Google Must Let Publishers Take AdX Demand Outside Google’s Stack

A federal judge is forcing Google to change how two products at the center of online advertising work together after finding that the company unlawfully monopolized the markets for open-web publisher ad servers and ad exchanges.

DFP, now sold through Google Ad Manager, is the ad server publishers use to manage advertising inventory and decide which ad ultimately fills an available position. AdX is Google’s advertising exchange, where buyers submit bids for that inventory. Google operates both products and also controls AdWords, the advertiser-facing system responsible for a large source of demand entering AdX.

The court previously found that Google unlawfully tied DFP to AdX, making the products harder for publishers to replace independently. The remedy rejects the Justice Department’s request to force a sale of AdX and instead requires Google to open the connections among its products. AdX must submit real-time bids through competing publisher ad servers, DFP and AdX must integrate with Prebid, publishers must be able to export their data, and AdWords can’t favor Google-owned ad technology based on ownership.

Google keeps the products, customers and advertiser demand that made its ad business powerful. Publishers gain more ability to keep one Google product while replacing another, which puts more pressure on each part of the stack to compete on its own economics and performance.

AdX Demand Becomes Less Tied to Google’s Ad Server

Google’s advantage came partly from controlling multiple steps in the same advertising transaction.

Publishers use an ad server to determine which sources of demand compete for an impression and which advertisement wins. Exchanges connect those publishers with advertisers and buying systems willing to bid for the impression. Google could operate the publisher’s ad server through DFP, bring demand into the auction through AdX and supply a large pool of advertiser spending through AdWords.

The court found that the relationship between those products restricted competition. Publishers seeking the most direct real-time access to AdX demand had a strong incentive to remain on DFP. Rival publisher ad servers therefore had to persuade a customer to switch while that customer risked losing an important source of advertising demand.

The remedy requires AdX to submit real-time bids to competing publisher ad servers in the same manner it submits them to DFP. Google must also provide the technical support required to make those connections work.

That changes the sales proposition for an independent ad server. A competitor can ask publishers to evaluate its software, fees, workflow and performance without requiring them to surrender AdX demand as part of the migration.

DFP will still benefit from familiarity, existing integrations and Google’s scale. Rival servers still have to prove they can protect yield and handle the operational demands of publishers selling inventory continuously. Access to AdX removes one of the penalties for testing them.

Prebid Has to Get an Integration That Can Actually Compete

Prebid gives publishers an open-source way to solicit bids from multiple exchanges before the ad server selects the winning advertisement.

Google must create integrations connecting AdX with Prebid and DFP with Prebid. The AdX integration will allow Prebid to solicit real-time Google bids for indirect open-web display inventory. The DFP integration will allow publishers to route competing demand through Prebid while continuing to use Google’s ad server.

The technical requirements make the order more consequential than a basic interoperability mandate.

The integrations must operate in a functionally equivalent manner and on materially identical terms to the way DFP works with AdX. Google can’t satisfy the order by creating a connection that adds latency, responds to fewer requests or provides weaker information simply because a publisher uses Prebid or another ad server.

Latency, information and response rates directly affect auction economics. A bidder that receives less information may value an impression differently. A bid arriving too late can’t win. An exchange receiving fewer opportunities has less chance to generate publisher revenue.

The quality of Google’s required integrations therefore becomes measurable. Publishers and the court-appointed technical monitor can compare bid response rates, latency, information flows and auction results across Google and competing routes instead of treating interoperability as a binary question.

Prebid becomes an important piece of infrastructure in that test. Publishers can keep DFP while using a more neutral route for competing demand, or keep AdX demand while testing another publisher ad server.

Publisher Data Becomes Easier to Move With the Account

Ad-server competition also depends on whether publishers can take their operating history with them.

Publishers build configurations, pricing rules, inventory records and auction data inside the systems they use. Switching becomes riskier when a new provider requires rebuilding that information or operating with less visibility into historical performance.

The remedy requires Google to let publishers access and export data from DFP and AdX. Google must also provide more auction information, including data covering winning and losing bids, and documentation explaining how DFP reaches auction decisions.

The court rejected the Justice Department’s proposal to open-source DFP’s final auction logic. Instead, Google has to disclose enough technical and transaction information for publishers and competitors to examine how ads were evaluated and selected without handing over the complete underlying system.

