, track for opened and unopened emails, and complete all of this in a matter of seconds,” he says. ### Klaviyo For e-commerce automations, Swift uses Klaviyo paired with Hubspot. Other marketers appreciate Klaviyo’s lead-nurturing tools, which are focused on segmentation and targeting, as well as their automated workflows, real-time tracking of behavioral triggers, and other tools. ### Hotjar Along with UTM-tagged click data, Swift says Hotjar helps “optimize which mid-funnel pieces actually move leads forward.” ### Realize Performance advertising platforms can help you create meaningful touchpoints at various phases of lead nurturing, via tools like its GenAI Admaker and Social Importer. It also allows you to ensure you show ads to the right audience at the right stage, thanks to its unique matchmaking AI that targets potential customers based on their online behavior. ### ActiveCampaign Some B2B marketers turn to this tool for email marketing and automation along with CRM integration, site tracking data, and other metrics. ## Key Takeaways Creating an effective lead-nurturing strategy helps to build relationships that convert leads to customers. Use high-quality content and other touchpoints to provide assistance with their decision, focusing on what differentiates you from the competition. “At the end of the day, the best lead-nurturing strategy is simple: Be helpful, stay consistent, and watch the data,” Marty says. “You don’t need fancy language or over-the-top designs. Just offer real value, respond quickly, and show that your firm is ready to help. That’s what turns leads into loyal clients.” ## Frequently Asked Questions (FAQs) ### What is the difference between lead nurturing and lead generation? Lead generation is largely about identifying the people you want to reach out to, while lead nurturing is beginning to engage with those leads. “I’ve run campaigns that drove hundreds of leads in a week, and honestly, most of them went nowhere,” Melian says. “I've discovered that lead generation is the invitation, whereas nurturing is the actual conversation.” ### What is a lead-nurturing CRM? A lead-nurturing CRM is a customer relationship management system that helps you track, manage, improve, and build relationships with your leads, throughout the entire lead-nurturing process. This starts as soon as you identify a potential lead, and carries through until they are a fully engaged and consistent buyer. CRMs help with lead scoring, segmentation, tracking a lead’s spot in the pipeline, automating messaging and touchpoints, and producing the data that is necessary for decision-making. ### What is lead scoring? Lead scoring is a process by which you assign value to certain leads, based on how important they are to your business, how likely they are to convert, and how much they will lead to significant sales or engagements with your company if they do convert. The lead-scoring process helps marketing teams identify where to focus their resources. ### How can I create a lead-nurturing plan that drives conversions? Follow each step of the lead-nurturing process, utilizing technology to ensure you’re focusing on the most important aspects and looking for gaps in your lead-nurturing plan. With consistency, patience, and a concrete strategy, you will see the fruits of your efforts in the loyalty of long-term clients and customers who are happy to refer others. --- ### B2C Marketing: Definition, Challenges, Trends URL: https://www.taboola.com/marketing-hub/b2c-marketing/ Last Modified: 2025-05-20 08:51:33 Learn how B2C (business-to-consumer) marketing can help grow your brand, how to navigate common challenges, and how to design your performance marketing strategy. The B2C, or business-to-consumer, marketing model is highly popular, and depending on how your business is structured, it may make the most sense for you. To maximize the effectiveness of your marketing, you’ll need to understand the differences between B2C marketing and B2B marketing, as well as the challenges that come with the B2C model and how to design your marketing strategy to overcome those challenges. ## What Is B2C Marketing? B2C marketing is an approach in which businesses sell products or services directly to consumers. Both online retailers and brick-and-mortar businesses use B2C marketing, which involves acquiring and retaining customers, and it often uses emotions, like desire, to drive sales. ## The Difference Between B2C vs. B2B While B2C marketing focuses on selling products or services directly to consumers, a B2B, or business-to-business, marketing approach involves selling products or services to businesses. There are several key differences between these marketing models: ### Audiences B2C and B2B marketing are designed to reach different audiences, with B2C marketing reaching a wide swathe of potential customers, most of whom will make the decision to convert (or not) quickly, with minimal touchpoints, depending on the price of the product or service on offer. B2B marketing focuses on a more specific audience, often decision-makers in other businesses, and generally requires a longer, more involved funnel. This audience is naturally smaller than a typical B2C audience. ### Purchase Volumes There’s a significant difference in the common purchase volumes of B2C and B2B marketing. B2C purchases are often a smaller volume, maybe even a single product, while B2B marketing purchases are much larger, and might involve ongoing subscriptions or plans for additional future purchases or upgrades. ### Marketing Strategies Since B2C and B2B marketing focus on different audiences, the strategies also differ. B2C marketing often centers on buyers who make relatively quick buying decisions. They’re likely to make a purchase soon after learning about a brand or product, and their buying decisions are usually based on a need or emotion. As a result, B2C marketing often focuses on the audience’s pain points and makes emotional appeals, while also trying to make the purchasing process as quick and easy as possible. B2B audiences are making larger and more substantial purchases, so B2B marketing strategies focuses more on building relationships and trust, rather than driving an initial sale. This type of marketing often involves establishing ongoing lines of communication with prospects, answering questions, and providing personalized quotes and support. B2B audiences are likely to perform more research over a longer period of time than a B2C audience, so providing information and assisting them during the process is a large part of B2B marketing. ## Challenges of B2C Marketing ### Reaching the Right Audience For B2C marketing to be effective, it needs to reach your target audience, and be crafted in a way that will appeal to them. That process starts with defining and understanding your audience, including their demographics, pain points, and the marketing channels they already use. You can gain insight by reviewing data on your current customers, and creating buyer personas to help you better identify and envision your target audience. Once you’ve identified your audience, you’ll need to personalize your marketing as much as possible so that it resonates. It’s also important to carefully choose your marketing channels so that you reach your target customers where they are. You may need to test different marketing strategies, messages, and channels to find what works best for your campaign goals. ### Converting Leads Lead conversion can be a challenge in B2C marketing, and getting potential buyers through that last step of checking out remains difficult: The average online shopping cart abandonment rate worldwide in 2025 is 70.19%. To combat this, it’s important to make your checkout process as smooth as possible. Eliminate any unnecessary fields from the checkout process and make sure that your website is optimized for mobile (Capital One Shopping research reports that 57% of e-commerce sales come from mobile devices globally). Implementing cart abandonment reminder emails and offering discounts to shoppers may prompt them to complete their checkout process, converting them into paying customers. ### Data Management B2C marketing can generate large amounts of data, and that data can inform marketing decision-making. But, businesses need to determine how to collect, organize, and use that data, at least in part to be sure they’re complying with privacy laws. Choosing a performance marketing platform that’s designed to leverage first-party data can help businesses connect with their audiences and drive sales. That platform should also have user-friendly tools for campaign management and optimization, which can empower a business to craft effective campaigns and scale its marketing efforts. ### Delivering an Optimal Customer Journey The marketing customer journey, which includes all of the interactions a potential customer has with your business, shapes their perception of your brand. It’s essential to create an optimal journey that encompasses different stages and touchpoints, helpfully leading the customer from their discovery of your business to the moment they make a purchase. Creating a customer journey map outlining those steps can help ensure you’re including all of the phases necessary in your marketing process. ### Competing With Other Brands Businesses often face steep competition in B2C marketing, making the landscape particularly challenging for small businesses. Larger brands with more substantial budgets have more marketing power, so smaller businesses need to find unique ways to stand out. There are several ways to accomplish this: Businesses can focus on crafting an engaging story and branding, e.g., highlighting the individuals behind the brand to create a more personal engagement with customers. Businesses can also find creative ways to provide value, such as through webinars or tips and information that their audience would find useful. Developing relationships with their audience and creating a sense of community can likewise drive brand loyalty and make a business stand out. ## How to Design a B2C Marketing Strategy Designing a B2C marketing strategy requires a deep understanding of your target audience and what will motivate them to purchase your product or service. There’s no one-size-fits all approach that works for every business, but these steps can help you design an effective strategy. ### Define Your Audience The better you can define your target audience, the better you’ll be able to craft a marketing strategy to connect with them. Review data on your current customers to better understand who makes up your customer base, and consider running surveys or holding focus groups with customers to better understand their backgrounds, pain points, and motivations to buy from you. Get specific with your target audience data, including details on demographics, income, purchasing behavior, interests, and the marketing channels they use. This detail will help guide your marketing strategy. ### Identify Your Marketing Channels Consider the marketing channels where you’re likely to find your target audience. These channels can range from social media and email marketing to print ads and direct mail. Identify which channels you want to focus on and develop a strategy for each channel. It’s important to familiarize yourself with the challenges and benefits of using each channel, and also consider the budget you’ll need for each. Be prepared to perform channel testing to further inform your marketing strategy, so you can identify which channels and which messages are giving you the best results. ### Design Messaging That Appeals to Emotion When it comes to crafting messaging for B2C marketing, appealing to buyers’ emotions is a solid technique. Emotion can help motivate decisions to buy, especially since B2C audiences often make decisions to purchase promptly and with minimal research. This is a good time to establish a unique and engaging brand tone that will be present in all of your marketing materials. ### Make Your Website Easy to Navigate Be sure to invest plenty of time working on your website and make sure that it’s simple for your audience to navigate. Your website needs to be visually appealing, well-organized, and easy to understand. It should clearly present the information your customers will want and need, and should be accessible to mobile users, too. Focus on the checkout process and make sure that the site accepts multiple popular and trusted payment options, like PayPal and Venmo. Do plenty of troubleshooting to make the checkout process as easy as possible, which can help reduce cart abandonment rates and increase your conversions. ## Future Trends in B2C Marketing ### Artificial-Intelligence (AI)-Assisted Marketing AI is rapidly shaping and enhancing B2C marketing. Thanks to AI, your business can access predictive analytics to refine your strategy, performing detailed behavior analyses to better understand your target audience and crafting marketing messages to engage with them, tailoring dynamic content to individual customers’ interests and buying habits, and more. These are just a few of the ways that marketers are already using AI, and this technology is sure to allow for even more powerful marketing capabilities in the coming years. ### Social Commerce As social media platforms have evolved, they’ve become shopping platforms themselves. Shoppable posts and live shopping experiences are meeting your audience where they already are, providing an easier and more efficient checkout experience. Social commerce has removed some of the barriers and hangups that are inherent with directing traffic to your site, making it easier to convert leads. ### Short-Form Video It’s hard to ignore the growing power and use of short-form video. Platforms like TikTok and Instagram Reels have allowed marketers to rapidly reach large audiences, especially when content goes viral. Your business can use short-form video to personalize your brand, deliver valuable informative content, increase social media engagement, and portray products in a captivating way. Investing in short-form video creation is an excellent way to create valuable content that you can use for multiple marketing purposes. ### Personalized Interactions Personalized interactions during the customer journey are becoming more important, and customers will soon come to expect these. For example, you can use data analytics to tailor product suggestions to customers based on their past internet activity and buying habits, and chatbots can provide more responsive and personalized support, even when live customer support isn’t available. Some advertising platforms even use predictive AI to target users by intent, rather than just identity, analysing audience behavior and calculating which ones are most likely to buy. ## Key Takeaways B2C marketing is a highly common and versatile marketing structure, and it’s the foundation of many thriving e-commerce businesses. Marketing directly to consumers offers many benefits, including an opportunity to build a relationship with your customers and boost customer loyalty. ## Frequently Asked Questions (FAQs) ### What is B2C product marketing? B2C product marketing is the process of marketing and selling products directly to consumers. For example, a small business owner producing water bottles could market their product directly to consumers by building a following on social media, advertising on websites, building an email list to reach potential customers through email, and more. ### What is an example of B2C? Amazon is a prime example of B2C marketing. Amazon’s business is structured on marketing directly to consumers, and that marketing strategy encompasses strategies like social media, influencer marketing, paid digital advertising, and more. ### What is the most popular B2C performance marketing software? There are many popular B2C performance marketing software options available, but rather than choosing the most popular platform, it’s important to select a platform that works best for your business and your marketing needs. When choosing performance marketing software, look for a platform that works from robust first-party data to maximize ROI, uses AI to optimize targeting and creative in real time, and gives you access to user-friendly tools for effective campaign management and optimization. --- ### ChatGPT: Features, Benefits, and Marketing Applications URL: https://www.taboola.com/marketing-hub/chatgpt/ Last Modified: 2025-08-10 08:56:19 Discover how ChatGPT works, its many real-world applications for marketers, and what, exactly, the latest updates mean both for businesses and everyday users. ChatGPT, an artificial intelligence tool, was the “shot heard around the world” when its GPT3.5 model was released to the public on November 30, 2022. Created by OpenAI, a San Francisco-based research group launched in 2015 that counts Elon Musk and Sam Altman among its founders, ChatGPT was the first generative AI tool to come into wide use for the general public. (The term “generative” refers to the chatbot’s ability to create original text, images, and videos based on user prompts.) People couldn’t decide whether ChatGPT was the best or worst thing that had ever happened to their industry. Students loved that they could generate entire papers in seconds; teachers hated it for the same reason. Doctors both loved and hated how competent it was at diagnosing illness based on symptoms and the patient profile. Many industries — including marketing — were excited that it could be used to automate mundane tasks, but were also worried that AI would replace their jobs. Its “wow” factor, of course, was that you could enter prompts in normal, idiomatic full sentences and the bot would reply in equally normal-sounding full sentences. It marked the mainstreaming of AI technology, and for the first time the average person could see for themselves what the benefits and dangers of AI could be. In this article, I’ll examine what ChatGPT does, how it works, how it can be used, and how marketers can use it to their advantage. ## What Is ChatGPT? Let’s start with the GPT part. An acronym for Generative Pre-trained Transformer, you can think of GPT as a Google search engine on steroids. To break down the three components of the acronym: ### Generative It generates original content. ### Pre-trained It has been front-loaded with data and its algorithms are already poised to respond to your prompts without the users having to train it themselves. ### Transformer Not to be confused with the types of transformers used in electrical circuitry, “Transformer” in this context was introduced by Google in 2017. According to the company’s official literature, a Transformer is “a novel neural network architecture based on a self-attention mechanism that we believe to be particularly well-suited for language understanding.” In other words, it processes all the words in a sentence or text all at once instead of one at a time, allowing it to parse meaning more quickly and with greater accuracy than preceding technology allowed. ChatGPT is a conversational AI model built by OpenAI, based on the GPT architecture. While GPT technology is not proprietary to OpenAI, ChatGPT is.  