Yahoo on October 22, 2026 published a blog post arguing that the way most connected TV advertising is now bought leaves advertisers with far less control than the word "programmatic" implies. The post, titled "CTV decisions belong to the advertiser" and filed under the company's enterprise blog without an author byline, makes the case that programmatic guaranteed deals should stop being the default for streaming TV budgets, and that biddable private marketplace deals should carry most of the spend instead.

In Short

Yahoo published a blog post on October 22, 2026 arguing that advertisers who buy connected TV through fixed-price guaranteed deals hand most decisions about who sees an ad, how often and at what price to the seller. It matters because, citing eMarketer, more than 89% of programmatic video is bought through direct deals, and Yahoo says the guaranteed version of those deals wastes budget on heavy viewers and repeated ads that most streaming viewers notice. Yahoo proposes moving 20% of a CTV budget into a test of biddable private deals, while its own supporting figures come from unnamed advertisers and Yahoo internal data.

Two deal types under one label

The post opens with a statistic that, on its face, suggests the TV market has fully adopted automated buying. According to Yahoo, eMarketer reported in July 2026 that 93.8% of CTV ad spending now transacts programmatically. The source is eMarketer's "US Programmatic Advertising Forecast and Ad Tech Trends H2 2026."

The second figure is the one Yahoo builds its argument on. More than 89% of programmatic video, according to the same eMarketer report as cited by Yahoo, is programmatic direct. That category, Yahoo argues, hides two very different ways of buying. "It can mean no auction at all, meaning none of the precision and efficiency that made programmatic worth the name in the first place," the post states. Or the decisions can still sit with the buyer, "just exercised behind a more closed-door approach with the publisher of your choice."

Yahoo separates the two in plain terms. A private marketplace, or PMP, is described as "an invite-only deal with a publisher you chose, at rates you negotiated." It remains biddable. "Your algorithm decides impression by impression: bid, bid less, or pass entirely," according to the post. A programmatic guaranteed deal, or PG, is "fixed price, fixed volume, committed in advance." Under PG, Yahoo writes, "Publishers require DSPs to bid on almost all bid requests," which leaves "limited advertiser choice and control."

There is a gap in the numbers worth noting. The 93.8% figure covers CTV, while the 89% figure, as Yahoo presents it, covers programmatic video as a whole. The post does not say how programmatic direct spending divides between PMP and PG, which is the split its argument depends on. Other data points in different directions. FreeWheel's report for the first half of 2026 put guaranteed deals at 27% of programmatic ad views in the United States against 71% in Europe, measured on FreeWheel's own platform and by ad views rather than spend. The two sources measure different things, so neither confirms nor contradicts the other directly.

Yahoo's central claim about inventory is that the choice between PMP and PG is rarely about access. "Outside sponsorships and tentpole events, most publishers now allow programmatic buying that is actually biddable, even against guaranteed spend commitments," the post states. "The gap between PMP and PG isn't inventory quality or premium access. Often it's the same inventory from the same publisher. The gap is who is in control of the buy." The post names no publishers and gives no count of how many allow biddable buying against guaranteed commitments.

Seven decisions, and who makes them

The most pointed section of the post is a list. Yahoo asks which of seven decisions a buyer actually made on the last CTV campaign: "which household saw the ad, how many times, how recently, whether linear had already reached them, what data defined the audience, what that impression was worth, and whether to buy it at all?"

On a PG deal, according to Yahoo, "most of those sit with the seller." The post draws a line between planning and buying. "Choosing the publisher and negotiating a CPM is good planning, but it's not advanced buying."

Without impression-level decisions, Yahoo argues, budgets drift toward the households that watch the most television. "Heavy TV viewers absorb the majority of impressions, so they take the bulk of your budget by default, and you may miss out on reaching net-new households," the post states. It offers a concrete illustration: an advertiser may buy three slots in the same program. "That does nothing for the brand and nothing for efficiency, it just irritates someone who already saw you."

The cost of repetition

Yahoo frames the damage in two ways, spend efficiency and customer experience, and borrows a term from Omnicom Media Intelligence to describe the second. Omnicom calls the tipping point "negative reach," which the post describes as "the moment repeated impressions start frustrating people and damaging the brand instead of building it."

The supporting figures come from Omnicom Media Intelligence's April 2026 paper "Why Frequency Matters: Combating Negative Reach," as cited by Yahoo. More than 60% of consumers report seeing the same ad multiple times in a single streaming session, according to that research. More than half say they would pay extra to avoid it. The post does not give the sample size, the market surveyed or the fieldwork dates.

The complaint is not new. Frequency capping has long been available inside individual platforms, but a cap set separately in each publisher's system cannot see exposures delivered by other publishers. That is the gap Yahoo says biddable deals can close.

Four controls Yahoo ties to private marketplaces

The post lists four capabilities it says PMPs make possible and PG deals largely do not.

Household-level frequency across publishers is the first, "rather than a separate cap sitting in each publisher's silo." Yahoo concedes the dependency directly. "This one depends entirely on durable household identity. Without it, cross-publisher frequency is an estimate dressed up as a control."

