Unlocking Contextual Targeting Opportunities on CTV
A version of this piece ran in Streaming Media on March 19, 2024.
TV is getting digitized at an awkward moment on the advertising timeline. Advertisers can finally target the big screen with the precision they got used to on the open web, and the tool they trusted for that precision, the third-party cookie, is on its way out. Inconvenient. Also a forcing function. If you can’t follow the person, you have to understand the content.
That’s contextual targeting. Match the ad to what’s on screen, not to a profile of who might be watching.
I’ve spent the last few years building that stack at VideoByte and now as GM of CTV at Kargo, so I’m not a neutral observer. I also think the case against behavioral ads was never the whole story on living-room TV. A phone is personal. A TV is often a room. Behavioral targeting can be right for one person on the couch and wrong for everyone else in the shot. Contextual can be relevant to the group, because it’s relevant to the show.
It also does work behavioral targeting was never designed to do. Brand safety and suitability become questions about the content, which is the question brands actually ask. You can insist on aligning with scenes and tones that make sense for the product, instead of hoping an audience segment doesn’t wander into the wrong pod. Attention follows too. An ad that belongs next to what you’re already watching has a head start on one that followed you in from a different app.
The objections are real
Contextual CTV is still early, and the hesitation isn’t imaginary.
Without shared content identifiers, buying context across partners is messy, and comparing outcomes is worse. The IAB and others are pushing standards. Standards only work if buyers and sellers actually use them. Brand safety is also blunter than people want when you only get show-level labels. An hour of mostly fine content can still have a commercial break after a scene some brands will not touch. Scene-level understanding is the difference between “avoid this title” and “avoid this moment.”
Then there’s the access problem. On the open web, you can analyze a page. On CTV, publishers lock the stream. Direct integrations become the tollbooth, which slows everyone down and keeps advertisers from getting smarter. Publishers aren’t villains for that. They’re protecting CPMs and trying not to let buyers cherry-pick the premium moments while leaving the rest. There is a compromise available. It requires negotiation, not a lecture about openness.
Stop treating CTV like big display
A lot of the “contextual doesn’t scale on CTV” talk assumes CTV is display advertising with a larger screen. It isn’t. Quality CTV inventory is still scarce enough that publishers can afford to collaborate, and the format itself allows things display never did.
AI is making the content side sharper: more nuance than keywords and genre tags, closer to what we built for scene-level matching. Virtual product placement can pick the moment a product belongs in the frame. Formats that sit beside the content instead of stopping it, what we called Moments at VideoByte, give contextual a different surface than a pre-roll pod. A truck ad in the browse results for action-adventure, a fashion ad next to a reality slate, an L-bar that doesn’t kick you out of the episode. Cookies trained the industry to chase the person. The interesting work on CTV is happening around the program.
None of that erases the hard parts. Standards, access, measurement, the politics of who gets to read the stream. I won’t pretend those are solved because a byline needs an optimistic last line.
I will say this: the death of the cookie is not a temporary outage on the old targeting model. It’s a chance to build a better one for a screen people watch together. Contextual is how you start.
David Naffis is General Manager of CTV at Kargo.