As described on OpenAI’s official site, ChatGPT “interacts in a conversational way. The dialogue format makes it possible for ChatGPT to answer follow-up questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests.” The distinguishing characteristics of ChatGPT — what makes it revolutionary from the user’s POV — are its machine learning (ML) capabilities and its large language model attributes (LLM): ### Machine learning A way for computers to learn from data and make decisions or predictions without being explicitly programmed. It’s like teaching a computer to recognize patterns and improve over time, similar to how humans learn from experience. ### Large language model A type of machine learning system trained on vast amounts of text to understand and generate human-like language. It’s like a hyper-intelligent chatbot that can answer questions, write stories, or have conversations by predicting which words come next. ## How Does ChatGPT Work? ChatGPT is trained on vast datasets of text to predict and generate human-like responses. It processes user inputs, interprets context, and delivers relevant outputs. Fine-tuned for conversational tasks, it continuously improves through user interactions and updates, ensuring accuracy and adaptability across diverse applications. But hey, why not go to the source itself? Here’s what happened when I asked ChatGPT to define itself and asked one follow-up question: ## What Are the Benefits of ChatGPT? ChatGPT offers numerous advantages, making it a go-to tool for various industries. ### Enhanced Productivity ChatGPT automates repetitive tasks like drafting emails, generating reports, or answering frequently asked questions (FAQs), freeing up time for strategic work. For marketers, this means more focus on campaign planning and audience targeting. ### Cost Efficiency By streamlining workflows, ChatGPT reduces the need for extensive human resources in tasks like content creation or customer support. Businesses can achieve high-quality outputs without significant investments in additional staff or tools. ### Scalability ChatGPT handles large volumes of queries or content generation, making it ideal for businesses scaling operations. Its versatility supports applications from small startups to global enterprises. ## Considerations When Using ChatGPT While powerful, ChatGPT has limitations and ethical considerations that users must address. ### Accuracy and Verification ChatGPT may occasionally produce inaccurate or biased outputs. Users should verify critical information, especially in professional settings, to ensure reliability. Out-of-Date Material ChatGPT does not have up-to-the-minute updates in real time. To find out what the current knowledge cutoff date is for the version you are using, simply type in, “What is your knowledge cutoff date” into the ChatGPT prompt. For example, I entered this prompt on April 28, 2025: ### Data Privacy When using ChatGPT, sensitive data shared in prompts may be stored or processed. Businesses should avoid inputting confidential information and review OpenAI’s privacy policies to understand data handling. ### Ethical Use Overreliance on AI-generated content can raise concerns about authenticity. Marketers, for instance, must balance AI use with human creativity to maintain brand integrity and avoid generic outputs. ## What Can Users Do With ChatGPT? ChatGPT’s versatility enables a wide range of applications for diverse users. ### Content Creation Users can generate blog posts, social media captions, or creative stories. ChatGPT tailors content to specific tones, styles, or audiences, simplifying the creative process. ### Task Automation From scheduling emails to coding simple scripts, ChatGPT automates routine tasks, boosting efficiency for professionals and hobbyists alike. ### Learning and Research Students and researchers use ChatGPT to summarize articles, explain complex concepts, or brainstorm ideas, making it a valuable educational tool. ## How Are Marketers Using ChatGPT? Marketers can harness ChatGPT to optimize their campaigns and engage audiences effectively. Here’s how: ### Content Marketing ChatGPT can enhance content marketing by automating content creation, optimizing strategy, and personalizing engagement. - Content creation: This is the low-hanging fruit, and the aspect you’ve most likely already experimented with. ChatGPT can generate blog posts, social media content, and ad copy quickly. It can also create charts, data visualization, videos, graphics, and infographics. - SEO: ChatGPT can optimize your content by identifying keywords and rewriting text for better rankings. It can also help you come up with SEO-friendly headers. You can refresh old content with ChatGPT-suggested improvements instead of always creating new material. ### Social Media ChatGPT can craft compelling captions, recommend trending hashtags, and even generate image or video content ideas that align with audience interests. It can also optimize the posts, depending on the venue. ### Customer Support Automation ChatGPT powers chatbots that handle customer inquiries, providing instant responses. This improves user experience while reducing support team workloads, enabling marketers to focus on strategy. ### Landing Page Text ChatGPT can be a powerful tool to help customize your landing pages for better conversions. You can use it to fine-tune existing pages or create multiple variations from the same content, helping you target different audiences more effectively. ## How Much Does ChatGPT Cost? ChatGPT offers Free, Premium, Pro, Team, and Enterprise plans: ### Free Gives you access to real-time data from the web with search, limited access to models GPT‑4o (as of this writing, the most sophisticated model), o4-mini (the “lite” or free version of GPT-4o), and deep research. Also gives limited access to file uploads, data analysis, image generation, and voice mode. ### Premium $20/month. Gives you access to all the features offered in the Free version but with greater access to research, uploads, etc. It also allows the user access to deep research and multiple reasoning models (o3, o4-mini, and o4-mini-high), as well as access to a research preview of GPT‑4.5, their largest model yet. ### Pro $200/month. Gives you access to all the Premium features, but with uncapped access; access to o1 pro mode, which uses more computing power to come up with the best answers to the hardest questions; extended access to deep research; and extended access to Sora video generation. ### Team $25-$30 per user/month. All the features in Pro, customized to your team’s workflow. Protects your data with multi-factor authentication, data encryption, and data excluded from training by default. ### Enterprise Custom-built, pricing on a per-case basis. ## Latest ChatGPT Updates OpenAI regularly enhances ChatGPT to improve functionality and user experience. To discover the latest information about updates, go to the official OpenAI site page that provides model release notes. ### New and Smarter Models OpenAI recently launched new versions of ChatGPT (like GPT-4.1 and GPT-4o) that are better at understanding questions, thinking through answers, and giving more detailed and accurate responses. ### Better Voice Conversations ChatGPT’s voice mode is now more natural and expressive, making conversations sound smoother and more lifelike. Some of the newest voice features are rolling out slowly, so they might not be available everywhere yet. ### Easier Connections to Other Apps OpenAI is making it easier to connect ChatGPT to other tools, like marketing software, customer service apps, and work platforms. This helps businesses use ChatGPT to save time and get more done. ## Key Takeaways ChatGPT is a versatile AI tool that enhances productivity, automates tasks, and supports creative and marketing efforts. While it offers significant benefits, like cost efficiency and scalability, users must consider accuracy, privacy, and ethical use. Marketers, in particular, can leverage ChatGPT for content creation, audience targeting, and automation, especially when paired with performance-focused AI ad platforms that optimize in real time. Staying updated on OpenAI’s advancements ensures users maximize ChatGPT’s potential. ## Frequently Asked Questions (FAQs) ### Who created ChatGPT? ChatGPT was created by OpenAI, a research organization co-founded by Elon Musk, Sam Altman, and others in 2015, focused on advancing AI technologies. ### What are the ethical concerns associated with ChatGPT? Ethical concerns include potential biases in outputs, overreliance on AI for creative tasks, and the risk of misinformation. Users should verify outputs and use ChatGPT responsibly to maintain authenticity. ### Is ChatGPT free? ChatGPT offers a free tier with limited features. Premium plans, like ChatGPT Plus, Pro, Team, and Enterprise, provide enhanced capabilities for a subscription fee. Find out more on OpenAI’s website. ### What is voice mode vs. advanced voice mode? Voice mode allows users to interact with ChatGPT via voice commands, enabling hands-free conversations. Advanced voice mode, available on select platforms, offers more natural speech, expressive tones, and faster response times, enhancing the conversational experience. --- ### Ad Agency: Your Partner for Reaching the Right Audience URL: https://www.taboola.com/marketing-hub/ad-agency/ Last Modified: 2025-05-20 08:34:16 Take a deep dive into the world of ad agencies, as we break down their departments, functions, and why companies hire them for strategic planning and more. Back in 2003, my college roommate and I would watch TV together and constantly mock the ads, picking them apart and noting how “we could do it better.” That’s what set me down the path into advertising. Fast forward to 2025, and I’ve been a copywriter for nearly 20 years, first going to school for it, then interning at different agencies, and finally finding my way in the employment world. One of the biggest changes I’ve seen over these past two decades is the shift from companies using ad agencies for all their creative needs, to building their own agencies in-house. When I started out, everyone was vying to get into a creative agency: In-house positions existed, but we craved the creativity of the ad agency environment, where you’d have multiple, external clients, instead of working for just one (and having them be your employer, too). This was around the time of Mad Men, so the mystique and glamor of agency life was well-publicized in the national consciousness. But, times have changed. Though I started off in the ad agency world, I’ve since moved to mostly in-house work, with a major benefit being that employment opportunities have expanded significantly. Instead of everyone fighting for a spot at the same agencies, in-house opened up way more options. All that being said, agencies are not only still around, but seem to be having a moment. With the current media landscape being as crowded as it is, standing out is harder than ever, and a full-service ad agency may be just what your business needs. So, let’s look at the ins and outs, and the strengths and weaknesses of it all. For this piece, I’ve enlisted the help of two experts: Kyle Sobko, CEO of SonderCare, who has worked with both ad agencies and internal teams over more than two decades in sales and marketing, and Aaron White, CEO and co-founder of the marketing platform Outbound. ## What Is an Ad Agency? Here’s how Sobko sums it all up: “An advertising agency plans and manages advertising campaigns across platforms like search, social, video, display, and broadcast. They handle creative direction, audience targeting, budget allocation, media placement, A/B testing, and ongoing performance optimization. Their role is to execute campaigns that meet defined goals, while coordinating all the moving parts across teams, platforms, and timelines. They work as both strategist and operator.” Agencies are a group of skilled professionals dedicated to linking products and services with the right audiences, and delivering the right message to grab their attention. They do more than just design eye-catching visuals or come up with snappy headlines and taglines, though — an ad agency steps up as your all-around marketing partner, offering their wide range of expertise to help businesses meet their goals, including things like boosting brand visibility, increasing sales, or even changing public perception after a negative incident. “An ad agency is a company that helps businesses create, plan, and manage their advertising efforts across various channels,” adds White. “It’s essentially a creative and strategic partner for getting a brand in front of the right audience.” Ad agencies are an entire external marketing team ready to reach the right audience you want. They bring the necessary knowledge, experience, and creative power to the table, crafting the brand's voice, personality, and visual presence, along with narratives that form an emotional connection. Smaller businesses who don’t have the budget to hire an in-house team can see big benefits from having a dedicated agency at their service, as can large companies who would rather outsource all the creative and data-driven metrics. ## What Does an Advertising Agency Do? “An ad agency handles everything from market research and creative concepting to media buying, ad placement, and campaign optimization,” says White. “It’s a full-service hub for brand messaging and promotion.” To say the least, they do a lot, and the responsibilities of an ad agency can be as diverse as the types of clients they serve. Overall, though, their primary function is to help businesses communicate effectively with their desired target audience, and getting to that goal together is where they really shine, as all the departments work in tandem. Here's a glimpse into the roles they all play, and why each one is important: ### Strategic Planning As a creative, we can’t just start thinking up ideas as soon as we get an account: We need planners to help us understand the client's business, target audience, market landscape, and objectives. They dive into the research, analyze the data, and develop the marketing strategies that serve as a map for everything going forward. The account planners create a brief, which lists important points like the brand's positioning, key messaging, and determining the most effective channels to reach the intended audience. As freeing as it may sound to “come up with anything” (and have an unlimited budget), having guidelines and guardrails to work within is especially helpful for the creative team. ### Creative Development This is where I’ve spent my entire career, often being the only copywriter on a team full of art directors. The creative department, at least from my point of view, is the heart of the agency. We’re responsible for thinking up and producing the attention-grabbing advertising across various media. This includes everything from memorable taglines and engaging sales copy, to designing and developing visual and video content. ### Media Planning and Buying Getting your message in front of the right people at the right moment is the goal of every client that’s pushing a product or service. You can have the perfect ad, with top-notch design and copy that does exactly what the client wants, but if it’s not reaching the right crowd, it falls flat and is quickly forgotten. The media team is all about identifying the best channels to send that message out into the world. Traditionally that's been through TV, online, social media, print, or outdoor placements, but in an ever-changing landscape, there’s always innovative opportunities popping up. They negotiate with media outlets to lock in the best rates, making sure the client's budget goes as far as possible while maximizing reach and impact. Think of them as the distribution experts, making sure the message hits home where it makes the most impact. ### Account Management The account managers are liaisons between us, the creative team, and the clients. We’ll most likely talk with clients at meetings and conference calls, but it’s the account people who are always in contact with them, ensuring smooth communication, managing projects, and building strong, lasting relationships. They go deep into understanding the client's needs, translating them into actionable tasks for the various agency teams, and keeping the client informed of progress and results. Creatives may get the recognition when a campaign is successful, but we couldn’t do it without the account team. ### Digital Marketing Having a solid online presence is essential nowadays. Many agencies have specialized digital marketing teams that take care of various tasks, including search engine optimization (SEO), pay-per-click (PPC) ads, social media management, content marketing, and email campaigns. They skillfully maneuver through the constantly changing digital world to assist brands in reaching consumers online and achieving digital success. ### Production Once the creative concepts are approved, which usually takes multiple rounds of back-and-forth edits, the production department takes the reins to bring them to life. This involves overseeing the creation of commercials, print ads, digital assets, and other advertising materials, ensuring they are produced to the highest quality standards, and stay within budget. Creatives tend to dream big, but it’s the production team that makes our vision a reachable reality. ### Research and Analytics Another area that helps us creatives significantly are the Research and Analytics team. Their insights spark ideas that often become entire campaigns. Throughout the process, R&A play a vital role, conducting market research to understand consumer behavior, tracking the performance of campaigns, and analyzing data to identify areas for improvement. This data-driven approach ensures that everything we’re doing is going to be effective and aligned with the client's goals. ## Benefits of Working with an Ad Agency Lots of companies have chosen to move everything in-house over the past few decades, so at this point, why would a business choose to partner with an advertising agency instead? “Working with an ad agency gives you