Stagwell Volunteered for Trade Desk's Koa Beta. Three Holdcos Already Quit.

Stagwell announced on April 21, 2026 that it will become the first global marketing network to adopt Koa Agents, The Trade Desk's new agentic AI for media planning and buying. The announcement landed roughly two months after WPP and Dentsu quietly exited Trade Desk's OpenPath product, and one month after Publicis told its clients to stop transacting on the DSP following an internal audit. Stagwell is now the largest holding-company-shaped buyer publicly defending Trade Desk in the middle of an open agency revolt.

What Stagwell actually signed up for

The deal embeds Stagwell's existing Koa software into Trade Desk's Kokai interface via the Open Agentic Kit framework. Phase one automates audience planning, segment building, and activation across the open internet. Phase two pushes into automated campaign setup, real-time troubleshooting, and proactive optimization recommendations. Stagwell's official announcement says closed beta lands later this summer.

Matt Adams, global CEO of Stagwell Media Platform, told Digiday the partnership offered "less customization" than Stagwell's parallel work with Google, Gemini, and DV360, which is why Trade Desk made sense as the place to lean in. "We're surfacing the insights and the options in a faster and better way," Adams told Digiday. "You'll still need the traders' judgement."

Slavi Samardzija, Stagwell's global chair of media and commerce, framed the integration as combining Stagwell's data infrastructure with Trade Desk's agentic capabilities. Trade Desk CEO Jeff Green is reportedly overseeing the project personally. None of this would be unusual on its own. The unusual part is the timing.

What three holdcos already walked away from

In February 2026, WPP and Dentsu both exited OpenPath, Trade Desk's direct-to-publisher buying product. Adweek's reporting cited hidden fees, undisclosed charges, and transparency concerns as the proximate cause. Trade Desk's stock dropped over 3% on the news. The company pushed back, pointing to a 95% customer retention rate, $443M in net profit, and 18% annual revenue growth as evidence the OpenPath grievances were narrow and not systemic.

Then in March, Publicis Groupe recommended its clients stop transacting on Trade Desk's DSP following an internal audit. Per the audit, Trade Desk had "improperly applied their DSP fee to other fees" and automatically opted the holdco and some of its clients into fee-based offerings without explicit authorization. We covered the underlying mechanics last month in our piece on who actually owns the supply pipe in the Publicis split.

So inside about ten weeks, three of the largest agency networks moved against the platform. Their stated reasons converge on one theme: the bill did not match the contract. That is also the part Stagwell does not appear to be challenging at all.

Three reads on why Stagwell went the other way

None of these readings are flattering to all parties, and Stagwell has not said which one applies. Honestly, it is probably some mix of all three.

The first read is negotiation. When everyone pulls budget at once, the platform that loses the budget gets cooperative. Better rates, better roadmap influence, better make-good terms when something breaks. Stagwell becoming the loudest pro-Trade-Desk voice in the room buys exactly that kind of access while the cost of buying it is at a multi-year low. Adams's offhand comment about "less customization" elsewhere reads, to me, like the renegotiation thesis stated out loud.

The second read is access. Amazon DSP is the obvious alternative, especially for retail-media-heavy clients, but Amazon DSP requires Amazon Ads relationships that do not equally favor every holdco. The inventory it provides access to (Prime Video, Twitch, Amazon retail media) is asymmetrically valuable depending on your client mix. WPP, Publicis, and Dentsu have deep CPG and retail relationships that justify the Amazon switch. Stagwell's client mix skews differently. If you cannot get the same Amazon terms as WPP, doubling down on Trade Desk is rational, not contrarian.

The third read is product economics. Stagwell already runs Koa as its own AI layer. Embedding it into a partner's stack costs less than rebuilding the AI surface for another DSP from scratch. From that angle, the Koa Agents deal is less a vote of confidence in Trade Desk's auditing and more a vote of confidence in Trade Desk's API surface area.

Brands probably should not read this announcement as Stagwell endorsing Trade Desk's transparency posture. It seems to depend more on Stagwell choosing the path where its existing product and a partner's renegotiation moment line up.

The question to put on your next agency QBR

If your agency runs more than 30% of programmatic spend through a single DSP, ask three specific questions before the next renewal cycle:

1. Which DSP is taking the spend, and what percentage of total programmatic budget does that represent? If the answer is "we do not break that out at the client level," that is the answer.
2. Does the agency have principal media positions on the same supply path? OpenPath was the OpenPath problem. The fee structure mattered because the agency was both buying and brokering on the same lane.
3. If the same audit Publicis ran landed on your account, what would change in reporting? Most agencies have not stress-tested this and will not have a fast answer.

I think most teams will skip this conversation because it is awkward and the agency will frame the audit as paranoid. From what I have seen, the agencies that handle this gracefully are usually fine. The ones that get defensive when asked basic supply-path questions tend to be the ones with something to defend.

There is a smaller, related question worth asking: whose AI layer is actually running your bid logic? Stagwell's Koa is Stagwell's. Reported holdco buying tools route through partner SSPs and DSPs in ways that are not obvious from the dashboard. Omnicom is partnered with Equativ. Whose model decides the bid matters because whose model decides the bid eats the margin first. We dug into a parallel version of this in Magnite's buyer agent bidding on its own Disney inventory.

What isn't actually happening yet

Despite the audit drama, advertisers have not actually moved budget. Digiday reported that no media buyer interviewed had switched DSPs after the Publicis announcement. The DSPs that pitched (Amazon, Yahoo, Viant, Nexxen, MNTN, StackAdapt) generated meetings, not migrations. Viant's 2025 revenue grew 19% to $344.2M, which is a fine year, but it is not the kind of share shift you would see if the holdco walkouts had real budget behind them.

And to be fair, this part is not entirely new. Programmatic agency standoffs usually look like noisy positioning followed by quiet renegotiation. The walkouts so far are about audit pressure, not about clients actually leaving. Which is what makes Stagwell's announcement strategically interesting and not strategically reckless. There is no real exodus to outflank. There is a fee renegotiation cycle to ride.

What I would watch for the rest of Q2

Two things over the next 90 days.

The first is whether Trade Desk publishes a substantive response to the Publicis audit, with line-item fee disclosures, rather than the directional defenses it has offered so far. If it does, the holdco walkouts get quieter, the rival DSPs stop getting pitch meetings, and Stagwell looks early instead of contrarian. If it does not, Stagwell looks more isolated and the renegotiation thesis gets stronger.

The second is whether anyone follows Stagwell publicly. Havas, IPG, and the smaller global networks have stayed quiet so far. If even one of them publicly co-signs Koa Agents, the narrative flips from "Stagwell against the holdcos" to "the consensus moved on without WPP and Publicis." My guess: at least two of the four remaining holdcos will quietly try to negotiate softer Trade Desk terms within 60 days, regardless of what they say in public.

For now, I think the safer bet for brands is the one that gets buried in every transparency story: pick your DSP based on the audit you would be willing to run on it, not the one your agency wants to defend.

Notice Me Senpai Editorial