Connect with us

NEWS

The 2026 CTV Forecasts the Market Is Now Scoring

IAB’s September outlook scores December’s CTV forecasts on AI discovery, agent trading, and a thinner idea of premium inventory.

Published

on

IAB raised its 2026 U.S. connected TV ad growth forecast to 15.6% on September 10, after a sports-heavy first half. The same day, Nielsen’s July Gauge put streaming at 49.0% of U.S. TV time, with YouTube at a record 14.2%.

Those figures now score a set of December 2025 calls from CTV and ad-tech operators who had already named AI discovery, data quality, and a thinner definition of premium as the year’s levers.

IAB Lifts 2026 CTV Growth to 15.6%

The Interactive Advertising Bureau raised 2026 U.S. ad growth to 12.3% in its September Outlook Study, up 2.8 points from the 9.5% it published in January. The revision rests on more than 200 brand and agency buyers and on a first half that IAB says was lifted by the Winter Olympics and the FIFA World Cup.

IAB SEPTEMBER 2026 CHANNEL GROWTH CALLS

Channel September forecast January forecast
Social media 16.5% 14.6%
Connected TV 15.6% 13.8%
Commerce media 13.6% 12.1%
Digital video excluding CTV 9.4% 9.6%
Podcasts 8.7% 8.6%
Paid search 8.1% 8.2%
Digital out-of-home 7.0% 7.4%
Linear TV -1.5% -1.7%

Linear TV is the only major channel IAB still expects to shrink. Digital out-of-home, which AdOmni’s Jonathan Gudai had cast as a creator stage for 2026, was the one digital line the bureau cut from its January call, to 7.0% from 7.4%.

The first half was strong, major live events delivered, and advertisers have increasingly powerful tools in their arsenal to find and engage customers.

David Cohen, CEO, IAB, September 2026 outlook

Cohen also warned that the economy still holds “real areas of uncertainty” and that “there are no easy wins.” Customer acquisition jumped nine points from January to 63% of media goals. Brand equity rose six points to 43%. Repeat purchases sat flat at 24%.

What the December Forecasts Got Right

On December 23, 2025, operators from Yahoo DSP, Ampersand, Media.net, Nexxen, Digital Culture Group, Swivel, and AdOmni laid out the year. The September print does not bless every line. It does sort them.

THE DECEMBER BETS, SCORED

  • CTV spend: IAB’s CTV call is now 15.6%, 1.8 points above January, so the growth bet held.
  • Live sports: Yahoo DSP’s Joshua John, head of strategy, had called sports the biggest CTV fight; 47.4% of his respondents said bundling would define it, and 34.5% expected worse fragmentation first. The World Cup Final on July 19 and the Winter Olympics are the events IAB now cites for the first-half lift.
  • FAST as a business: Nearly 60% of John’s respondents already treated free ad-supported streaming as a premium ad model, not a funnel. That view aged well as streaming’s share of TV kept climbing.
  • AI discovery: John’s survey found 75% of executives expected OS-level AI assistants to decide which shows and services appear on TV home screens. The 2026 fight that buyers actually funded is brand visibility inside chat tools.
  • Agent trading: Swivel’s Frans Vermeulen said agent-to-agent deals would open paths beyond classic programmatic. The protocols shipped. The default auction did not vanish.
  • Premium as attention: Nexxen’s Oscar Rondon said attention would replace clicks and impressions as the performance signal. Pause ads and the TV home screen are now sold on that claim.
  • Voice as a breakout buy: Digital Culture Group founder Crystal Foote said voice-activated targeting would break out in 2026. Shoppable voice prompts still exist. Microphone-based targeting took a public hit.

Maria Vilchez Lowrey, chief growth officer at Direct Digital Holdings, had said 2026 might surprise on overall spend while consolidating around social, CTV, and search. The September table matches the first two. Paid search, at 8.1%, is the quiet line in that trio.

Discovery Shifted Into Chat, Not the Home Screen

John’s 75% figure was specific. It was about TV home screens, not search bars. Apple has since pushed Apple Intelligence into on-screen brand discovery, and Siri’s next models were described as built with Google’s Gemini. That is an OS assistant. It is not the thing IAB’s buyers now rank as their top media problem.