access to specialized talent, fresh creative ideas, and media buying power you wouldn’t have in-house,” says White. “Plus, it frees you up to focus on running your business while experts handle the strategy.” Agencies can have more, and farther reaching, benefits that push a company's marketing success further than if they did it alone, such as the following: ### Expertise and Experience While it may take an in-house team some time to find their footing and assemble a group that works well together, agencies bring a wealth of specialized knowledge and years of experience right off, and in every department. They’re experts that are dedicated to each aspect of the advertising timeline, bringing a level of skill and insight that may be difficult for a company to replicate internally. ### Fresh Perspectives and Creativity In-house teams can often become “too close” to the project, since everything they do is for the same client. The benefit of that is a fast-track to learning about your own product’s offerings, but it can be hard to take a step back and see the big picture when you’re continuously working on the same company’s offerings every day. Agencies, however, have an objective, outside perspective, bringing fresh ideas and creative solutions. They’re not bound by internal biases or ingrained ways of thinking, allowing them to develop innovative and impactful campaigns that can cut through the clutter. ### Cost-Effectiveness It may sound backward, but working with an agency can be more economical than building and maintaining a full in-house marketing team. Agencies have established relationships with media vendors, often snagging better rates, and they can scale their services to meet the client's specific needs and budget. ### Access to Resources and Tools Agencies invest in the latest marketing technologies, tools, and research resources, providing clients with access to capabilities they might not otherwise have. ### Focus on Core Business By outsourcing advertising and marketing to an agency, businesses can free up internal teams to shift their skills to other strategic priorities. ## What are the Types of Advertising Agencies? There’s not just one style of advertising agency. In fact, there’s quite a few, and they can be pretty varied, with different types catering to a client’s unique requests. ### Full-Service Agencies These are the most all-in-one shops for your advertising needs. They provide a wide array of services, from strategic planning and creative development to media buying and digital marketing. Basically, they can take care of all the big stuff that clients might need on their advertising journey. ### Specialized Agencies When you need a specific niche, like digital marketing, social media, public relations, or even healthcare advertising, call on a specialized agency. Their deep knowledge and focus in a particular area can be a game-changer for clients with special requirements. ### Creative Boutiques Usually smaller in size, these agencies are all about creative development. They come up with fresh and impactful advertising ideas and often collaborate with other agencies for media buying and additional services. But, creative is at the center of it all. ### Media Buying Agencies If your creative is all ready to go, where to put it is the next step, and that’s where a media buying agency can be helpful. They’re experts in planning and purchasing media space (and time) for clients, and usually have strong connections to media vendors, allowing them to get better buying options than if you go it alone. ### In-House Agencies While technically part of a client’s organization, larger companies sometimes set up their own internal advertising teams to manage their marketing needs directly. I’ve worked on plenty of these teams and it has its advantages, like being fully immersed in the product and company, compared with being at an agency where I’d be working on multiple different clients. ## Common Ad Agency Services Ad agencies offer a wide array of services, the main ones being: - Brand strategy and development. - Market research and analysis. - Creative concept development and execution (copywriting, graphic design, video production). - Media planning and buying (online and offline). - Digital marketing (SEO, PPC, social media, content marketing, email marketing). - Website design and development. - Public relations and communications. - Campaign management and optimization. - Performance tracking and reporting. ## Key Takeaways Advertising agencies do so much more than just brainstorm creative ideas: They’re strategic partners, helping businesses navigate the increasingly complex world of marketing, and connect with their target audiences effectively. The skills and services they offer help brands achieve their goals, whether it's enhancing visibility, driving sales, or a specialized request that a company can’t do alone. Agencies craft compelling narratives, choose the right media channels, and analyze a campaign’s performance for the best outcome possible. A lot has changed since the days when Mad Men was set (and even since the days when the show aired), but one thing is still true as ever: Ad agencies play a crucial role in the modern marketing landscape. ## Frequently Asked Questions (FAQs) ### What is the future of ad agencies? The process of creating and launching a campaign isn’t as simple as it used to be. Going forward, agencies will need to be flexible and embrace new technologies if they want to stay current and make a real difference. Along with that, expect to see more teamwork and specialization within the agency landscape. “The future of ad agencies lies in AI and data. Agencies that can move fast with changing platforms, and leverage AI in their workflows, will thrive,” says White. The evolution of agencies is already happening, and is quickly moving toward strategies that are more focused on data, personalized advertising, and a seamless mix of online and offline channels. “I believe we will see a structure more similar to software-as-a-service (SaaS) — monthly subscriptions, embedded teams, transparent reporting, flexible scopes,” says Sobko. “We are currently still testing the model of being an embedded partner, working like an extension of our growth team by opening Slack channels and sharing access to their dashboards and KPI reports on a weekly basis, etc. They are not pitching us: They are inside the business. It’s not fancy, but I see that’s the direction we are going — far less presentations and reporting, just execution and quicker feedback loops for accountability.” Sobko adds that, “I’ve had a front-row seat to the agency model’s transition over the years. It’s not about flashy campaigns anymore. The agencies that are going to stay relevant are the ones who demonstrate real-time data, and who can affect a company’s operations — not just advertising creative.” ### What are the Big 4 advertising agencies? When you hear the term "Big 4," it’s most likely referring to the four largest advertising holding companies globally, which own numerous individual agencies (or if they don’t already own, will probably acquire them soon). These are generally considered to be: - WPP: A multinational communications, advertising, public relations, technology, and commerce company. - Omnicom Group: A global advertising, marketing, and corporate communications company. - Publicis Groupe: A French multinational advertising and public relations company. - Interpublic Group of Companies (IPG): An American multinational advertising agency holding company. ### Do ad agencies make money? Even though they’re not a company’s only option for launching a campaign anymore, they absolutely still make money. At the end of the day, ad agencies are businesses that generate revenue through various means. “There are a few different ways ad agencies make money,” says White. “Some charge service fees, retainers, commissions on media buys, or even performance-based models. The most successful ones create long-term partnerships by delivering results, not just ads.” Some revenue models include: - Commission-based: Historically, agencies earned a percentage of the media spend they placed for clients. While less common now, it still exists in some forms. - Fee-based: Agencies charge clients a set fee for their services, often based on the scope of work, time involved, or project deliverables. - Retainers: Clients pay agencies a recurring fee for ongoing services over a specific period. - Performance-based: In some cases, agency compensation is tied to the achievement of specific marketing goals or key performance indicators (KPIs). - Markups: Agencies may add a markup to the cost of certain services, such as production or third-party vendor fees. But the true profitability of an ad agency depends on factors such as its size, specialization, efficiency, client relationships, and the overall economic climate at the time. From the initial partnership between a company and agency, to the planning, creative, media buy, and launch, the advertising business is endlessly fascinating and fun. Agencies may not have the glitz and glamor and three-martini lunches of yesteryear, but they’ve evolved to meet the current landscape, and will continue to do so, no matter how much times change. --- ### Artificial intelligence (AI): The Digital Marketer's Guide URL: https://www.taboola.com/marketing-hub/artificial-intelligence/ Last Modified: 2025-06-16 04:03:27 Learn exactly what artificial intelligence (AI) is, how AI works, its many major benefits and challenges, and discover real-world AI use cases across industries. Artificial intelligence (AI) refers to machine “thinking” that simulates human intelligence. To be precise, AI in the purest sense — technology that perfectly mimics the functioning of the human brain — is still just fantasy (or rather, Sci-fi): What we call AI today can only work with the information it is fed. Even if the intelligence of current AI isn’t human, though, it does have the ability to make itself smarter, which is still inarguably impressive. Machines based on AI technology can process enormous amounts of data with tremendous speed, which is why they’re so adept at pattern recognition, language learning, and complex decision-making. They’ve already begun transforming many fields, from medicine to agriculture to education. In the performance marketing world, AI tools have already proven adept at identifying high-intent users, adapting in real-time to user behavior, and optimizing their journey through the conversion funnel. ## How Does AI Work? In the simplest terms, AI is a union of volume, speed, and algorithmic logic. It involves crunching tons of data, spotting patterns, and making decisions. Here’s how an AI system usually works: - It takes in and metabolizes huge amounts of data. For example, in 2024, Meta fed hundreds of thousands of books into its AI models. - It uses machine learning (ML) to pick out trends or connections. (More on that below). - It uses neural networks — technology that mimics the human brain’s ability to recognize patterns — to make decisions like suggesting or automatically taking actions based on what it’s learned. - It keeps getting smarter as it processes more data over time. Tools like ChatGPT and DeepSeek are known as Generative AI tools, so named because they can generate new text, images, and videos. The “wow” factor — their ability to have normal, human-sounding conversations — comes from the Large Language Models (LLMs) on which they’re based. ## Types of AI ### Narrow AI (Weak AI) Of the three types of AI discussed in this section, Narrow AI is the only one that exists in the real world. Narrow AI is built to perform a single job: These systems are tightly programmed for their task and can’t step outside those parameters. One can find numerous use cases for narrow AI in the fields of e-commerce and digital marketing: - Amazon and Netflix product recommendations: When you shop on Amazon and see suggestions like, “People who bought this self-cleaning litter box also bought this cat litter,” that’s Narrow AI at work. The recommendation engine crunches your browsing history and purchase data, as well as the purchases of other users whose buying history resembles your own. - Smart Bidding: Google Ads uses Narrow AI for Smart Bidding. It analyzes data such as who’s clicking your ads, when, and on what device, then automatically adjusts your bids to get the most conversions within your budget. ### General AI (Strong AI) Think of Ava, the main android character in the movie Ex Machina: that’s an example of General AI, bordering on Superintelligent AI. General AI, sometimes known as Strong AI, is hypothetical at this point, but it’s the holy grail of artificial intelligence. Machines driven by General AI don’t just follow narrow scripts, but think and adapt across all sorts of tasks, including writing fiction, problem solving, and offering advice in a way that surpasses its training. In theory, these would be as versatile as the human brain. Researchers are making progress, but we’re currently nowhere near machines that can match the full range of what the human brain can do. ### Superintelligent AI Dystopian depictions of sentient AI machines that want to be autonomous from humans are generally referring to Superintelligent AI machines. Examples would be the setting of William Gibson’s 1984 sci-fi novel Neuromancer. This technology is not real, but if it were, machines built on its principles would leave human intelligence in the dust. Ethical questions would arise about whether they are capable of good and evil. Superintelligent AI could lead to what is known as singularity — a hypothetical scenario where artificial intelligence can autonomously improve upon itself without human intervention and advance uncontrollably. ## AI Models ### Machine Learning Models Machine learning models are trained to recognize patterns and make decisions without being explicitly programmed for every scenario. They can be supervised (trained on labeled data), unsupervised (finding hidden patterns in unlabeled data), or reinforced (learning through reward-based systems). An example of machine learning is a spam email filter. It uses a learning algorithm, like a decision tree or neural network, trained on a dataset of emails labeled as "spam" or "not spam." The model learns patterns, such as specific words or phrases, to classify new emails. For instance, if an email contains "Earn $$$$," the model might flag it as spam based on the parameters of its programming. Over time, user feedback — marking emails as spam or not — refines the model's accuracy. ### Deep Learning Models Deep learning, a subset of machine learning, uses neural networks with many layers (hence "deep") to analyze various factors of data. It's particularly effective in tasks like image recognition, language processing, and fraud detection. An example of deep learning is the facial recognition feature on your smartphone, which analyzes and identifies facial features and compares them against the countless other facial images in its database. Over time, with more input, the feature “learns” to recognize faces with increasing speed and accuracy. ### Large Language Models (LLM) Large language models (LLM) are a type of neural network, trained on vast amounts of text data to understand and generate human-like language. An example of a large language model is GPT-4, developed by OpenAI. This technology serves as the basis for ChatGPT and similar generative AI tools. So, if you’re using an AI-driven performance marketing tool that takes advantage of LLM, you can enter prompts that generate natural-sounding copy for email campaigns, calls to action (CTAs), and social media. ### Natural Language Processing Models (NLP) NLP models allow machines to understand, interpret, and generate human language. Tools like chatbots, translation services, and voice-activated assistants rely heavily on NLP. In marketing, NLP can dynamically tailor messaging based on user sentiment or interaction history. ## Benefits of AI ### Enhanced Efficiency AI dramatically reduces the time needed to analyze data, draw insights, and make decisions. For example, performance marketers using AI-enhanced platforms can automatically optimize campaigns at scale, saving resources while maximizing return on investment (ROI). ### Personalization at Scale AI enables highly personalized experiences for users. By analyzing behavior patterns and preferences, AI can serve custom content and product recommendations — a strategy exemplified by platforms that leverage predictive audience targeting based on first-party data. ### Real-Time Optimization Modern AI-based tools, such as those for marketers, enable real-time adjustments based on user actions. These tools make it possible to adapt to user behavior instantly, driving better outcomes. ## Challenges and Risks ### Bias and Fairness AI systems can perpetuate or even amplify existing biases in the data they are trained on. This could lead to discriminatory outcomes in hiring, lending, or advertising. Developers must remain vigilant about training data and continuously monitor outputs for fairness. In a 2023 experiment with GPT-3 API technology that came to be known as “AI Seinfeld,” Twitch ran a 24/7 channel devoted to a fake sitcom called “Nothing, Forever,” which ran AI-generated episodes of a show that appeared to be based on the 90s sitcom Seinfeld. The scripts for these visually clunky episodes were auto-generated by the bot, which did not contain sufficient content filters for offensive material. Unfortunately, the chatbot started to spew highly offensive content that was not culled from any real Seinfeld episode, nor was it based on malicious user input: It had processed “comedy” material from the internet at large. That’s why any AI-generated content you use needs to first undergo thorough human vetting. ### Privacy and Security As AI systems rely on data, ensuring user privacy and securing data from breaches is a major challenge. Companies working with proprietary first-party data must implement strong encryption, anonymization, and compliance practices. ### Job Displacement This aspect of AI scares people, but its scope has been exaggerated. Automation powered by AI might replace certain jobs, particularly in industries reliant on repetitive tasks. However, it also creates opportunities by enabling new roles focused on AI development, oversight, and integration. Mastering AI-based tools for your own industry is a great way to stay ahead of the pack. ## Use Cases of AI ### Marketing and Advertising AI is revolutionizing how brands engage with consumers. From predictive targeting to real-time creative