Linear suppression is the second: steering impressions away from homes a TV campaign already reached and toward incremental reach among CTV households.

Algorithmic pricing is the third, so that "the bid reflects what a given household is worth to you instead of one flat rate for every impression, regardless of how many times that household has seen your ad."

Supply path optimization is the fourth, which Yahoo describes as "choosing the cleanest route to the impression rather than inheriting whatever path the deal came with." The supply path a bid request travels determines how many intermediaries take a fee before money reaches the publisher.

The identity caveat in the first point carries more weight than the post gives it. Household-level controls across publishers are only as reliable as the identifiers that connect a television in one app to the same television in another. An Adstra and InterMedia study reported in July 2026 found that just 23 percent of residential IP addresses reached their intended geographic target, and that device-level CTV identifiers were 71 percent consistent month over month against 57 percent for IP addresses. Yahoo does not say which signals underpin its household matching beyond naming its own identifier, ConnectID.

The numbers behind the case

Yahoo offers two customer results, neither with a name attached.

The first concerns a consumer packaged goods advertiser that, according to Yahoo, shifted to biddable supply and "cut its CTV eCPM in half." The post describes the comparison as "Same advertiser, same brand, comparable premium-TV mix. The buying model was the variable." The footnote cites Yahoo internal data from August 2026.

The second is the figure in this article's headline. "Another advertiser reached 60% more households for 32% less investment year over year," the post states, describing it as "Same buy, same publisher mix, but with much greater control of frequency and the ability to suppress linear viewers." This claim carries no footnote.

Taken at face value, reaching 60% more households with 68% of the previous budget implies a cost per household reached roughly 57.5% lower than the year before. That is a PPC Land calculation from Yahoo's two percentages, not a figure Yahoo published.

What the post leaves out matters as much as what it includes. There are no absolute spend levels, no CPM values before or after, no publisher names, no description of how households were counted or deduplicated across devices, and no information on whether the comparison periods had similar market conditions. A year-over-year comparison can also be affected by market-wide price changes unrelated to deal type, and the post does not say whether its figures account for them. Skepticism about vendor-reported results is common in this market: a Jamloop survey of 120 senior brand and agency marketers published in July 2026 found that just 33% fully trust most platform-reported performance claims.

Starcom's account

The post includes one named voice from outside Yahoo, presented under the heading "An advertiser's view." The speaker is Mathew Sanders, VP Director, Platform Investment at Starcom, which is a media agency within Publicis Groupe rather than an advertiser.

"We moved most of our CTV spend from PG to PMP last year, including our Upfront deals," Sanders is quoted as saying. "PMPs have given us more control over which households we reach and how often across our entire campaign. That has made our spend work harder for us in terms of unique reach and incremental reach to linear. Our PMPs have actually paced more smoothly than our PG deals did, since we control daily, weekly, and monthly pacing, and we've still met our spend commitments with publishers. Apart from a few tentpole and live-sports buys, we're reaching the same premium inventory we had before. Our buying approach is really the only thing we changed."

The statement is notable for two details. It says spend commitments to publishers were still met under PMP buying, which addresses a common seller concern about biddable deals. And it says pacing improved, a claim that runs against the usual assumption that guaranteed deals deliver more predictably. The quote does not say which clients' spend moved, what share "most" represents, or which year "last year" refers to, and it contains no numbers.

What Yahoo DSP offers

The post is, in its final third, a description of Yahoo's own buying platform. Yahoo sells a demand-side platform, so the argument for biddable buying is also an argument for the tools it charges for. The post closes by offering help from "Our Yahoo DSP teams."

Household identification. According to Yahoo, its identifier Yahoo ConnectID "now covers approximately 90% of CTV households," citing Yahoo internal data from Q2 2026. The post does not define the market, the method or what "covers" means in practice, whether a match, a reachable household or a modeled one. When Yahoo DSP licensed Nexxen's ACR audience segments in October 2025 for the US, UK and Germany, ConnectID was described as reaching 232 million logged-in users in the United States, a different measure from household coverage. ConnectID, according to the post, helps advertisers recognize the same household "across participating CTV publishers and devices," which limits the claim to publishers that take part.

Buyer controls. The post describes four features aimed at PMP buying:

  • Mixed campaigns, which combine PG and PMP CTV lines in the same campaign so they can be managed and measured together.
  • Flexible Frequency Groups, which allow custom frequency strategies across campaigns, packages and lines. Yahoo's examples include separating CTV from non-CTV, managing PMP and PG independently, and setting different caps by brand or line of business.
  • Peer39 CTV Quality Score, which Yahoo DSP "has integrated." The score is described as "a unified metric built specifically for CTV measurement that's calculated using a weighted blend of key signals, including completion rate, overnight impressions, and verified valid channels." The post does not publish the weights.
  • Jounce Media Supply Insights, which provide supply chain reporting "with detailed publisher and seller transparency," along with supply path optimization segments, inside the platform.

