Industry story
Trade Desk and Viant Quietly Building Walled Gardens of Their Own
dsp identity measurement walled-gardens
An AdExchanger opinion piece argues that the ad-tech industry's self-styled 'independent' platforms are steadily accumulating control over the same decisioning layers — identity, measurement, optimization, supply access — that they once criticized walled gardens like Meta and Google for monopolizing. The article cites The Trade Desk's build-out of OpenPath (supply access), Unified ID 2.0 (identity), Kokai (optimization), and Ventura (a connected-TV operating system), alongside Viant's acquisitions of IRIS.TV (content classification), Lockr (identity), and TVision (measurement). The authors contend that the meaningful question is no longer whether a platform owns media, but whether advertisers can independently audit its optimization logic — and that without such auditability, 'neutrality becomes difficult to distinguish from opacity.'
Analysis
Showing the shorter version.
Trade Desk and Viant Are Building Stacks That Grade Their Own Homework
The independence pitch that built the open web is quietly turning against buyers. An AdExchanger opinion piece lays out how The Trade Desk (the largest independent ad-buying platform) and Viant have assembled nearly the same stack Google owns: identity, supply access, measurement, and optimization under one roof. The Trade Desk has OpenPath for supply, Unified ID 2.0 for identity, Kokai for optimization, and Ventura as a connected-TV operating system. Viant bought its way to the same place with IRIS.TV for content classification, Lockr for identity, and TVision for measurement.
The piece frames this as walled-garden behavior. That's a stretch. Neither company owns a logged-in audience the way Meta or Google does. UID2 is open-source. OpenPath still clears on open-auction rails. True lock-in requires owning the user; these companies don't. The structural-equivalence argument is overcooked.
The practical problem, though, is real. When the platform that decides where your money goes also measures whether it worked, every reporting discrepancy gets resolved in their favor. Not because of malice, just because the numbers that matter flow through their systems. That doesn't show up in a pilot. It shows up at 90 days, when TVision says the campaign worked and your own log-level data says something quieter.
The fix is contractual and boring: audit-rights language and third-party measurement rights before the next upfront locks you in. Whatever you accept at renewal holds for a year or more. If those rights aren't in your current Trade Desk or Viant contracts, they probably disappeared quietly and you'll need a renegotiation to get them back.
The market has already priced The Trade Desk's platform ambitions at a premium. Viant is still priced like a niche CTV shop, not the measurement-plus-identity bundle it has become. The squeezed party in either case is the standalone measurement or identity vendor with one feature and no bundle to hide inside.
Our call: By the end of the 2026 upfront cycle, independent measurement and audit-rights language becomes a standard buy-side ask in Trade Desk and Viant renewals, and at least one major holding company or large advertiser publicly pushes for third-party auditability of platform optimization. Confidence: medium. Buy-side procurement has a decade-long track record of formalizing exactly these asks into contract templates once a trend gets a name, and the incentive to demand audits rises exactly as the stack closes.
The independence pitch that built the open web is quietly eating itself. An AdExchanger opinion piece lays out how the platforms that sold themselves as the anti-Google option are now assembling the same stack Google owns: identity, supply access, measurement, and optimization, all under one roof. The Trade Desk has OpenPath for supply, Unified ID 2.0 for identity, Kokai for optimization, and Ventura as a connected-TV operating system. Viant bought its way to the same place with IRIS.TV for content classification, Lockr for identity, and TVision for measurement.
What's actually being decided here: not whether these companies own audiences (they don't), but whether an advertiser can still independently check the math when the vendor that decides where your money goes also measures whether it worked. This is a Type 1 problem for buyers. Once audit rights are gone from the contract, you don't get them back without a renegotiation you'll lose. The forcing function is the upfront and renewal cycle: whatever language you accept now locks in for a year or more.
The Market Analyst. The stock market already believes the platform story for The Trade Desk. It trades at a premium to plain DSP peers (a DSP, demand-side platform, is the tool buyers use to purchase ads) precisely because investors bought the operating-system thesis. What's not baked in is the buy-side or regulatory pushback that forces open the black box. That's the downside nobody's modeling. Viant is the mirror image: still priced like a niche CTV shop, not the measurement-plus-identity bundle it's quietly become. In plain terms: one company's ambition is fully paid for, the other's isn't. The squeezed party is the mid-tier point solution, the standalone measurement or identity vendor with one feature and no bundle to hide inside.
