How AI Is Transforming Streaming Adtech

ArticleID

176403 - click to preview View Count: 1453

Title

Author(s)

Steve Nathans-Kelly (5483)

Alternate URL

Standout URL

Images

Note

Image Size for Social Media Use
1200 x 628 pixels

Make Top Featured Article

Article Helper

Related Issues

Issue Helper

Summary

AI&rsquo;s potential to transform CTV and OTT ad preparation, ad serving, ad placement, contextualization, video analysis, performance, and reporting is fairly well understood and rarely understated. But how are streaming platforms, content owners, and advertisers actually operationalizing AI today? <a href="https://www.linkedin.com/in/aliajdaniels/" target="_blank" rel="noopener" title="Revry&rsquo;s Alia J. Daniels">Revry&rsquo;s Alia J. Daniels</a>, <a href="https://www.linkedin.com/in/crystal-foote/" target="_blank" rel="noopener" title="Digital Culture Group (DCG)&rsquo;s Crystal Foote">Digital Culture Group (DCG)&rsquo;s Crystal Foote</a>, <a href="https://www.linkedin.com/in/oph%C3%A9lie-boucaud-572a07bb/" target="_blank" rel="noopener" title="Dataxis&rsquo; Ophelie Boucaud">Dataxis&rsquo; Ophelie Boucaud</a>, and <a href="https://www.linkedin.com/in/sargeway/" target="_blank" rel="noopener" title="Sargeway&rsquo;s Sarge Sargent">Sargeway&rsquo;s Sarge Sargent</a> discuss how publishers are leveraging AI&rsquo;s advantages and also addressing some of its limitations in this clip from <a href="https://www.streamingmedia.com/Articles/News/Online-Video-News/Instant-Replay-Streaming-Media-Connect-August-2026-176152.aspx" target="_blank" rel="noopener" title="Instant Replay: Streaming Media Connect August 2026">Streaming Media Connect 2026</a>.<h2>Operationalizing AI at Revry: Preserving Nuance, Protecting Privacy</h2>Framing the discussion around the impact of AI integration on the growth of <a href="https://www.streamingmediaglobal.com/Articles/Editorial/Short-Cuts/What-Is-Contextual-Advertising-and-How-Is-it-Changing-TV-166275.aspx" target="_blank" rel="noopener" title="What Is Contextual Advertising and How Is it Changing TV?">contextual advertising</a>, and the ways LLMs and <a href="https://www.momentslab.com/blog/multimodal-ai-and-media-assets-the-future-of-content-intelligence" target="_blank" rel="noopener" title="Multimodal AI and Media Assets: The Future of Content Intelligence">multimodal models</a> are reshaping the ways CTV ads are prepared, sold, and delivered, Boucaud asks Daniels, "From <a href="https://www.revry.tv/" target="_blank" rel="noopener" title="revry tv">Revry</a>'s perspective, what have you been implementing? What have you been experimenting with?""For us,"Daniels replies, "it's a combination of where can we operationalize utilizing AI without losing the nuance that still requires still a human touch. AI is fantastic at a lot of things, and I think there's still a bit of room to grow. But because I'm coming from this perspective of we are targeting a specific, multicultural audience, we operate from a <a href="https://www.streamingmedia.com/Articles/Post/Blog/How-Streaming-Services-Can-Deal-With-Consumer-Privacy-Concerns-148522.aspx" target="_blank" rel="noopener" title="How Streaming Services Can Deal With Consumer Privacy Concerns">privacy-first</a> place. With anything that we're doing, we start from there. After that, it's about, how nuanced can AI be with regards to certain cultural contexts that it may not be fully aware of? So we're still really focused on utilizing a human touch when we're talking about specific content and how our advertisers are serving against that content.""It's still a moment where we need a lot of calibrating," Boucaud agrees, "especially for more <a href="https://www.streamingmedia.com/Articles/Editorial/Featured-Articles/Mining-for-Riches-in-the-Niches-Dynamic-Approaches-for-Targeted-Growth-172817.aspx" target="_blank" rel="noopener" title="Mining for Riches in the Niches: Dynamic Approaches for Targeted Growth">niche segments</a> that may not be so well catered to by more generic models."<h2>Quantifying Cultural Relevance at Digital Culture Group</h2>Turning to Foote, Boucaud asks, "Crystal, can you tell us a bit more about how you've been experimenting with" AI in contextual advertising at <a href="https://www.digitalculture.group/" target="_blank" rel="noopener" title="digital culture group">DCG</a>, "especially with brands? What is the outcome so far?""Our engineering team created a nine-metric system, which is now our currency, called the <a href="https://www.digitalculture.group/resonance-convergence-coefficient" target="_blank" rel="noopener" title="DCG Resonance Convergence Coefficient Score">Resonance Convergence Coefficient Score</a>," Foote explains. "Within that nine-metric system, we're looking at platform relevance, cultural relevance. We're looking at vernacular when it comes to certain audiences to make sure that before we even go live with a campaign, we take that brief or that RFP, we put it into our AI platform, <a href="https://www.digitalculture.group/post/what-is-audience-resonance" target="_blank" rel="noopener" title="Audience