Program-Level Signals Tell CTV Buyers Which Shows Ran Their Ads And Which Ones Paid Off

Contextual targeting in CTV works only when buyers know which show their ad ran on. Most of the time, they only know the app. Buyers can no longer rely on data about who is watching, so they increasingly target based on what’s on screen. A single streaming app can air a live game, a recap show, and a documentary on the same day. Program-level data tells buyers what their ads ran next to and which programs drove results.

Chris Chandler is Head of CTV and Audio Partnerships at Peer39, which provides contextual and brand suitability data for programmatic advertising. He works with streaming and podcast publishers to bring program-level content signals into the company’s data, then helps their sales teams and media buyers understand how those signals appear in demand-side platforms. Before joining Peer39 in 2024, he built and led the first strategic accounts team at Nexxen’s supply-side platform and worked in business development at SpotX, where he partnered with CTV media owners including MLB and Fox. More than a decade on the sell side of programmatic video gives him a clear view of where app-level targeting falls short for buyers.

"Buyers should be able to inspect the content, trust the description, and then learn from where their ads ran," Chandler says. Many buyers can’t do that yet, because program-level data is missing or labeled differently from one publisher to the next. Closing those gaps decides whether contextual can earn a bigger share of CTV budgets.

Details that change the buy

The signals that guide a CTV buy start with format. Buyers want to know whether the content is a film or a TV episode, whether it’s live or on demand, and what genre it falls under. Ratings and content warnings help them judge brand suitability. With sports, a buyer may also care which sport is airing, whether it’s the game itself or a pre- or postgame show, and whether the broadcast is live or a replay from the day before. Each of these signals can change what an impression is worth to a given advertiser.

Live events don’t always end on schedule, and a time slot can’t tell a buyer what’s airing once a game runs long. "If a game goes into overtime and you’re buying that inventory after the scheduled time slot is over, is it still the live feed of the game or is it the postgame show? Things like that can make a big difference," Chandler notes.

One movie, two genres

These details only help if the labels match from one publisher to the next, and often they don’t. Something as small as episode numbering can break targeting and reporting, with one publisher writing "Season 1, Episode 2" and another writing "S1E2." Genre labels vary even more, partly because publishers may license the same title from more than one content provider, each sending its own metadata. "One example was Braveheart. A streamer might get it from two different content providers. In one it’s listed as an action-adventure film, and in another it could be a period piece or religion. Those little nuances make it hard for the publisher. They don’t know which ones to use," Chandler says.

Some publishers also receive incomplete metadata from their providers, face contract limits on what they can pass along, or carry labels with errors. Third-party data providers can fix this by standardizing signals so the same program reads the same way in every DSP, and by checking that content matches its label before buyers can target it. Standardized labels also make it easier for publishers to report on their inventory and sell it directly.

That work only helps inventory that has labels at all. Roughly 60% of CTV bid requests still contain no usable program-level signal, according to Peer39’s June 2026 report "Getting FAST Right." Publishers and data providers have made progress over the past two years, but covering the rest of the market will take time.

Reading the results

Where program-level signals exist, reporting shows buyers what they bought. Buyers new to contextual often start by copying the buys they already run elsewhere. Aggregated reports then list the programs and apps where their ads appeared, and those lists can turn up surprises, such as mobile app inventory the buyer thought was CTV. The reports also reveal which signals were useful and which were too generic to help. When buyers tie that data to conversions or other outcomes, they know what to change. “It’s difficult to make those optimizations without those signals,” Chandler says.

What buyers do with the reports depends on their goal. A campaign built for reach can favor low-cost impressions across broad programming. A campaign built for conversions needs to know which content drives action and which content only fills a delivery quota. "A buyer might find that a particular type of programming consistently performs well while another is just cheap inventory. Better content reporting lets buyers adjust where those dollars go," Chandler adds.

The gap between seeing results and acting on them is also getting shorter. Buyers have usually learned how their ads performed on each program after a campaign ends, too late to change that campaign. Some agencies are now testing integrations that send those results back into the DSP while a campaign is still running, so spending can change before it wraps. Program-level data then guides each bid as well as the report at the end.

Contextual targeting in audio trails CTV by a year or two, but audio may catch up faster. Audio publishers already make much of their content data public, while CTV publishers tend to keep tighter control of theirs for business or contract reasons. The audio market is also more concentrated, with four or five publishers dominating and two of them already passing contextual signals. For podcasts, contextual data can go down to the individual episode, so a show that doesn’t suit every brand overall can still offer an episode with a doctor as a guest that fits a health food company.

Buying is changing as well. Some publishers are building AI-based tools that let a buyer type in what they’re looking for and receive a deal with contextual data already attached. This way of buying could become common within the next year or two, for streaming shows and podcast episodes alike. "As more decisions get automated, the quality of information fed into those decisions matters even more," Chandler says.