Measurement. Unified TV reporting, according to Yahoo, "provides visibility into total, overlapping, and incremental reach and frequency across CTV supply sources and linear TV investments."

None of these features carries a launch date, a price or a regional availability statement in the post. It is not clear from the text which, if any, are new as of October 22, 2026.

The post fits a sequence of moves by Yahoo DSP into television over the past year. On January 6, 2026, Yahoo DSP put three AI agent capabilities live, including campaign activation over Model Context Protocol. In May, Walmart Connect's first-party data became available in Yahoo DSP for reaching VIZIO CTV audiences. In June, Channel 4 named Yahoo DSP among five new DSP partners for its video-on-demand inventory, with the integration scheduled for June and July 2026. And in July, Truthset's Data Rated Audiences became available for activation against ConnectID across CTV, mobile and the open web.

Where guaranteed deals still fit

Yahoo does not argue for abandoning PG altogether. "PG has a real job: genuinely scarce inventory. Tentpoles, live sports, first position in pod. Guarantee those. Everything else should have to earn the impression," the post states.

The objection, as Yahoo frames it, is to habit rather than to the deal type. "Our objection isn't to PG. It's to PG as a default, because a default is a decision nobody made."

The proposed next step is modest in scale. Yahoo suggests advertisers "carve out 20% of your budget to put PG and PMP to the test." The post does not describe how such a test should be designed, what metrics would decide it, or how long it would need to run.

Competing pitches for deal-based CTV buying

Yahoo is not the only buying platform positioning around deals in CTV this year. Viant extended its Direct Access product to private marketplace and programmatic guaranteed deals on September 24, 2026, with support through FreeWheel, Magnite's SpringServe, Publica and custom OpenRTB integrations. Viant cited publishers that account for 94% of programmatically transacted demand, a figure it attributed to Jounce Media without a date or method.

The Trade Desk took a different route. In July 2026, it opened OpenPath, its direct publisher integration, to CTV pause ads, a format that had been in open beta since May 12, 2026 and limited to private marketplace and guaranteed deals.

Supply verification has also moved closer to deal buying. Jounce Media classified 98% of the Jamloop CTV spend it examined as premium direct, a review published with Jamloop's "Verified Supply" product on October 1, 2026. The same firm's data is what Yahoo offers inside its own platform through Jounce Media Supply Insights.

Why this matters

The debate Yahoo describes is about where control sits in a market that has largely stopped using open auctions for premium television. If more than 89% of programmatic video is bought direct, the practical question for advertisers is not whether to buy programmatically, but whether the deals they sign leave any room to decide impression by impression.

Why would an advertiser accept a deal that obliges it to bid on almost every request? For buyers coming from upfront negotiations, PG offers certainty: fixed price, fixed volume, a known publisher. Yahoo's argument is that this certainty is bought with frequency control, audience definition and pricing flexibility, and that the trade is often unnecessary because the same inventory is available on biddable terms. Sellers can have their own reasons to prefer PG, such as predictable revenue and price protection, and the post does not present the publisher side of that trade.

How much weight the argument can carry depends on evidence the post does not supply. The household coverage figure is undefined. The two customer results are anonymous. The Starcom statement contains no figures. And the company making the case sells the platform that would carry the extra decisioning. Still, the core distinction Yahoo draws, between a deal that lets an algorithm decline an impression and one that obliges the buyer to take almost all of them, matches the standard definitions of the two deal types. Whether the difference produces the gains Yahoo describes is something only a controlled comparison could show, and the 20% test it proposes is, at least, a way to find out.

Timeline

Summary

Who: Yahoo, through a blog post on its corporate site promoting Yahoo DSP, with a supporting quote from Mathew Sanders, VP Director, Platform Investment at Starcom. The post cites eMarketer and Omnicom Media Intelligence research, and describes integrations with Peer39 and Jounce Media.

What: Yahoo argues that programmatic guaranteed deals leave most CTV buying decisions with sellers, and that biddable private marketplace deals let advertisers control household frequency across publishers, suppress homes already reached by linear TV, price impressions individually and choose supply paths. It cites two anonymous customer results, a halved eCPM and 60% more households reached for 32% less investment, and claims ConnectID covers about 90% of CTV households. It proposes a 20% budget test while keeping PG for tentpoles, live sports and first position in pod.

When: The post is dated October 22, 2026. Its supporting data comes from eMarketer (July 2026), Omnicom Media Intelligence (April 2026) and Yahoo internal data (Q2 2026 and August 2026).

Where: The post was published on Yahoo's corporate blog. It does not specify a market for its figures; the eMarketer data it cites covers the United States.

Why: More than 89% of programmatic video is bought through direct deals, according to eMarketer as cited by Yahoo, and Yahoo says the guaranteed share of that spending costs advertisers efficiency through overexposure to heavy viewers and duplicated reach. The argument also supports demand for Yahoo DSP's own buying tools, and its customer evidence is unnamed and partly unsourced.