optimization, AI-driven tools help advertisers find high-intent users and drive them through the performance funnel. This new wave of AI goes beyond email blasts and social media strategies, offering scalable, measurable performance outcomes. ### Healthcare AI assists in diagnostics, personalized medicine, and treatment recommendations. Machine learning algorithms can predict disease outbreaks and aid in drug discovery at speeds previously unimaginable. ### Agriculture Machine learning models forecast crop yields based on historical data, weather patterns, and field conditions. Using drone images, they can identify crop diseases and pests, thereby allowing for early intervention. ### Financial Services Banks use AI for fraud detection, risk assessment, and personalized banking experiences. AI can rapidly sift through millions of transactions to identify suspicious patterns. ### Autonomous Vehicles Self-driving cars rely on AI to interpret sensory information, navigate roads, and avoid obstacles, representing one of the most advanced applications of real-time decision-making AI. ## Key Takeaways Artificial intelligence simulates human intelligence and can adapt based on new data. Modern AI thrives on data ingestion, pattern recognition, and predictive analytics. There are different types of AI, from narrow to theoretical superintelligent systems. AI brings efficiency, personalization, and real-time optimization — especially powerful in performance marketing. Challenges include bias, privacy concerns, and job displacement. AI use cases are transforming industries, from advertising to healthcare. ## Frequently Asked Questions (FAQs) ### What is the history of AI? The concept of artificial intelligence dates back to ancient myths of mechanical beings. One of the earliest AI creations was the so-called Mechanical Turk: In 1770, a Hungarian inventor named Wolfgang von Kempelen wowed the crowned heads and illuminati of Europe by demonstrating his invention, a life-size humanoid automaton wearing a turban that could play master-level chess against humans: A sentient machine. The human player would make their move, then the robot would take his, with the pieces gliding around the board seemingly by themselves. Due to the Orientalist, vaguely Ottoman costume — the automaton resembled Zoltar, the fortune-telling mechanical man from the movie Big — the machine came to be called the Mechanical Turk. It defeated nearly everyone, including, allegedly, Benjamin Franklin. One might even say it was the world’s first example of artificial intelligence. Unfortunately, it was a hoax and was operated by a chess player hiding behind a false panel in the contraption. Modern AI began in the 1950s with pioneers like Alan Turing and John McCarthy. Early enthusiasm led to periods of stagnation known as "AI winters," but advancements in computing power and data availability in the 21st century have brought AI to the forefront of innovation. ### Are there ethical implications with AI? Absolutely. AI can reinforce biases, infringe on privacy, and make opaque decisions with serious consequences. Ethical AI development emphasizes transparency, fairness, accountability, and the minimization of harm. Marketers, for instance, must ensure their targeting practices respect user consent and avoid discrimination. ### Strong AI vs. Weak AI: What's the Difference? Weak AI (or Narrow AI) is designed for specific tasks, like recommending products or filtering emails. Strong AI would theoretically possess consciousness and general reasoning abilities across a broad range of activities, akin to human intelligence. Currently, all deployed AI systems are forms of Weak AI. --- ### Drip Campaigns: The Pre-Planned, Targeted, and Effective Way to Engage URL: https://www.taboola.com/marketing-hub/drip-campaigns/ Last Modified: 2025-05-20 10:33:42 Drip campaigns use pre-written emails sent to potential customers on a schedule, or triggered by specific actions, and are intended to prompt customer conversion. Picture the mighty stalactites and stalagmites seen far below the planet’s surface, gracing so many striking caves. How were those impressive structures created? One drip at a time, of course. Marketers can also use the drip approach for creating truly strong bonds with their audiences, cementing brand awareness, customer loyalty, and purchases, as well. Drip campaigns use a measured approach and pre-planned materials to reach out to potential leads at just the right times and, when executed well, can greatly increase the returns marketers get on their ad spend. Below, I’ll discuss drip campaigns in advertising in detail, covering best practices, what to avoid, what has worked for others, and what marketing experts have to say on the subject. ## What Is a Drip Campaign? “A drip campaign is an automated sequence of emails triggered by customer behavior (or a preset schedule), like signing up, abandoning a cart, or making a purchase,” says Nina De la Cruz, a content marketing strategist at Getsitecontrol. “Each message is designed to guide the subscriber through the buying journey, from awareness to conversion.” Drip campaigns can also include a series of scheduled emails sent to subscribers or past customers on a semi-regular basis, the intention being to maintain customer awareness and contact that lowers the barriers to action, like revisiting a website and making a purchase. While drip campaigns can take on many forms and use varied approaches, they all have one thing in common: The emails (or in some cases text messages) are written ahead of time and are sent out via automation, not by a human’s action at the time. In that way, drip campaigns greatly free up marketers to focus on other aspects of their work. ## How It Works There are essentially two types of drip campaigns: One sends out emails on a pre-planned schedule, the other automatically sends out an email when a specific trigger occurs, such as when someone signs up for a service or leaves a website with a full shopping cart but no purchases. As an example of the first type of drip campaign, online consultant and podcaster Yann Ilunga of YannIlunga.com says that, “"Drip campaigns often refer to a series of touch points, typically emails, that occur at a predetermined cadence. An email course is an example of a drip campaign. A five-day email course consists of a series of five emails that are sent daily over the span of five days.” Brian Akdemir, director of e-commerce at Bahdos, explains the “trigger” type of drip campaign like this: “A lot of drip campaigns are set off by certain actions from users, like looking around a website without buying anything. In these situations, the drip sequence is meant to get the potential customer's attention again and get them to take the action you want them to, whether that's buying something, signing up for your newsletter, or something else. The behavioral data lets the marketer change the drip campaign's message and timing based on the person's interests and where they are in the buying process.” ## Drip Campaign Examples ### Lead Nurturing Imagine an email that you get every Tuesday from a company that makes herbal tea, often featuring special offers, new product announcements, and how-to tips. You can call that a newsletter, but you can also call it a drip campaign. Any series of pre-planned marketing emails that are sent via automation, be it just a series of four or five emails, or be it a weekly, bi-monthly, or other such schedule, is a drip campaign. ### Abandoned Cart Reminders This is one of the most common trigger-style drip campaign approaches. When a customer visits a website, puts items in their digital shopping cart, and then leaves the website without making a purchase, an automation triggers the sending of an email to remind the person of their intended purchase, and often sends along an incentive such as a 10% off discount. ### Welcome Letters When a new person signs up for a service, subscribes to a newsletter, makes a purchase, or takes some other action, it’s common for an automated email to be sent out welcoming them to the brand and thanking them for joining the fold. There are often several follow-up emails still in this welcoming tone. ### Political Campaigns “Political campaigns widely use drip campaigns to build support for a candidate,” says Baruch Labunski, founder and CEO of Rank Secure. “They are highly effective in swaying opinion, especially regarding controversial subjects.” ## Drip Campaign Benefits ### Hands-Off Automation Because drip campaigns use automation to send out emails or texts, they leave marketers free to focus on other aspects of their business, from studying data, to writing copy and creating graphics, to calculating the latest return on ad spend (ROAS). ### Enhanced Brand Awareness A drip campaign keeps a brand top-of-mind for its audience, serving as just enough of a touchpoint to keep people thinking about the products or services on offer without being annoying by approaching too often. ### Increased Engagement and Re-Engagement Drip campaigns are a proven way to increase audience engagement, and also to bring people who have disengaged from a brand back into contact with it. Drip campaigns are almost always more effective than one-off email blasts or other more limited marketing strategies. ## How to Create a Drip Campaign ### Define Your Goals What do you want to achieve with your drip campaign? Are you hoping to drive sales, generate leads, nurture prospects, or onboard new customers? What specific actions do you want recipients to undertake? Purchasing a product, signing up for a service, downloading a file, or visiting a specific site? What will be your measures of success? ### Choose Between Time-Based or Trigger-Based Emails This isn’t an either-or situation, as you can by all means have time-based drip campaigns and also have trigger-based communication at the ready, but when it comes to planning a specific drip campaign, you have to know whether it’s going to work best based on scheduled emails or based on customer action. ### Choose Your Software There are many different platforms that can be used to create drip campaigns, as you’ll see below, so take the time to figure out which one is right for your organization, and then use it to create and schedule the campaign itself. ## Drip Campaign Best Practices ### Personalize the Messaging People hate getting junk mail in their digital inbox just as much as they hate regular snail mail junk. You have to make sure the messaging each lead gets feels directed right at them, with customization details that make it something they’re happy to receive. “From a messaging perspective, it’s best to make your drip campaign touchpoints as personalized as possible, from both a content and timing perspective, so you’re reaching the right person at the right time with the right message,” says Meagan Sweigart, principal and fractional marketing consultant at Kinetic Marketing Communications. ### Make the Emails Relevant Based on Timing Drip campaigns are anything but one-size-fits-all: They have to make sense based on myriad factors, including timing. “Drip campaigns are customized based on where the client is in their journey,” says Amra Beganovich, founder of Colorful Socks. “For holiday sign ups, we send an email series that includes gift ideas, bestseller highlights, and a final reminder prior to the shipping cutoff, for example.” ### Use A/B Testing Because drip campaigns by their very nature involve multiple emails (or other forms of connection), they provide marketers valuable data that can be used in making comparisons. “For instance, we can look at which subject lines or calls to action work best for our audience and then use what we've learned in future drip campaigns,” says Akdemir. “We can't get this level of control with more traditional marketing methods, which is why drip campaigns are such a useful tool.” ## Popular Software for Drip Marketing ### Constant Contact Constant Contact is a cloud-based digital and email marketing platform primarily catering to small- and mid-sized businesses, helping them build and manage their digital marketing strategies. It offers a range of tools, including email marketing, social media marketing, marketing automation, and lead generation, all integrated into a user-friendly interface. ### Mailchimp Mailchimp is an all-in-one marketing platform, particularly known for its email marketing capabilities. The company allows businesses to create, launch, and track email campaigns, offering features like customizable templates, automated workflows, and detailed analytics. ### HubSpot HubSpot is a versatile platform used to manage myriad aspects of a business, including marketing. HubSpot allows you to create and send email campaigns to engage your audience over time or based on triggers. ## Metrics to Measure Success of Your Drip Campaign ### Improved Open Rate If your email marketing campaign is working well, then those emails are actually being opened. Check the open rate, which is the percentage of people who actually open an email contrasted against the total number of people to whom it’s sent, and see if drip campaign emails enjoy better rates than those of other messages. ### Click-Through Rate Even more important than an email open rate is the click-through rate, often abbreviated as CTR. This is the percentage of people who take an action step, clicking on a link or button within the email, and being taken to a new page where conversion is possible. ### Conversion Rate The most important metric for a marketer is always the conversion rate, as this is the rate at which people actually take the action the marketer was hoping to prompt, such as a sale, a sign up for a subscription, joining a platform, and so forth. ## Key Takeaways “A drip campaign releases communication over a period of time based on timeliness or customer action,” says Beganovich. “These sequences can be initiated via certain behaviors, like cart abandonment, or they can be sent on a schedule, like a welcome series sent out over a few days. The idea is to lead them from awareness to purchase through clear messaging at spaced intervals that builds trust and relevance, not trying to stuff it down their throats.” Drip campaigns maintain brand awareness and audience contact without annoying potential leads like over-served ads, pop-ups, and other types of advertising. Because drip campaign emails (or texts or social media DMs) are automated, they save marketers time and money, as the marketer can create all of the media for the campaign, set the schedule and/or automations, and then be entirely hands-off, focusing on other tasks. ## Frequently Asked Questions (FAQs) ### Drip campaign vs. email blast: What’s the difference? A drip campaign is a series of pre-written emails sent automatically to a specific audience over time, or that are triggered by a specific action or event, like a newsletter sign up or an abandoned shopping cart. In contrast, an email blast is a one-time email sent to a large group of recipients at the same time, typically for a broad announcement or promotion. ### Drip campaign for abandoned cart recovery: How does it work? A drip campaign for abandoned cart recovery works by automatically sending a series of emails to shoppers who have added items to their cart but haven't completed the purchase. These emails aim to gently nudge the shopper back to complete the transaction, offering reminders, incentives like a discount, or by addressing potential issues they may have encountered during checkout. ### A/B testing for drip campaigns: How does it work? A/B testing in drip campaigns involves creating two or more variations of an email sequence and sending them to different segments of your audience to see which performs better. This helps you optimize your drip campaign by identifying which variations, like subject lines, copy, and images, lead to higher open rates, better click-through rates, and ultimately more conversions. --- ### Cookieless: What It Means, How It Works URL: https://www.taboola.com/marketing-hub/cookieless/ Last Modified: 2025-06-22 08:09:35 Discover what cookieless means, how it affects your marketing and advertising efforts, and what actionable steps you can take to thrive in a privacy-first world. For marketers, cookies are more than a tasty snack. They’ve long been a way to gather information on consumers: With the right technology in place, marketers using tracking cookies can monitor how users are interacting with their site and other sites across the web, then deliver targeted ads. But, in recent years, cookies have undergone intense scrutiny. Privacy restrictions are making it tougher than ever to monitor user behavior, and that’s driving a shift toward a cookieless future. As advertisers adapt to a cookieless world, what does that mean for your marketing campaigns? Let’s take a look. ## What Is Cookieless? Cookieless is a term often used in the context of “a cookieless future.” It refers to a digital environment where tracking users through third-party cookies is no longer an option. Cookies, which are small pieces of code that are stored on a website visitor’s browser, have become the top way advertisers track the users most likely to buy from them, but privacy concerns have reduced the viability of that strategy. “Cookieless refers to marketing techniques that do not depend on third-party cookies to track user behavior across multiple web pages,” says Amra Beganovich, founder of Colorful Socks. “It represents a move toward guarding user privacy and returning greater control to the consumer. Gone are the days of passively collecting data: Brands must now fight for their insights, getting directly involved with their audience.” ## The Impact of Cookieless on Marketing The move toward a cookieless internet has already had a noticeable impact on advertising campaigns. Brands that have previously relied on cookies for targeting their campaigns face new challenges. They include: - Audience targeting: Marketers have previously relied on third-party data to build “lookalike audiences.” Without the ability to gather that data, identifying and segmenting relevant users has become more challenging. - Attribution and measurement: Marketers now need to find ways to track user behavior without relying on cookies. This can make monitoring return on investment (ROI) tough. - Customer acquisition costs: If marketers can’t target as efficiently as in previous years, they may unknowingly be spending more money to acquire each new customer. ## Cookieless Tracking Methods As the reality of a cookieless future has become more apparent, advertisers have discovered alternative ways to track consumers for better targeting. They include: ### First-Party