I recently spoke with Josh Heller, chief revenue officer at Coactive AI, about how AI is changing the way publishers and advertisers understand video and images. His central idea is straightforward: content can be understood as data, then used to make advertising more relevant. Instead of relying only on broad labels, publishers can analyze what appears in their libraries and make those insights useful to marketers. That could help buyers find moments that fit a campaign while giving publishers new ways to package inventory.
Finding More Value in Publisher Content
Publishers are looking for ways to make more from the audiences they still reach. Josh sees an opportunity in the long tail of inventory that often gets sold indirectly. With a more detailed understanding of their content, publishers could package it around themes advertisers care about, not just broad categories or genres.
A campaign might be looking for moments of happy families, social diversity, or female empowerment. Finding those moments across a library could make existing inventory more valuable and help buyers find content that fits their goals.
Josh also sees content intelligence helping beyond ad sales. It could support research, recommendations, and marketing by helping publishers understand what’s resonating and why. The same insights that help sell inventory could help a publisher promote its content and make better decisions across the business.
Context Adds to Audience Data
Audience data can tell us who’s watching, but it doesn’t necessarily tell us whether an ad will connect. People with similar demographics may be in very different mindsets. The content they choose (and the moment they’re watching) can provide another useful signal.
That doesn’t mean choosing between audience and context. Josh’s point is to use them together. Marketers can test how different audiences respond to different creative in different content environments, then use the results to improve both placement and creative. Rather than assume demographics tell the whole story, they can learn which combinations actually perform.
AI Can Scale Contextual Advertising, but Trust Matters
There’s a practical challenge: understanding video at a granular level takes time. Josh described AI analyzing frames, dialogue, on-screen text, music, and sound effects, as well as how those elements change over time. That can help publishers respond to contextual requests much faster than a manual process allows.
But speed isn’t enough. Buyers need confidence that the content they’re purchasing matches their requirements. Josh put it bluntly: “The seller’s been grading their own homework.” A verification layer could help buyers check whether a placement fits their definitions of relevant, suitable, or safe.
That distinction matters. A program about sports isn’t necessarily live sports, and an old broadcast of a major event may not be what an advertiser intends to buy. The more precise the campaign, the more important it is to agree on what the signals mean, and to verify them.
Start with Creative Variations
One-to-one advertising may be the aspiration, but marketers don’t have to jump straight from one ad for everyone to a custom ad for every viewer. Josh suggests starting with a manageable number of creative variations and testing them against different audiences and content moments.
That approach gives teams a practical way to learn what works before making bigger investments. If a particular creative performs well with a certain audience in a certain context, that’s useful guidance for future placements and for developing the next round of ads.
Treat Content as a Data Asset
Josh’s broader prediction is that publishers will increasingly treat content as a data asset, not just something to distribute. Advertisers could define the kinds of moments they want and find them across publishers and platforms. That will take collaboration among publishers, agencies, and brands, but it could make content easier to understand, verify, and activate.
The opportunity isn’t simply to add AI to advertising. It’s to bring content, audience, and creative signals together, and use them to make advertising more relevant for viewers and more valuable for the businesses involved.
Discussion Points Include
- Finding value in long-tail publisher inventory
- Combining content signals with audience data
- Building buyer trust through verification
- Testing creative variations against content moments
- Treating content as a data asset