The Skeptic. The whole piece rests on one move: owning adjacent layers equals walled-garden control. That conflates two different things. Vertical integration is not lock-in. UID2 is open-source. OpenPath still clears on open-auction rails. IRIS.TV licenses its data broadly. Meta and Google are walled gardens because they own the logged-in user and can simply refuse to share. The Trade Desk and Viant own no such audience. And the auditability complaint, while fair, applies to every opaque algorithm in ad tech, including the bid-shading logic buyers happily ran for a decade without a peep. In plain terms: this is a useful think piece stretching a real trend into an alarming headline.
The Operator. Forget the framing fight. Tuesday morning, the problem is reconciliation. When optimization, measurement, and classification all live in one vendor, and the numbers don't match, guess whose favor the discrepancy gets resolved in. Not yours. That break doesn't show up in the pilot. It shows up at 90 days, when your TVision measurement says the campaign worked and your own log-level data says something quieter. The fix is boring and it's contractual: audit-rights language and SLAs before the next upfront locks you in. In plain terms: if the referee also plays for one team, you want the tape.
The Customer / End User. Nobody on the buy side asked for this. Advertisers wanted an alternative to the duopoly, and the pitch worked because it promised transparency Google and Meta wouldn't give. Now the alternative is assembling the same opacity with a friendlier logo. The advertiser's real question shifted. It used to be "who owns the media?" Now it's "can I see how you decided to spend my money?" In plain terms: the buyer traded one black box they distrusted for another they haven't learned to distrust yet.
The tensions. Two real disagreements. First, the Skeptic versus the Strategist read baked into this cluster: is this structural power or just a tidier stack? The Skeptic is right that no endemic audience means no true wall. But the Operator is right that you don't need to own the audience to resolve every reporting dispute in your favor. Control over the scoreboard is a softer moat than owning the stadium, and it's still a moat. Second, the Market Analyst versus the Customer: the market rewards this integration as margin and pricing power, while the buyer experiences it as declining ability to verify. Both are describing the same feature. One calls it a platform, the other calls it a black box.
Synthesis. This hinges on one belief: does bundling the decisioning layers reduce a buyer's ability to independently audit outcomes? If yes, the "independent walled garden" frame has teeth and the auditability startups and clean rooms get a genuine tailwind. If no, it's just vertical integration and the alarm is oversold. The honest answer is that it depends entirely on contract language nobody's reading closely yet. The council leans toward the Operator: the structural-equivalence argument is overcooked, but the practical loss of audit rights is real and it's happening at renewal, in the fine print, while everyone argues about metaphors. Verify one thing before the upfront: do your current Trade Desk and Viant contracts give you log-level access and third-party measurement rights, or did those quietly disappear? De-risk by keeping at least one measurement vendor outside the stack you're buying from.
Prediction: By the end of the 2026 upfront/newfront negotiating cycle (Q4 2026), independent measurement and audit-rights language will become a standard buy-side ask in Trade Desk and Viant renewals, and at least one major holdco or large advertiser will publicly push for third-party auditability of platform optimization.
Confidence: Medium. The incentive to demand audits rises exactly as the stack closes.
Why: The AdExchanger piece names the specific accumulation, OpenPath, UID2, Kokai, Ventura at The Trade Desk, and IRIS.TV, Lockr, TVision at Viant, that puts optimization and measurement under one roof, and buy-side procurement teams already treat measurement reconciliation as a fight they lose when the vendor grades its own homework. When the referee and the player merge, the predictable buyer response is to demand independent scoring, and holdco procurement has a decade-long track record of formalizing exactly these asks into contract templates once a trend gets a name. The opposite outcome, buyers accepting single-vendor measurement without a fuss, runs against how procurement behaves whenever switching costs and opacity rise together.
Revisit by 2026-12-31: We're right if a major agency or advertiser publicly calls for independent auditability of DSP optimization, or audit-rights language becomes a reported standard in these renewals. We're wrong if the upfront closes with no visible buy-side push and measurement stays bundled without complaint.
Also covered this issue
-
LUMA Q2 2026 Report: Ad-Tech M&A Up 5%, CTV and AI Drive Deals
adexchanger
Fox buying Roku for twenty-two billion resets the valuation template for every independent CTV platform and forces immediate data-sharing renegotiations.
Comments