Data Collection As long as you’re respectful of privacy concerns, you can collect information on your own website visitors. You can incentivize customers to provide contact information and answer questions about their preferences. You can also monitor purchases and use that information for targeting. ### Contextual Targeting Contextual targeting displays ads based on the content on the page that surrounds it. This ensures that the people seeing the ad are interested in similar products or services. No data collection is required for this approach. ### Predictive Targeting The rise of artificial intelligence (AI) has coincided with the fall of third-party cookies, and marketers are leaning into the technology. Predictive targeting uses AI and machine learning to anticipate what a customer is most likely to do in the future. Built on years of data, this technology can accurately predict which users are most likely to convert and deliver ads based on that. ### Server-Side Tracking Some marketers are using server-side tracking to get around the need for third-party cookies. With server-side tracking, user data is sent to your website’s server rather than placing software on the user’s computer. This gives you more control over the data while also better aligning with privacy standards. ## Strategies for a Cookieless Future How can you continue to thrive in a cookieless world? Here are some strategies you can put into place. ### Invest in First-Party Data Infrastructure The best place to gather information on users is on your own website. Offer value to both existing customers and new website visitors in exchange for collecting information. Loyalty programs, exclusive content, and exclusive discounts can all be ways to encourage consumers to provide their information. ### Strengthen Creative Assets Marketers are leaning more heavily into creative these days, investing time and energy into ads that get results. Think beyond static ads and consider investing in video and carousel ads. Look for platforms that allow you to use a variety of effective ad types. Leverage Predictive and Contextual Targeting As AI continues to saturate every facet of advertising and marketing, using the technology is becoming essential if you want to remain competitive. The right solutions can analyze intent signals to predict future actions. Add contextual targeting to this strategy, and you can also ensure users see ads relevant to the content they’re currently consuming. Optimize with AI and Automation The more you can put on autopilot, the better. That frees you up to focus on other things. AI assistants can handle everything from creating ads to adjusting bidding to boost the likelihood of conversion. Use Privacy-First Metrics Marketers still need to measure, and that’s become trickier in a cookieless world. You can overcome this challenge by working with partners who make privacy a top priority in gathering data. ## Key Takeaways The term “cookieless” refers to an environment where third-party cookies are no longer an option. To respond to the drive toward a cookieless future, marketers must find ways to target customers while also respecting their privacy. To remain competitive, marketers must shift toward using first-party data, contextual targeting, and predictive technologies to target customers. ## Frequently Asked Questions (FAQs) ### Why are third-party cookies being phased out? In recent years, attention has shifted to protecting consumer privacy. Both regulators and consumers have called for protocols that protect data from marketers and advertisers. As a result, some services have begun blocking third-party cookies. While Google has backed off its cookie phaseout plans, the company is shifting toward other cookieless options, including allowing consumers to hide their IP addresses. “The phaseout is about shifting power back to the user,” says Jensen Savage, chief executive officer at Savage Growth Partners. “It forces marketers to be more intentional and respectful with data, which is better for user privacy, but has also proven to be a huge obstacle for marketers.” ### What are the alternatives to third-party cookies? Advertisers have access to other options for reaching consumers online. They include collecting data on visitors to your own website, displaying ads to match the content of the site the user is visiting, and using patterns to create predictive audiences. “The best alternative, in my opinion, is contextual advertising,” says Alex Smith, manager and co-owner of Render3DQuick. “The ad fits naturally with what the person is already interested in at that moment, without needing to know anything about who they are or where else they’ve been online.” ### How will cookieless tracking affect ad targeting? Ad targeting isn’t going away any time soon. Advertisers will merely need to shift the way they target users. Instead of tracking people across the web, marketers will now need to understand intent signals and context. Luckily, tools are available that use the latest technology to match creatives to the high-intent users most likely to find them relevant. “I tell clients to think of it as moving from a laser-focused sniper approach to a broader storytelling approach,” says Rodrigo César, CEO and co-founder at SSinvent. “Without third-party cookies, advertisers will have to lean more heavily on first-party data, look-alike audiences, and contextual signals rather than granular, individual-level targeting.” ### What is the role of contextual targeting in a cookieless world? In a cookieless world, contextual targeting becomes an essential part of an advertiser’s toolkit. Instead of focusing on a potential customer’s identity, you’ll need to look at what that person is doing in the moment. Thanks to AI-powered content analysis, though, this type of targeting has become easier than ever. Tech can be used to align ads to a user’s current consumption patterns, allowing for more precise targeting. “Contextual targeting is going to play a much bigger role now,” says Chris Coussons, CEO and founder of Visionary Marketing. “It’s something I remember using more in the earlier days of digital marketing, before we had all this granular tracking. Now it’s back in a smarter way. Pairing strong content relevance with well-designed creatives can still drive great results, especially if you understand your audience properly.’’ ### How can businesses collect and use first-party data effectively? Businesses have access to a variety of tools for collecting data from their own website visitors and customers. It’s important that the solutions you use to gather that data are privacy-compliant, but from there, you can incentivize people to willingly provide their information. “The key is value exchange,” Savage says. “People will share their data if you give them a good reason, like helpful content, exclusive perks, or personalized experiences. Once collected, the magic is in using that data to serve, not sell, your audience.” --- ### Engagement Rate: Importance, Calculation, Benchmarks URL: https://www.taboola.com/marketing-hub/engagement-rate/ Last Modified: 2025-05-05 09:30:06 You might think an ad being served to thousands of people on social media or displayed on a website where people linger on each page is an indication of marketing success, but prepare to be disappointed. Although there might have been plenty of ad impressions created while those ads were served far and wide, or were on display for extended periods, those efforts and ad spend might have been worthless if there was no user engagement. Audience engagement with online materials — in this case ads — can take many forms. A click, an upvote, a like, a share, a download, and other actions all count as engagement. When you track audience engagement rates, you can get an accurate sense of how well your ads are performing, or you can discover that people really aren’t responding to them, meaning you need to change course. ## What Is Engagement Rate? “Engagement rate tells you how much your audience is actually interacting with your content, not just scrolling past it,” says Michele Iapicco, CEO of Simplified Media Agency. “It’s usually calculated by taking the total number of likes, comments, shares, saves, and clicks, divided by your total impressions or followers.” Engagement makes the chance for conversion, be it a sale, a sign up, or a follow significantly more likely. When you know the rate of engagement, you can determine how your ads are performing. Over time, you can see if your engagement rate is improving or getting worse. ## Why Engagement Rate Matters “Engagement rate is one of the most important performance metrics in digital marketing, because it goes beyond reach and tells you how your audience is actually interacting with your content,” says Ryan Croy, founder of Public Haus Agency. “A high engagement rate usually means the content is resonating, which is essential for brand awareness, customer trust, and algorithmic visibility. Engagement rate gives you a truer sense of impact than follower count ever could.” Basically, engagement rate reflects how well your content resonates with your audience and points to meaningful interactions. A high engagement rate suggests that your content is valuable, relevant, and interesting to your intended audience, potentially leading to increased brand awareness, customer loyalty, and conversions. It also helps you understand what content resonates most with your audience and can help guide your content strategy. ## Engagement Rate Calculation “Engagement rate is calculated by dividing total engagement, which includes likes, comments, shares, saves, and so on, by your total audience or impressions,” says Croy. To calculate engagement rate, divide the total number of interactions — that’s those likes or shares and such — by the number of followers or impressions, and then multiply by 100. For example, if a post has 100 interactions and 1,000 followers, the engagement rate is (100 ÷ 1000) x 100, which equals 10%. ## Engagement Rate by Channel Different online platforms have very different engagement rates, and many of them are surprisingly low. ### Instagram Instagram is known for its visual focus, which often leads to higher engagement rates. Nano-influencers with 1,000 to 10,000 followers on Instagram can see engagement rates as high as 2.19%, while other influencers generally see rates between 0.8% and 1%, according to data sourced from Social Insider. ### TikTok TikTok's short-form video format also contributes to higher engagement. Popular with younger audiences, its videos can range from a few seconds to an hour. Its average engagement rate is around 4.07%, per data from Brandwatch. ### Facebook While Facebook has a large user base, its average engagement rate is lower compared to Instagram and TikTok. While some studies have shown that Facebook has an average engagement rate of 5.07%, other data suggests an average engagement rate of just 0.15% for Facebook posts, according to information from Social Insider. ### X (Twitter) Formerly known as Twitter, this platform was rebranded as X in 2023. X's real-time, news-oriented nature can result in lower engagement rates, with an average engagement rate of 1.6%, per data from Influencity. ### YouTube YouTube's engagement rate varies depending on subscriber count. Channels with 1,000 to 5,000 subscribers can see an average engagement rate of 5.60%, while those with 100,000 to 1 million subscribers have an average of around 2%, per InsightIQ. ### LinkedIn LinkedIn's focus on professional networking typically results in lower engagement rates compared to more visually driven platforms. Its average engagement rate is around 4.8% for LinkedIn posts, per information sourced from the platform itself. ### Social Media Engagement Rate As we see from the numbers above, social media engagement rates are surprisingly low. Even a 3% engagement rate for a platform like Instagram or TikTok would be considered quite good. That's not necessarily a commentary on the quality of advertising you are creating, but speaks more to the sheer volume of content on social media, and the more than 5 billion people using it worldwide. ### Website Engagement Rate People tend to engage more meaningfully with websites than they do with social media, which makes sense: While social media often involves mindless scrolling, website visits are usually made with more purpose. The average website engagement rate is generally considered to be between 60% and 70%, per data from Fathom Analytics. However, engagement rates vary significantly by industry and the specific types of content or features offered. Engagement with websites can be measured in a variety of ways, such as time spent on a page, bounce rate, scroll depth, and more. ### Email Engagement Rate The average email engagement rate, measured by open rate, is around 36.5% across all industries. Click-through rate (CTR) averages around 2.66%, per Mailchimp, and the click-to-open rate (CTOR) is typically between 10-25%, depending on industry, per Salesforce. Note that newsletters with a predictable weekly cadence have higher open rates than email blasts that occur at more random times. ## Engagement Rate Industry Benchmarks Engagement rate industry benchmarks can help businesses understand how their advertising performance compares to others in their field. Generally, a good engagement rate on social media is considered to be between 1% and 3.5%. However, this can vary depending on the specific platform, audience, and industry. For Facebook, for example, 1% is a good benchmark, whereas for Instagram, 3% is the goal. Note that engagement differs based on industry — here are a few examples of that: ### Finance For businesses in the finance industry, social media engagement tends to be around 1.48%, per data sourced from Influencity. ### Gaming Engagement rates for gaming brands are usually around 0.95% per MarketingCharts, with TikTok gaming content converting a bit better on average. ### Tech Tech brands usually see between a 1% to 3.8% engagement rate on social media, per SocialChamp. ### Food and Beverage Food and beverage companies have an average social media engagement rate around 1%, though brands with smaller followings tend to have more loyal and engaged followers, per data from Brandwatch. ### Fitness and Lifestyle Brands in the fitness and lifestyle spaces usually see social media engagement rates around 1.65%, per MarketingCharts. ### Travel Travel-themed brands usually have an average engagement rate of 1.5%, per Influencity. ### Automotive Auto-themed brands have an overall social media engagement rate around 0.06%, according to RivalIQ. ## Factors Affecting Engagement Rate Several factors influence engagement rates, including content quality, platform, posting frequency, audience relevance, and algorithm changes. The effectiveness of calls to action, the overall user experience, and even external factors like news events can also play a role. Here are some of the most common issues. ### Content Quality If the ads you’ve created aren't interesting or engaging, then of course the engagement rate will be very low. To be sure, even the best ads might go largely missed due to myriad factors, but it's a sure bet that advertisements with uninteresting copy, poor graphics, and a convoluted layout just aren't going to get shares, likes, or clicks. ### Post Frequency and Timing Consistent posting, at the right time of day and day of the week, can greatly influence engagement. That said, posting too frequently can overwhelm audiences, while posting too infrequently can cause content to be forgotten. ### Platform and Algorithm Different platforms have different algorithms and user behaviors. Instagram, for example, tends to have higher engagement rates than Twitter, and platforms like TikTok often have very high engagement rates for specific content types. Algorithm changes can also significantly impact engagement. ### Audience Relevance Understanding your target audience's demographics, interests, and behaviors is vital for creating content that resonates with them. You can find this data using tools like social media analytics or trend research tools like Google Trends. ## How to Improve Engagement Rate “Engagement rate is a measure of how well you’ve earned someone’s attention,” says Joel Luks, an adjunct professor with the University of Houston who teaches marketing and advertising. “Not just made them look, but made them care enough to do something about it and take action. It’s the difference between being seen and being relevant.” Improving your engagement rate means you have improved the way your audience perceives and interacts with your brand. ### Create Great Content It’s easier said than done, but if you can create high-quality ads, you will see a better engagement rate. Focus on compelling or amusing copy, attractive images and graphics, a clean layout, and clear and readily actionable calls to action. ### Be Consistent You need to be consistent in your advertising campaigns, both in the look and feel of the ads, and in the frequency and method of delivery. You don't want to run ads for several days, then go completely dark for a week, then reappear for one day, for example. Be consistent and you will steadily build better brand awareness. ### Segment Your Audience Audience segmentation is the process of dividing a large audience into smaller, more specific groups based on shared characteristics like demographics, behaviors, and interests. This allows advertisers to tailor their marketing efforts and messaging to make it resonate more effectively with different subgroups within their target audience. By understanding the distinct needs and preferences of various segments, companies can create more personalized and targeted campaigns, ultimately leading to improved engagement and conversion rates. ### Constantly A/B Test It’s vital to continuously test different aspects of your ads to see where improvements can be made. Take advantage of advertising platforms that offer AI-powered testing, which will let you optimize your creative in real time, leading to more effective results. ## Tools for Measuring Engagement Rate There are a few different tools at your disposal for measuring engagement rate, and you may want to try several out before settling on one — or you may want to use several! ### Engagement Calculators GRIN offers an influencer engagement rate calculator to assess the interaction rates of influencers. Hootsuite provides an engagement rate calculator for platforms like Instagram and TikTok. Phlanx features an engagement calculator to assess social media performance. Modash calculates Instagram engagement rates using a formula based on