Adapting to AI-driven search is the leading investment challenge, cited by 44% of those buyers. Low-quality AI-generated content, which IAB labels “AI slop,” follows at 38%. Optimizing content for AI-generated answers is the No. 1 area of increased focus, at 76%, ahead of LLM models at 72%. Use of generative AI inside media campaigns actually slipped, to 69% from 78% in January.

Marketers have spent a lot of time asking, ‘How can I use AI?’ Now the question is becoming, ‘How do I reach a consumer who uses AI?’

Chris Bruderle, Vice President, Industry Insights and Content Strategy, IAB

Bruderle said people are discovering products and judging brands inside those environments, and that marketers want to know how a brand shows up in the conversation. 86% of buyers already are changing, or expect to change, how they measure media because of conversational AI tools and AI agents in the next 12 months. The most common response, at 48%, is to measure brand visibility and citations inside the AI tools. Forty-five percent still cannot compare that path with the journeys they already know how to count.

Nielsen’s July Gauge, released the same day as IAB’s outlook, is the viewing half of that split. Streaming rose to 49.0% of total TV, up 0.5 points, with usage up 3.4% from June. YouTube’s 14.2% share of TV was a record, on viewing up 6% month over month, and its lead in the Media Distributor Gauge widened to 5.0 share points. FOX finished July at 7.8% of TV after the World Cup run. NBCU-Versant reached 9.2%. Peacock sat at 2.6%. Disney’s streaming properties held 4.7%.

John had also found that nearly half of his respondents already treated TikTok and YouTube as younger viewers’ primary TV, and that 38.3% said streamers had to copy creator-driven models. IAB’s September list of ad types getting more focus puts creator and influencer partnerships first, at 54%.

The Agent Layer Inside Programmatic

Vermeulen’s 2025 call was that buy-side and sell-side agents would negotiate new deal types instead of only bidding into the old auction. The industry did not skip the stack. It wrapped agents around it.

THE AGENT STACK IN 2026

  1. January 6, 2026: Yahoo DSP puts native agents into planning, activation, troubleshooting, and audience discovery at CES, under a “Yours, Mine, and Ours” setup that lets advertisers bring their own models, use Yahoo’s, or mix both through MCP and APIs.
  2. May 2026: Prebid.org ships an AdCP adapter so publishers can take agent bids without rebuilding the whole sell side.
  3. June 18, 2026: Yahoo opened the Yahoo DSP Agent Network, an open frame that plugs partner agents into audience, activation, creative, and measurement workflows.
  4. September 22, 2026: IAB Tech Lab’s Agentic Advertising Management Protocols page lists AAMP 2.0 deal types (programmatic guaranteed, preferred deals, PMPs, open marketplace, linear GRPs), AAMP 2.3 production-stack claims, AAMP 3.0’s OpenProposal brief spec, buyer and seller agent SDKs, and a public Agent Registry.

Adam Roodman, GM of Yahoo DSP, said agentic AI “should make advertising simpler, not harder, and that starts with openness.” The same June release said agent-to-agent capabilities were already in development. That is the honest score of Vermeulen’s bet: the pipes exist, and they still sit inside DSPs, SSPs, and trade-body schemas.

Karan Dalal, COO of Media.net, had said SSPs would become decision engines that filter impressions before the auction, and that transparency would turn from hygiene into a measurable edge. AAMP’s registry and “show, not tell” deal record are the institutional version of that argument. Independent analyst Eric Seufert put the other half in a September 23 post, walking a pair of San Francisco tech brothers through agentic commerce until the dialogue lands on a single line: “Everything is an ad network.” The buy side is learning the same lesson. Agents that skip the exchange still have to sell attention. The platforms that already sell it are building the agent layer so the path still runs through them.

Nexxen Sells Pause Ads and Home Screens

Nexxen’s December note was that CTV had a supply surplus, so “premium” had to be recalibrated, and that attention would replace raw impression counts. The company’s 2026 product map is that argument in inventory form.

In April it became the first exchange partner to put automatic content recognition, or ACR, into Google’s Instant Deals, so Display and Video 360 buyers could request audience or content forecasts and turn on ACR data inside Display and Video 360 without a custom deal cycle. Through May it extended TV home-screen ads beyond VIDAA to TCL FFALCON and TiVo Ads, selling the banner a viewer sees before any app opens. Rondon has described that screen as a stretch of “pure, intentional attention,” with viewers spending up to 10 minutes deciding what to watch. In September the company opened pause ads, the still that replaces a frozen frame when someone hits pause, to programmatic buying through its SSP and outside DSPs.