Resonance Index (ARI)">Audience Resonance Index (ARI)</a>, an aggregator of all the different APIs that we have for our AI and our new models that we created to provide intelligence for audience targeting, which we didn't have before. We were just an ad tech company, and we were putting together different data points and data taxonomies to create these audiences and find them in the digital ecosystem. But now the SaaS platform we created using AI is making more informed decisions and the outcomes are incredibly better because of it."Boucaud concurs that older approaches to ad content matching "were based more on assumptions and hypothesis and understandings of what an audience would like, but without really grounding it in data. That's a really interesting change."<h2>AI as Scene-Analysis Workhorse</h2>Boucaud then shifts the conversation in Sargent's direction and broadens the question a bit: How is AI making the whole ad tech industry evolve?" Earlier in the discussion, she notes, he mentioned the increasing role of AI in "making sure we understand the context of a scene or the mood. How do we do that? What is AI bringing in terms of in video analysis that we were not able to touch before?""Before, everything was done manually," Sargent recalls. "So what's actually changed is the workhorse. Now you see AI involved in multiple different layers from the ad creative side and creating ad assets or dynamically creating them out of existing libraries of assets, so combining multiple different previous campaigns into one on the fly."Sargent illustrates his point with an example from a current project. "I'm working with a client now who is launching an AVOD platform, and they're using <a href="https://www.streamingmedia.com/Articles/ReadArticle.aspx?ArticleID=175220" target="_blank" rel="noopener" title="Elevating Content Verticalization With Emotion-Aware AI: A Q&amp;A With Vionlabs' Arash Pendari">Vionlabs</a> to introspect each thematic asset and break it down. Not only does it pick out where the cues should be placed, but also [identifies] the tone and sentiment at those ad markers. One of the things that's missing is what Crystal's company does: adding in cultural significance and faith-based significance and also vernacular. So that I see is missing in what I'm looking at today that Vionlabs is pulling out." Moving on from the "content prep side" to the "ad server side," Sargent continues, "ad servers are now adding into their ecosystems the ability to pull out context or to accept context. This is important so that they know exactly what's in the actual content that the user is looking at. When they get to the ad break," he explains, this makes it possible to "match the context at that ad marker and then use that to make a decision." Referencing contextual advertising technology developed when he worked at Disney, Sargent says, "Disney has this down pat. While I was there, they launched what they call <a href="https://press.disneyadvertising.com/disney's-magic-words:-connecting-advertisers-to-the-moods-and-moments-in-premium-content" target="_blank" rel="noopener" title="disney Magic Words">Magic Words</a>, which is basically contextual keywords about the content, and then adding that into the ad server since they have their own ecosystem. That's all fair game for them. But if you're using <a href="https://www.streamingmedia.com/Articles/Post/Blog/Boosting-Ad-Fill-Rates-with-Advanced-Ad-Prefetching-A-Solution-Overview-166521.aspx" target="_blank" rel="noopener" title="Boosting Ad Fill Rates with Advanced Ad Prefetching: A Solution Overview">Google Ad Manager</a> or <a href="https://www.streamingmedia.com/Articles/ReadArticle.aspx?ArticleID=171627" target="_blank" rel="noopener" title="AI and Contextual Advertising: A Q&amp;A With Comcast Advertising&rsquo;s Kristin Shumaker and FreeWheel&rsquo;s Larry Allen">FreeWheel</a>, you need a way to signal that ad server what's actually in your content. Then at that ad break, it signala 'this is the context of the scene either before the ad break or the scene after the ad break' so the ad server can make that decision." Moving on from the ad server realm to what he describes as "the marketplaces," Sargent says, "now on the ad agency side, they're using AI to pull out the context in the actual ad asset itself. And then you need to package that and deliver that so that the ad servers in the marketplaces know exactly what's in your ad. So it's not just a Coke ad, but it's a Coke ad about X, Y, and Z, or 'here's the tone of that Coke ad,' so it should match the tone that the ad server wants to serve into."<a href="https://www.streamingmedia.com/Conferences/Connect2026-November/Default.aspx" target="_blank" rel="noopener" title="Join us November 9&ndash;11, 2026 for more thought leadership, actionable insights, and lively debate at Streaming Media Connect 2026! Registration is open!">Join us November 9&ndash;11, 2026 for more thought leadership, actionable insights, and lively debate at Streaming Media Connect 2026! Registration is open!</a>

Authors