engagements per follower. Influencer Hero, HypeAuditor, and DashThis each offer free engagement rate calculators with various features. Some of these tools are free or very inexpensive, costing less than $10 per month, but those usually offer limited capabilities; more expansive plans can cost more than $200 monthly. ### Social Media Analytics Hootsuite provides analytics dashboards to track performance across social media networks, including engagement metrics, while Mailchimp provides tools and resources for calculating engagement rates, including metrics like likes, comments, and shares. ### Web Analytics Tools Google Analytics provides detailed insights into website visitor engagement and behavior, while Userpilot focuses on product engagement, using metrics like product adoption, stickiness, and growth to assess engagement. ## Key Takeaways Knowing your engagement rate lets you see how well your ads are performing, and where you might need to tweak and change your approach. “Engagement rate is one of the go-to metrics in social media,” says Darija Grobova, PR manager with Omnisend. “Basically, it tells you how many people interact with your content compared to how many saw it — think likes, comments, shares, and saves. But, having a high engagement rate doesn’t necessarily mean your content is helping your business — a post can go viral and bring you zero sales. Audience size can also alter the numbers; smaller accounts can often have higher engagement rates because their audience is tighter and, in turn, more connected.” Keep in mind that low engagement rates are common with social media, so don't be put off by numbers that seem quite low, even down in the single digits percentage-wise. On the other hand, website engagement tends to be much greater and meaningful conversions are possible there. The same is true with email marketing, where properly timed and well-executed emails can see much greater engagement. ## Frequently Asked Questions (FAQs) ### How do you calculate engagement rate on Instagram? To calculate your Instagram engagement rate, divide your total engagement numbers (likes, comments, shares, saves) by your total followers, and then multiply the results by 100. This gives you the percentage of your followers who have interacted with your content. ### What is a good email click-through rate? A good email click-through rate (CTR) generally falls between 2% and 5%, but it can vary significantly based on industry and email type. Some industries, like non-profits, may see higher average open rates and CTRs (40.04% and 3.27%, respectively), while others, like e-commerce, may have lower rates (29.81% and 1.74% respectively). A CTR of 20% or higher is considered excellent and is much higher than most industry averages. ### How important is engagement rate for SEO? Engagement rate is a very important factor in SEO — although to be clear, Google doesn't directly use it as a ranking factor. High engagement signals to Google that users find your content valuable, relevant, and satisfying, which can indirectly improve your rankings, though. By improving user experience and content quality, higher engagement rates can lead to better search engine visibility and authority. ### How does video content affect engagement rate? High-quality (and usually short-form) video content generally leads to higher engagement rates compared to other content formats, like text or images, because it's more immediately engaging and memorable. Videos are more likely to be watched and shared, and people can retain more information from them than from reading text. ### What is the difference between reach and engagement? Reach refers to the total number of unique users who see your content, while engagement refers to the number of interactions those people have with your content. Reach is about visibility, while engagement is about active participation. You can't have engagement without reach, but reach alone can mean very little if people don’t take action. ### How do you track engagement rate over time? To track engagement rate over time, you'll need to regularly calculate your engagement rate using the formula (Total Engagements ÷ Total Followers or Reach) x 100. Then, you'll track these engagement rates over time, ideally by creating charts or graphs to visually represent the trends. This allows you to see how your engagement fluctuates and identify any patterns. ### How does personalization impact engagement rate? Personalization significantly improves engagement rates by making the user experience more relevant, timely, and enjoyable, which leads to increased interaction, retention, conversion, and overall customer satisfaction. By tailoring content, recommendations, and interactions to individual preferences and behaviors, personalization drives users to engage more deeply with your brand and the products or services you’re promoting. --- ### Click-Through Rate: Importance, Benchmark, Platforms URL: https://www.taboola.com/marketing-hub/click-through-rate/ Last Modified: 2025-11-04 12:15:25 Getting people to view your ad when they’re doom-scrolling is not enough to boost a business' bottom line: Your audience needs to engage with content. That’s why the number of people who click on a link — or click-through rate (CTR) — is a crucial indicator of how many potential customers you have and how much money you may be leaving on the table. “Click-through rate is one of the simplest yet most powerful metrics in digital marketing,” says Matt Wilcox, digital marketing strategist at The SMB Strategist. Essentially, it shows how many people clicked on your content, versus how many people saw it. While that may seem pretty basic, CTR shows if you’re effectively grabbing consumer attention. “Think of it like a first impression,” Wilcox adds: If people click on your ad, they’ve engaged with it. It’s a way of saying they’re listening and interested in learning more. If they scroll past, it’s a good sign that you must change your marketing approach to achieve the desired results — more sales and conversions. ## What Is Click-through Rate (CTR)? Click-through rate is the number of people who click on a link in your advertisement or content. If 100 people see your ad and six click on it, that’s a 6% click-through rate. CTR comes into play during the early stages of a digital ad campaign. “Most campaigns work like a funnel,” Wilcox explains. First, someone sees your ad, then they click on it. From there, the goal is to get that person to take action, whether buying a product, booking a service, or signing up for a giveaway or newsletter. “CTR doesn’t tell you whether someone converted or bought anything, it only tells you if your ad made them curious enough to learn more,” he adds. ## Why Is It Important to Monitor CTR? Monitoring CTR is essential because it’s a crucial step in acquiring a new customer, also known as a conversion. Ultimately, a conversion is the end goal of the marketing funnel, but it’s impossible to convert new customers if they’re not clicking through in the first place. “That’s why CTR is most useful when trying to measure how effective your ad creative is,” Wilcox says. Click-through rate is not just a barometer for how compelling your messaging is, but also how effective your ad placement and targeting are, says Ethan Hartman, the founder of Mutewind Digital. “A higher CTR indicates your content is attention-grabbing and more people who find it in their search results are ‘clicking through.’” ## What Is a Good CTR? The simple answer is that a “good” click-through rate is between 2% and 5%, “But it really depends on the industry, the type of content, and where it's being shown,” adds digital marketing expert Sophie Musumeci, founder of Real Entrepreneur Women. Wilcox says there’s no “one-size-fits-all” answer because it varies by platform, industry, and competition. For example, on Google Ads, where users are already searching for something, “you might expect higher CTRs than on Facebook or Instagram, where ads are more of a distraction from scrolling,” he says. Likewise, a local business in a niche market might get a CTR as high as 30% if they’re the only ones running ads, whereas a company in a competitive market might be successful with 10%. “What matters more than comparing to others is tracking your own results over time and experimenting with different messages, images, and headlines to improve,” Wilcox says. ## Click Rate vs. Click-through Rate ### Click Rate Is a More General Term Although click rate and click-through are similar, click rate is often used in email marketing to indicate how many people opened what a company sent out. It gives businesses an idea of how many people saw their content, rather than how many people reacted to brand messaging by clicking on a link. ### Click Rate Is Necessary for Calculating Click-through Rate In order to know your company’s click-through rate, you must know how many people are viewing your emails, blogs, ads, or videos in the first place. CTR is calculated by dividing the number of people who click on a link by the number of people who saw the content, multiplied by 100. That percentage represents your click-through rate. CTR = (total clicks ÷ total impressions) x 100 ### Areas for Improvement While click rate and click-through rate are related, they can signal different growth opportunities. Wilcox notes that low click rates suggest that email subject lines are not achieving their intended effects, or that audience targeting is ineffective. They’re making a bad first impression, in other words. In contrast, a low CTR may indicate that the content could be more compelling, or the formatting may be off. Overall, businesses must know how many people are viewing their emails, blogs, videos, and ads, but it’s equally important to know how many people are interacting with them. So, these metrics are different, but similarly essential for improving your marketing approach. ## How to Monitor CTR: 4 Platforms To Know Of ### Social Campaigns: Facebook Ads Manager Monitoring your click-through rate depends on what platform you’re using. Facebook Ads Manager is an easy way to track CTR for Facebook content. By simply clicking on the performance tab, companies can access a dashboard that calculates the number of clicks divided by the total number of views, which determines the overall click-through rate. ### Search Campaigns: Google Ads Google Ads can be a similarly effective resource for tracking CTR outside Facebook. To access this and other metrics, click on the Campaigns section and then select ads. This will show you how content is performing, including the click-through rate. ### Email Marketing Campaigns: Hubspot and Mailchimp Email marketing platforms like Hubspot and Mailchimp have accessible dashboards where companies can track CTR and other metrics. Like Facebook Ads Manager and Google Ads, email marketing platforms allow businesses to identify patterns about what content resonates and where subject lines and other content should be tweaked. ### Performance Marketing Campaigns: Realize Beyond tracking your basic click-through rate, more comprehensive tools like Realize can give businesses more data on how to target your audience and optimize your marketing approach. Available under the Campaigns tab, the platform will tell you how many people are clicking on content links, who is clicking multiple times, and how long they’re staying on your landing page. This is valuable insight for retargeting ads and testing new headlines, calls to action, and other content. ## Key Takeaways Click-through rate, or CTR, is the number of people who click on a link in an ad, blog post, email, video, or other digital content, divided by the total views times 100. A high CTR shows companies that your first impression effectively engages your audience. A low CTR indicates a need to reevaluate subject lines, headings, formatting, calls to action, and other aspects of your creative. ## Frequently Asked Questions (FAQs) ### What is CTR and why is it important? CTR, or click-through rate, is the number of people who click on a link in your content, divided by the total number of people who view it, times 100. For instance, if 1,000 people read your email newsletter, and 50 people read it, that is a 5% CTR. Knowing your click-through rate is important because it shows how effective a brand’s first impression is, how many potential customers could be converted, and how much money companies may miss out on without optimizing their approach. By tracking CTR meticulously, businesses can test different content approaches to find what engages their audience. ### What if you have a high click rate, but low sales? Even if you have a high CTR, if those numbers are not translating to sales, “it’s a sign something is breaking after the click,” Hartman says. In other words, it’s an indicator that something at the bottom of your marketing funnel is not working. Musumeci agrees that a high click-through rate combined with low conversions is a signal that there is a disconnect between the content and the consumer. “Either your offer isn’t clear, your landing page isn’t converting, or your audience isn’t warmed up enough,” she says. For instance, sometimes landing pages can be slow to load, and a company can be losing conversions for reasons outside of brand messaging. Regardless, it’s vital to figure out what the disconnect is. As Musumeci says, “What matters most is what happens after the click.” ### Is a 6% click-through rate good? Generally speaking, a 6% CTR is considered good. In fact, once a business has reached this threshold, Hartman recommends focusing on conversion rate optimization. Again, this is especially important if sales do not match up with clicks. “Review your messaging, call-to-actions, and page layout,” he suggests. “Try different offers or page layouts. You can also split your traffic into groups and speak more clearly to what each group is looking for.” --- ### Cross-Device Attribution: How It Works, Benefits, Challenges URL: https://www.taboola.com/marketing-hub/cross-device-attribution/ Last Modified: 2025-06-22 08:10:21 Imagine you and your marketing team created a social media advertising campaign that you thought would be a smash hit, driving scores of customers to buy the latest version of a product. You launch the campaign, track lots of people viewing the ads on their phones, but then see those views just aren’t converting to sales. “We failed!” you might think with dismay. But, what if lots of the people you reached on a mobile device turned to their computers to do a bit more research, then went ahead with a purchase? Your ads worked, but because you weren’t tracking user activity beyond their phones, you never knew it. “The user journey consists of many different touchpoints,” says Josh Silverbauer, CRO at From The Future. “In 2025’s shopping landscape, people routinely switch between devices — from phone to laptop to tablet — as naturally as moving from dining table to sofa. The single-device, single-browser tracking model has long lost its accuracy. This is why implementing a tracking architecture that can connect user data across devices is crucial.” Here’s what you should know about cross-device attribution. ## What is Cross-Device Attribution? You have to track your users far and wide if you want actionable data about their consumer habits, and that means looking at all of their touchpoints. “Cross-device attribution tracks how a single user moves between devices before taking action,” says Alex Smith, manager and co-owner of Render3DQuick. “Someone might see a display ad on their phone in the morning, compare service options on a tablet later in the day, and submit a quote request on their work desktop the next morning.” “If each step looks like a different user, you lose the connection between cause and effect,” Smith adds. “It becomes harder to pinpoint which channel actually influenced the conversion. In performance marketing, that disconnect leads to inaccurate reporting and weakens your ability to make smart decisions about where to invest.” ## How It Works Cross-device attribution helps marketers understand a customer's journey across multiple devices and platforms by linking their interactions. It achieves this by associating a unique identifier with each user, which enables tracking of their behavior even when they switch devices. This allows for a more holistic view of user behavior and provides more accurate performance tracking of marketing campaigns. Unique identifiers like account logins, device IDs, or other methods allow marketers to recognize the same user across different devices and platforms. There are two main approaches: deterministic and probabilistic, which are explained below. ## Benefits of Cross-Device Attribution “Cross-device attribution uncovers the complete buyer's journey,” says Amra Beganovich, founder of e-commerce brand Colorful Socks. “A customer might encounter our ad on Instagram while scrolling on a phone, open an email later on a tablet, and complete a purchase on a desktop. Without cross-device tracking, we would attribute the sale to the last click, and miss the earlier touchpoints that had played a role in the sale.” ### Complete View of the Customer Journey Cross-device attribution provides a holistic view of a customer’s journey across myriad platforms and devices, showing how users engage with a brand as a complete picture. ### Touchpoint Value Analysis Not all touchpoints are of equal value for all customers or for all types of advertising media. Cross-device attribution lets you see what is working best on mobile, on desktop, on tablets, and beyond. You can get a sense of what ads convert best with various users across different platforms. ### Improved Message Targeting By better understanding customer behavior across devices, marketers can tailor their messaging and offers to individual users and their specific needs and preferences. ### Ad Fatigue Prevention When customers are hit with the same (or similar) ads over and over again, they can backfire, driving people away from a brand. By creating ads tailored properly for each touchpoint, businesses can avoid placing repetitive ads on multiple platforms, reducing ad fatigue and increasing engagement. ## Limitations of Cross-Device Attribution As with all facets of marketing (and life in general), there are always challenges to effective cross-device attribution, but if you know what to watch out for, you can minimize them. ### Privacy Concerns Any form of tracking a person can create both legal and ethical concerns. You need to make sure you are not violating any laws, such as GDPR (General Data Protection Regulation) or the CCPA (California Consumer Privacy Act). Keep in