That is a different premium than a marquee show. It is a moment: power-on, pause, the minutes spent hunting. If the 2025 complaint was too much undifferentiated CTV, the 2026 answer from at least one large seller is to productize the seconds when the viewer is actually looking.

Why Data Quality Still Beats New Tools

Justin Rosen, Ampersand’s SVP of data and insights, said 2026 would be the year data dominates TV advertising, as marketers stop fussing over tools and start fussing over datasets, device graphs, and ID links. IAB’s September shopping list reads like his brief. Demo-targeted and cohort-based advertising is getting more focus from 53% of buyers. Publishers with first-party data sit at 48%. Contextual sits at 45%.

Lucas Bertrand, CEO of Looper Insights, had been blunter in December. He said the real gain would come from fixing genre mislabels, messy metadata, opaque supply paths, and brand-suitability blind spots, “issues that no amount of AI hype can mask.” Pixalate’s Q2 2026 benchmarks, drawn from more than 80 billion programmatic impressions, put invalid traffic at 25% of U.S. open programmatic CTV. That is the clog Bertrand described, still sitting in the pipe after a year of agent demos.

Rosen’s other 2025 call was that vertical video would become the open web’s new front door, after a Media.net study found 90% of consumers open to vertical video on publisher sites. That remains a format bet, not a spend line IAB broke out in September. The budget shift IAB did record is toward creators, first-party publishers, and context, which is another way of saying buyers will pay for a cleaner claim about who saw the ad.

Dalal’s transparency point sits on the same seam. As identity thins out, he argued, sell-side decisioning becomes a form of addressability, scored on context, intent, and quality rather than user IDs. The buyers who told IAB they want publishers with first-party data are making that trade with their budgets, whether or not they use his vocabulary.

Voice Targeting Hit a Credibility Wall

Foote’s December forecast treated voice as the next contextual channel. People were already asking Alexa for a recipe, queuing a podcast on a smart speaker, or checking weather through Siri, and she said those utility moments were turning into ad touchpoints. If brands lacked infrastructure to retarget from voice commands, she said, they were behind.

Shoppable CTV prompts, the “say this to add it to a cart” overlay, are still in the market. The uglier cousin of that idea is not. On August 27, the Federal Trade Commission finalized consent orders with MindSift LLC, Cox Media Group, and 1010 Digital Works LLC, with $930,000 in payments among them. Sales copy had claimed an algorithm listened through phones, televisions, and smart speakers to catch pre-purchase talk. The commission found the product never captured any voice data.

That case does not disprove Foote’s shoppable-CTA path. It does explain why “voice targeting” arrived in 2026 with a credibility problem attached. Foote’s other December line, that cultural intelligence had to sit inside the AI rather than in a review layer, is the piece she actually shipped. She has described a nine-metric Resonance Convergence Coefficient Score and an Audience Resonance Index that runs a brief through the firm’s models before a campaign goes live, scoring platform fit, cultural fit, and vernacular.

The remaining 2026 job, on IAB’s own survey, is measurement. Eighty-six percent of buyers expect to change how they count media because of conversational AI and agents. Forty-eight percent say they will measure citations inside the tools themselves. Forty-five percent still cannot line that path up with the journeys they already know how to count. That gap, not another forecast deck, is what the rest of the year still has to close.

Harry is the editor and publisher of MIND CRON, an independent title built on ten years of journalism that took him from the reporter's notebook to the editor's chair. Breaking news is where his rules are strictest. A story goes out when the primary document is in hand or two independent sources confirm the same fact, and not before, however loud the rumour. Anything still moving is labelled as developing, each update carries the time it was made, and the original wording stays visible so readers can see what changed. That discipline applies whether the story is a market shock in business, an outage in technology, a result in sports, a launch in gaming or a recall in auto, and it is no looser for science, entertainment, lifestyle, travel or the wider news pages. Numbers are checked against the source before publication. Errors are corrected openly under a public corrections policy. Tips from readers are checked the same way as everything else, and Harry reads and answers that mail himself at support@mindcron.com.

Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Trending