mind that some people choose to protect their privacy online, such as by using tracker-blocking browser extensions, which can make it harder for you to keep tabs on them. ### Data Gaps It is essentially impossible to track the entirety of a person’s online activity, so even with best practices in place, you still might miss some of the touchpoints of a user. You have to account for that as you analyze your data and acknowledge potential blindspots. ### Resources While you can automate a fair amount of cross-device attribution practices, implementing and maintaining cross-device attribution systems requires significant technical expertise, resources, and some ongoing hands-on management. ## Best Practices “Cross-device attribution matters because your dream clients are real people navigating real life — they’re scrolling on their phones between school drop-offs, checking emails on laptops during lunch breaks, and watching replays on tablets at night,” says Sophie Musumeci, founder of Real Entrepreneur Women. To be sure you don’t let a lead slip away, follow these best practices for cross-device attribution. ### Establish Your Objectives Clearly define what you want to achieve with your cross-device attribution efforts, whether it's increasing brand awareness, driving conversions, or improving customer retention. Then, set specific, measurable, achievable, relevant, and time-bound (SMART) goals and key performance indicators (KPIs) to track your progress. ### Map Out the Customer Journey As you collect data, consider touchpoints within the wider context. Map out many individual customer journeys, from awareness through to conversion and across platforms, so you can identify and study trends. ### Utilize Multiple Attribution Models Explore different attribution models like last-click, first-click, linear, time decay, and position-based models to see which best reflects your audience’s journey. Consider using a mix of attribution models to gain a more comprehensive understanding of customer behavior. ## What Are the Attribution Models Used in Cross-Device Attribution? As briefly noted above, there are two primary cross-device attribution models. Let's take a closer look at each of them now. ### Deterministic Attribution This approach relies on specific identifiers to link a user's activity across different devices. For example, if a user is logged into their Google account on their phone and computer, the system can track their activity on both devices. The benefit of a deterministic model is that it’s usually more accurate and precise. That said, this approach relies on having access to specific identifiers, which may not always be available or privacy-compliant. ### Probabilistic Attribution When direct identifiers are not available, probabilistic models use a broader range of data points to estimate the likelihood of a match. This might involve analyzing IP addresses, Wi-Fi networks, device types, location data, and online behavior patterns. Probabilistic models are useful when deterministic identifiers are unavailable, allowing for a more comprehensive view of user behavior across devices. But, keep in mind that this approach is less precise than deterministic modeling, and may sometimes produce incorrect data. ## How Can First-Party Data Be Leveraged for Better Cross-Device Attribution? As well as being less subject to wide-ranging privacy laws than third-party data, first-party data can significantly improve cross-device attribution by providing a more accurate and reliable view of customer interactions across different devices. By leveraging data collected directly from customers through your own channels, like websites and apps, you can gain a clearer picture of their journey and can attribute conversions more precisely. This leads to better insights into campaign performance, improved targeting, and more effective marketing strategies. ## Key Takeaways “Cross-device attribution is really about connecting the dots,” says Brian Kroeker, president of Little Rock Printing. “For example, from a customer first engaging with an Instagram ad on their phone, to browsing a site later on a tablet, to them completing a purchase later on a desktop. Without the right attribution in place, you risk giving credit to the last touchpoint while ignoring all the earlier ones that influenced the sale.” Or, on the other hand, you might miss a conversion and think your marketing efforts were unsuccessful. Proper cross-device attribution creates a full picture of the customer journey. It can make your ad spend more effective, improving return on investment (ROI), as well as helping with audience segmentation and targeting, and providing you with a wealth of data to analyze and act on. ## Frequently Asked Questions (FAQs) ### What does cross-device attribution mean? “Cross-device attribution is the process of understanding and tracking customer interactions across multiple devices and touchpoints in their buying journey,” says Michael Capote, CMO of German Car Depot. “Since customers are now moving between multiple devices such as smartphones, tablets, laptops, and smart TVs for their daily activities, it’s important to track their entire journey to achieve effective marketing measurement and optimization.” ### How can cross-device attribution help with frequency capping across devices? Cross-device attribution helps with frequency capping across devices by allowing advertisers to understand and track user interactions across multiple platforms. This information can help marketers implement a more comprehensive frequency cap that applies across all devices associated with a user. That way, users won’t see the same ad multiple times on different devices within the same household or user profile, which can otherwise cause ad fatigue. ### What tools and technologies are used for cross-device attribution? Cross-device attribution uses a variety of tools and technologies to track user behavior and attribution across different devices. Key tools include attribution platforms, data management platforms, and analytics tools from Google Analytics, Ruler Analytics Limited, HubSpot, Dreamdata, Adobe Analytics, and other companies. ### How does cross-device attribution differ from single-device attribution? Cross-device attribution focuses on tracking a customer's journey across multiple devices like phones, tablets, and computers, while single-device attribution only tracks interactions on a single device. Cross-device attribution offers a more complete view of the customer journey toward conversion, and allows for a more accurate assessment of campaign performance. Single-device attribution, on the other hand, may miss key touchpoints and provide an incomplete picture of the customer’s behavior. --- ### Content Management System (CMS): How It Works, Types, Best Practices URL: https://www.taboola.com/marketing-hub/content-management-system/ Last Modified: 2025-05-20 10:25:51 No matter what type of business you operate, having a well-organized website with clear, intuitive navigation is a must-have today. Behind most successful websites is a powerful content management system (CMS) that helps businesses and people create, manage, and modify content for the web with relative ease. ## What Is a Content Management System (CMS)? A content management system is a software that allows users to create, edit, organize, and publish digital content through a user-friendly platform without needing extensive coding knowledge. CMS platforms power a majority (68.7%) of all websites, with the CMS market expected to reach $28.09 billion by 2029, according to DiviFlash. Many CMS platforms have extensive libraries of pre-designed templates, plug-ins, add-ons, and external tools to help users manage content without the help of a web developer. A CMS separates content creation from the technical aspects of building a website: This empowers non-technical users to focus on creating valuable content while the system handles the behind-the-scenes functionality. Some examples of popular CMS platforms are WordPress, Shopify, Wix, and Squarespace, but there are hundreds of other options out there, depending on users’ needs. ## How Does a CMS Work? A CMS operates on a two-part system that makes website management accessible if you don’t have coding training. This includes: ### 1. Content Management Application (CMA) This is the user-facing portion of a CMS that allows content creators to add, edit, and delete content through an intuitive dashboard. It’s your site’s control panel where you write posts, upload images, and arrange your site’s content. ### 2. Content Delivery Application (CDA) This back-end component takes the content you’ve created, stores it in a database, and displays it according to your site’s templates when a visitor browses your site. When someone visits your site, the CMS retrieves the requested content from the database, merges it with the corresponding template, and displays the complete page to the visitor — all in real time. The separation between content and presentation ensures consistent styling while making content updates simple and efficient, without heavy lifting on the publisher’s part. ## Why Is a CMS Important? A content management system offers numerous competitive advantages for businesses, organizations, and individuals with a dedicated following. CMS platforms make web publishing accessible to the masses by removing technical barriers. Not a wizard at HTML, CSS, or JavaScript? No problem: A CMS eliminates the need for these skills for loading and updating content. This means your organization can save serious cash on development and maintenance costs. Additionally, most CMS platforms let you choose roles and permissions, allowing multiple team members to collaborate simultaneously with dedicated access levels. Plus, the templates and themes that come standard in most CMS platforms ensure uniform branding and design across all pages, maintaining a cohesive brand style and look no matter who’s creating the content. ## Pros and Cons of a CMS Pros Cons User-friendly interface that requires minimal technical knowledge. Performance lag compared to static sites, potentially reducing page load speeds. Collaborative workflows allow multiple users to work simultaneously. Security vulnerabilities if not regularly updated and properly maintained. Design consistency through templates and themes, and testing for device responsiveness. Plug-in conflicts that can cause compatibility issues or site errors. Content scheduling for automated publishing at optimal times. Customization limitations for highly specialized or unique functionality needs. Built-in SEO tools to optimize content for search engines. Vendor lock-in with some proprietary systems might make migration difficult and costly. Version control to track changes and revert updates if needed. Regular maintenance is required to keep the system secure and functional. Cost efficiency through reduced development and maintenance expenses. Learning curve for more complex CMSs like Drupal or enterprise-level systems. ## 5 Types of CMSs ### 1. Open-Source CMS Open-source platforms are available for anyone to use and modify. They typically have large community support and extensive plug-in options. - Examples: WordPress, Joomla, Drupal. - Advantages: No licensing fees, ability to tailor to your needs, community support. - Best for: Organizations with some technical resources or smaller budgets. ### 2. Proprietary CMS These CMSs are commercially developed systems that require licensing fees but often bake in ongoing professional support. - Examples: Adobe Experience Manager, Sitecore, HubSpot. - Advantages: Professional support, enterprise-grade security, integrated features. - Best for: Large enterprises with specific content requirements and dedicated budgets. ### 3. SaaS/Cloud-Based CMS These platforms operate completely online, with hosting and technical maintenance handled by the CMS operator. - Examples: Squarespace, Wix, Shopify. - Advantages: No installation required, automatic updates, managed security. - Best for: Small businesses and users who need a simple, no-frills solution. ### 4. Headless CMS A newer kind of CMS platform that separates content management from content presentation, allowing content to be used across multiple channels, delivering a high ROI on content production. - Examples: Contentful, Strapi, Agility CMS. - Advantages: Omnichannel content delivery, developer flexibility, and future-proofing. - Best for: Organizations with multiple digital channels (web, mobile, IoT). ### 5. E-Commerce CMS Specialized systems that cater solely to online stores with integrated payment processing and product inventory management. - Examples: Shopify, Squarespace, WooCommerce. - Advantages: Built-in e-commerce functionality, payment gateway integration. - Best for: Online retailers of any size. ## How to Choose the Right CMS in 8 Steps The CMS market is massive: There are over 800 CMS platforms globally, and more than 78 million websites use a CMS, according to Meetanshi. With so many options, choosing the best CMS entails careful consideration of your company’s needs. Here are some tips for narrowing down your choices: ### 1. Assess Your Technical Skills Be honest about your team’s technical knowledge and abilities. Some CMS platforms require users to be more tech-savvy than others. ### 2. Define Your Content Needs Consider what types of content you’ll publish, how often, and what special features you might want or need down the road. ### 3. Consider Your Budget Factor in not only the initial costs of the CMS, but also ongoing expenses for hosting, plug-ins, themes, and maintenance. ### 4. Explore Ease of Use The interface should take the guesswork out of the process for your content creators, especially if they have limited tech skills. Look for demos or trial versions to take the CMS for a test run. ### 5. Check Customization Options Ensure the CMS can grow with your business and audience to accommodate future needs (such as payment integration, for example). ### 6. Research Security Features Security should be a top priority, especially if you’ll handle sensitive personal or financial information from customers or site visitors. ### 7. Assess SEO Capabilities Built-in SEO tools don’t always help your site’s visibility in search results. Clarify what the CMS’ SEO tools can actually do, how they work, and how much technical SEO maintenance may be required. ### 8. Consider Integration Requirements The CMS should play nicely in the sandbox with other tools your business uses, such as a CRM or marketing automation software. ## CMS Best Practices ### Content Organization - Create a logical content hierarchy that makes content navigation simple for site visitors. - Use consistent naming conventions for pages and assets. - Build an easy-to-use taxonomy with relevant content categories and topic tags. ### Security Measures - Keep your CMS and all plug-ins and themes updated. - Use strong passwords and two-factor authentication for all admins and content creators. - Have regular content backup procedures in place. - Consider security plug-ins or services for additional protection. ### Performance Optimization - Optimize images before uploading to avoid loading or display issues. - Use caching plug-ins or features. - Minimize unnecessary plug-ins, which can impact page load speeds. - Choose quality hosting that can handle your site’s traffic. ### Responsive Design - Choose templates that work well on all device types and sizes. - Test your site regularly on different devices. - Consider mobile users in your content creation process. ### Content Quality Control - Create clear workflows for content creation and approval. - Document style guidelines for consistency. - Implement regular content audits to ensure high quality. - Define user roles and permissions appropriately. ## Key Takeaways A CMS separates content creation from technical implementation, making website content management more accessible to non-technical users. Choosing the right CMS involves balancing tech requirements, ease of use, customization needs, and budget. ## Frequently Asked Questions (FAQs) ### What are the advantages of using a CMS? The key advantages of using a CMS to power your website include ease of use, collaborative workflows, consistent design, lower web development costs, improved SEO capabilities, and scalability. Modern CMS platforms increasingly rely on AI-powered features such as generative content creation and predictive analytics to round out their offerings. ### What are the disadvantages of using a CMS? Potential drawbacks of a CMS include performance overhead compared to static sites, security vulnerabilities if not properly maintained, feature limitations on some platforms, and potential for plug-in/compatibility issues. Some organizations might also end up needing extensive customization features that go beyond a CMS’ usual capabilities. ### What are the most popular CMS platforms? WordPress remains the most popular CMS by a long shot, powering 61.3% of all CMS-based websites. Shopify ranks second at 6.7%, followed by Wix at 5.2%, according to W3Techs. ### How do you choose the best CMS for your needs? To choose the right CMS for your needs, define your requirements, including content types, tech capabilities, budget, and future growth plans. Consider how easy the platform is to use and rework, security, and integration features. Always request a demo or a trial period before making a final decision. ### Is a CMS necessary for all websites? Not all websites need a CMS. Simple websites with sporadic updates might be better served by static site generators or hand-coded solutions. However, the overall trend is moving toward CMS adoption: The percentage of hand-coded websites fell from 76% in 2011 to just 33% in 2022, according to WP Beginner. --- ### Display Ads: Definition, Benefits Beyond the Banners URL: https://www.taboola.com/marketing-hub/display-ads/ Last Modified: 2026-06-25 09:41:37 The digital landscape right now is massive and overwhelming, to say the least. It’s a vast and endless ocean of information and interactive opportunities pulling users in all directions, so attracting the eyeballs (and clicks) of your target audience can feel especially challenging. While search engines help focused users find what they're actively looking for, what about capturing their attention during passive browsing, app use, or gameplay? That's where display advertising comes in and can be enormously helpful, directing people to your product, site, app, or business. As a copywriter and content writer for nearly two decades, I’ve seen the evolution of display ads and what they’re capable of, and the clever ways advertisers work around their limitations (more on that here). On the surface, some of them may still look the same as the early 2000s — just a rectangle rotating through slides — but the work that goes on behind it is completely different, and would be disorienting to even think about 20 years ago. For those just starting out learning about display ads, or for people even older than me: If this were the pre-internet days, think of display ads as the digital equivalent of billboards or magazine ads, only much more strategically placed, and zeroing in on the right audience as much as possible. The results speak for themselves: In 2024 alone, spending on display ads topped over $300 billion, and made about 14 billion impressions. Display ads, especially nowadays, have gone beyond proving themselves as a useful tool for building brand awareness, driving website traffic, and generating leads. So, what exactly are display ads in detail, how do they work, and most importantly, why should marketers care? I’ve taken a deep dive into the ins and outs of display advertising, its history, various forms, and overall potential, as well as how they can help your business. I’ve also enlisted the help of two experts for this piece: Rodrigo Cesar, digital marketing specialist/CEO at SSInvent, and Mimi Nguyen, founder and head of marketing of Cafely. ## What Are Display Ads? At their most basic core definition, display ads are a type of online advertising that combines visuals like images, videos, and animations along with text to share a carefully crafted marketing message. But, unlike the text-only ads that pop up next to search engine results based on specific keywords, display ads usually appear on websites, apps, and videos through various ad networks. These networks, one of the biggest being the Google Display Network, link advertisers to a wide range of websites to showcase their ads to users. “As the CEO of a digital marketing agency, I spend a lot of time helping clients grow their visibility online, and display ads are one of the foundational tools I often use to make that happen”, says Cesar. “They're essentially visual ads — banners, squares, pop ups, even video — that show up across websites, apps, and platforms. If you’ve ever seen an ad at the top of a news site or in the sidebar of a blog, you’ve seen a display ad in action.” “They work by targeting you in a variety of ways,” says Nguyen, “like who you are, where you’re located, or what websites you’re browsing — even showing you things you’ve already looked at before. As soon as you click on one, you’re taken straight to the advertiser’s page.” For a real-world example, let’s say you're reading an article about the best running trails in your state. A banner ad for a new brand of running shoes catches your eye at the top of the page. That’s a display ad doing its thing, cleverly using the context of your browsing to show you a product that might just pique your interest. They know you’re already reading about running, specifically looking at trails, and the next logical step would be some good shoes, so a captivating display ad is designed to draw you in. It’s not a guaranteed win for them, though: You may already have running shoes you’re happy with, tune out the ad, and keep reading the article. Still, if the display ad gets in front of the eyes of multiple thousands of runners, there’s a good chance some are going to hit that combination of needing new shoes, clicking through, and ultimately buying a pair — or may in the future as the ad follows them around the web. “What makes display ads powerful is their reach,” Cesar adds. “They’re not about waiting for someone to type something into Google. Instead, they help brands show up where their potential customers already are.” ## How Do Display Ads Work? “These ads are placed through networks like Google Display Network or Facebook’s Audience Network and are often targeted based on behavior, interests, demographics, or past website activity,” says Cesar. Though a banner ad may look simple (it was, after all, one of the earliest forms of internet advertising), a display ad has quite an epic journey from advertiser to your screen, involving both technology and marketing strategy. “I use them to raise awareness, build familiarity, and retarget people who may have visited a client’s site, but didn’t take action,” Cesar continues. “It’s like giving your brand a second or third chance to be remembered.” Here's a brief breakdown of the process: ### Advertiser Creation First, the advertiser designs their display ad, which can be an image, animated GIF, video, or some other sort of creative interactive element. They also clearly define their target audience based on factors like demographics, interests, browsing behavior, and website visits. ### Ad Network Connection The advertiser uses an ad platform, such as Google Ads, or an advertising service to link up with an ad network. These networks collaborate with endless websites and apps that have set aside space for showing ads. ### Bidding and Matching When a user visits a website within the ad network, an ad auction often takes place in milliseconds. Advertisers bid on the opportunity to show their ad to that specific user based on the targeting criteria they've set. The ad network's algorithm then determines the winning bid and the most relevant ad to display. ### Ad Delivery The winning ad is then served and displayed in the designated ad slot on the website or app that you’re currently viewing. ### User Interaction If the user finds the ad compelling and clicks on it, usually they’ll be directed to the advertiser's landing page, where they can learn more about the product or service and potentially make a conversion (generally a purchase, sign-up, or download). It's a fascinating process, and not just for us marketing nerds. The whole experience is frequently automated via programmatic advertising, where technology manages the buying and selling of ad space instantly, using sophisticated algorithms and data insights. Tools like Realize leverage a dedicated performance AI to simplify this process, ensuring your display ads target the right audience and are tailored on the fly to enhance performance. Realize, in fact, analyzes vast amounts of first-party data to understand user intent and engagement patterns, allowing advertisers to craft ever more relevant and effective display ad experiences. ## Types of Display Ads Display ads come in a variety of shapes, sizes, and formats, each designed to capture attention in different ways. “There are several types I work with, depending on a client’s goals,” says Cesar. “Traditional banner ads are still widely used, but responsive display ads — which automatically adjust in size and layout depending on where they appear — are becoming more popular because of their flexibility. I also rely heavily on retargeting ads, which follow users who have previously engaged with a site or product.” Here are the core types of common display ads you’ll most likely encounter: ### Banner Ads Probably the first thing that comes to mind when you hear the words “display ad,” these are the classic horizontal ads that typically appear at the top or bottom of a webpage. They come in various standard sizes, the most common ones being: - Skyscrapers: Tall, vertical ads that run along the sidebars of websites. - Square and Rectangle Ads: Versatile ad units that can be placed in various locations within a website’s content. - Leaderboards: Wide banner ads that are often placed at the very top of a webpage. - Half-Page Ads: Vertical ad units that can take up a lot of space and be visually impactful. - Mobile Banner Ads: Smaller ad formats generally optimized for display on mobile devices. ### Rich Media Ads These ads go beyond static images and incorporate interactive elements, animations, video, or audio to create a more engaging experience. They’ve evolved a lot over the years and provide opportunities to play around with the format and make it more fun, innovative, memorable, and even viral. ### Video Ads These are short video commercials that can play within or alongside website content, or within video players. “Video display ads have gained a lot of traction, especially on platforms like YouTube and in mobile apps, where video performs extremely well,” says Cesar. ### Interstitial Ads Full-screen ads that appear between page loads or app interactions. ### Native Display Ads While closely related to native advertising (which blends seamlessly with surrounding content — more on that later), native display ads are still served through display ad networks, but are designed to match the look and feel of the website they appear on. ## Pros and Cons of Display Ads Like any marketing tactic, display advertising is going to have its ups and downs. “Display ads come with both advantages and challenges,” advises Cesar. “On the positive side, the visual nature of display advertising means we can create eye-catching creative that sticks in people’s minds. The reach is massive, and when used for retargeting, display ads can be incredibly efficient at bringing back potential customers who need a little nudge. On the downside, users have become very good at tuning out ads.” “Another issue,” he continues, “is that people aren’t always in a buying mindset when they see display ads, so conversions may be lower compared to search ads. And, of course, ad blockers can limit visibility altogether, depending on the user.” Here’s a breakdown of the benefits and setbacks: ### Pros ### Broad Reach and Awareness Display networks can have an extensive reach, allowing advertisers to connect with a wide audience across a whole multitude of websites and apps. This makes them an excellent option, not just for building brand awareness, but also reaching potential customers who might not be actively searching for your products or services (yet). “Reach would be the number one upside — you just can’t beat being in front of 90% of all internet users,” says Nguyen. “Smart targeting is another good one, as you can zero in on exactly who you want to reach. As someone who loves playing around with picture and video editing, I love the creative freedom I get with working on Cafely’s own display ads.” ### Visual Storytelling One of the big benefits of display ads is showing instead of just telling. Rather than shoving slides in front of users’ eyes, these allow for more compelling storytelling, and the showcasing of products and services in a more engaging way. Using capabilities like images, videos, and animations can capture attention and convey key points faster, not to mention being more visually stimulating than text alone. It can be a powerful, memorable way to make an impression, and one that creates a positive connection to your product. ### Precise Targeting Options One of the biggest ways modern display advertising differs from the early days is how platforms offer more sophisticated targeting capabilities. Advertisers can now reach increasingly specific demographics to pinpoint the audience they want. It’s not just the basics like age, gender, and location, but interests, behaviors, even retargeting users who previously interacted with their website. This helps ensure that your ads are shown to the most relevant audience, especially if the ad platform you’re working with uses AI that allows for targeting by intent, rather than just identity. ### Cons ### Banner Blindness “‘Banner blindness’ means that if the ad isn’t truly relevant or engaging, it’s likely to be ignored,” Cesar points out. Over time, users (myself included) have become incredibly skilled at tuning out banner ad placements. Banner blindness can significantly reduce the effectiveness of display ads if they aren’t visually compelling and strategically placed. ### Potentially Lower Click-Through Rates (CTR) Compared to search ads, display ads often have much lower CTRs, often between 1% and 3%, and even that’s considered really good, as it’s usually less than 1%. Users browsing content may not be in a buying mindset at the moment they see your ad, and the visual distraction of ads can sometimes be more of a pain than an invitation to click. It’s the same feeling as trying to have a conversation with someone who's engrossed in a book — you might need a really compelling opening line or they’re going to ignore you, or get annoyed and form a negative association with your product. ### Ad Blocking Software “Banner blindness is a real challenge,” says Nguyen. “But, there’s also ad blockers, so there’s a decent chance some people will never see your ads at all. Plus, placement problems could mean your ad might show up next to content you don’t want.” While ad blockers were once just an application that your tech-savvy friend had, it’s now become extremely common. A significant segment of users have ad-blocking software running all the time, which can be a huge hurdle for advertisers to get around, as ad blockers prevent display ads from even being shown. Needless to say, this can severely limit the reach of your campaigns, and force you to look at alternative strategies to connect with your audience. ## How to Create Display Ads: Best Practices “As someone who creates display ads for clients daily, I always start with the basics,” says Cesar. “Make it visual, make it fast, and make it clickable.” ### Size “I usually stick with the most effective and widely supported formats like 300x250 (medium rectangle), 728x90 (leaderboard), and 160x600 (skyscraper),” says Cesar. “Responsive display ads have become a go-to because they adapt to the screen size — essential for today’s mobile-first browsing behavior.” Nguyen agrees, and says that “there are a few golden rules to keep in mind: First is that size matters, literally. Best to stick to standard sizes, because they show up in more placements and tend to perform better.” ### Copy This is the area that’s been the center of my working life for a long time, and one thing I can say with absolute confidence is: Keep it short and simple. “You have about one to two seconds to make an impression,” agrees Cesar. “I write headlines that speak directly to a pain point or a desired outcome. Something like ‘Tired of Slow Wi-Fi?’ or ‘Grow Your Business 3X with SEO’ works better than generic slogans. The body text supports the headline with a benefit or value proposition, not fluff.” “It needs to be ultra-clear,” adds Nguyen. “No need to try to be clever. I go for ‘scannable and strong,’ like ‘Bold flavor, zero crash’ because that tells you exactly what to expect from our coffee.” ### Call to Action (CTA) This is what wraps it up, and is the final push to drive a user to click. “Your CTA should be clear and action-oriented — ‘Get a Free Quote,’ ‘Shop Now,’ ‘Learn More,’ or ‘Claim Your Offer,’” suggests Cesar. “I always test different CTAs to see what resonates, but the golden rule is, don’t make the user think. Make it obvious what they should do next.” “It should pop,” says Nguyen, “both visually and contextually. Saying something like ‘Try Now’ or ‘See Flavors’ will perform better than the generic ‘Learn More’ buttons.” ## How to Target the Right Audience With Display Ads Casting the widest net possible may feel like a logical first instinct, particularly on a limited budget. But sometimes, less is more. Though the number of eyes on your ad may be smaller with increased targeting, it’s often much more efficient, and worth it in the long run. “We’ve had moderate success using contextual targeting — placing ads on sites that already attract coffee lovers, wellness readers, or busy professionals,” says Nguyen. “It’s not just about who they are, though, but what they’re doing in the moment. Behavioral targeting is another go-to: Showing our ad to someone who’s already visited our site or searched for similar products.” “For a health food brand,” says Cesar, “we targeted users reading fitness blogs and nutrition forums using affinity and in-market audiences. We combined that with keyword targeting for paleo and keto diets. The CTR increased by 40% once we aligned the creative with the content they were already consuming.” ### Tools to Use With the evolution of display ads and ad networks, there’s a lot of tools to choose from. That can be daunting for a newcomer, but it’s best to break it down by what your specific goals and needs are. “I worked with a SaaS company that wanted to reach startup founders,” recalls Cesar. “We used LinkedIn’s Audience Expansion tool in combination with Google Display placements on business news sites like TechCrunch and Fast Company. By narrowing the audience to specific job titles and interests, we tripled conversions compared to a broader campaign.” ## How to Retarget With Display Ads Didn’t get someone to click through and make a purchase the first time? All hope isn’t lost. With retargeting, you get a second (and often third) opportunity to present your product again, from a new angle, with altered copy and images. “Though retargeting might seem redundant to some, it’s been one of our consistently effective tools,” says Nguyen. “There was one particular time when we did a campaign targeting people who left items in their cart — just a simple nudge with a limited-time discount as incentive — and our CTRs nearly doubled.” ### Adjusting Your Ad “Retargeting is where display ads really shine,” says Cesar. “We helped an e-commerce brand recover 22% of abandoned carts just by showing users dynamic product ads of the exact items they left behind. We also added urgency with copy like, ‘Only 3 left in stock!’” “We showed display ads to visitors that had already browsed our site but didn’t buy,” recalls Nguyen. “We showed them the exact blend they were looking at and it gave our conversions a significant boost.” ### Utilizing the Retargeting Funnel Cesar provided the following example for making the best use out of a retargeting funnel. “For a B2B client, we created a three-stage retargeting funnel: Visitors who didn’t fill out a form saw a reminder ad within three days, a testimonial-based ad a week later, and a time-sensitive offer ad at the two-week mark. This layered approach led to a 54% increase in lead quality.” ## How to Measure Display Ad Performance Metrics are your guiding light through this entire process, and tracking the right metrics is absolutely crucial to understanding the effectiveness of your display campaigns. “When I’m running a campaign,” says Cesar, “I always pay close attention to several key metrics to gauge how things are going. Click-through rate is important — it tells me if people are even engaging with the ad at all. But, I also look at conversion rate to see if those clicks are actually turning into sales or leads. Cost-per-click and cost-per