Data portability shifts another portion of the switching cost away from access and toward execution. A rival still has to migrate configurations correctly, maintain reporting continuity and prove that the new setup can monetize inventory as effectively as the incumbent.

Those are competitive problems a software company can attempt to solve. Losing auction history or access to valuable demand because the publisher changed providers creates a much larger structural barrier.

Publisher Pricing Can Put Pressure on AdX’s Economics

The order also blocks Google from restoring several auction practices challenged during the case, including First Look, Last Look and Unified Pricing Rules.

Unified Pricing Rules limited publishers’ ability to set different floor prices for different buyers and exchanges. Removing those rules gives publishers more control over the minimum prices at which separate sources of demand can compete. Google had already removed Unified Pricing Rules from Ad Manager before the final remedy.

The court’s opinion cites an estimate from plaintiffs’ liability expert Tim Simcoe that eliminating Unified Pricing Rules could reduce AdX’s effective take rate from 20% to roughly 16.6%, putting it closer to some competing exchanges. The actual result will depend on publisher pricing decisions and auction behavior, but the estimate identifies a direct economic pressure created by allowing publishers to treat demand sources differently.

That pressure becomes more meaningful alongside interoperability. A publisher that can compare exchanges through Prebid, route AdX through another server and set different pricing rules has more ways to determine whether Google’s exchange is producing enough incremental revenue to justify its economics.

Google’s scale will continue to generate demand. The remedy gives publishers more tools for putting a price on access to it.

AdWords Loses Some of Its Ability to Reinforce Google’s Infrastructure

Advertiser demand helped make the DFP-AdX relationship difficult for publishers to avoid.

The court described AdWords demand as central to the competitive problem. AdWords had integrated with third-party exchanges, but most of its open-web display activity continued to flow through AdX. Restricting access to that demand helped reinforce Google’s exchange and, through the tie with AdX, its publisher ad server.

The remedy prohibits AdWords from deciding whether, how or how much to bid based on whether a publisher uses Google-owned ad technology. AdWords also can’t prioritize AdX simply because Google owns it and can’t create a direct bidding integration with DFP for the indirect open-web display inventory covered by the order.

Google can still make bidding decisions based on legitimate performance considerations such as information quality, fraud protection, privacy and expected advertiser return. The final judgment still has to define some of those boundaries more precisely.

DV360, Google’s demand-side buying system, isn’t subject to the same restrictions. The court concluded that DV360 didn’t play a sufficient role in the liability findings to justify extending those remedies to it.

Google therefore retains substantial buyer-side scale even as the court restricts how AdWords demand can reinforce ownership of the rest of its stack.

Independent Ad Tech Still Has to Convert Access Into Revenue

The ruling gives rival ad servers and exchanges better access to publisher transactions. It doesn’t supply them with customers.

Publishers will still evaluate whether another ad server improves revenue enough to justify migration risk. Exchanges have to bring incremental demand. Prebid integrations have to perform under production traffic. Reporting has to remain reliable. Support has to work when a problem affects revenue in real time.

Google begins that competition with established publisher relationships, familiar workflows and enormous transaction volume. The court’s opinion says DFP held 91% of the publisher ad server market in 2022, while AdX handled 63% to 71% of global open-web display exchange transactions among exchanges that supplied data between 2018 and 2022.

Independent advertising technology already faces a broader version of that challenge. The Trade Desk’s struggle to restart growth despite a strong balance sheet shows how companies without owned media, commerce or search businesses have to assemble comparable reach through integrations and partnerships. Google, Amazon and other vertically integrated companies can connect advertiser relationships, first-party data, inventory and transaction infrastructure inside businesses they control.

The court has reduced one specific advantage created by Google’s control of adjacent products. Competitors still have to build businesses capable of taking advantage of the opening.

6 Years of Technical Oversight Creates a Measurable Compliance Test

Google will operate under a court-appointed monitor and three-member technical committee for six years. The court can extend the term if the judgment hasn’t been fully satisfied.

The monitor will have authority to interview Google employees, inspect documents and access systems, algorithms and source code. The oversight applies globally rather than only to Google’s U.S. advertising operations.

That access is important because compliance will happen inside software.

Google Ad Manager processes millions of ad requests every second. Small changes to latency, bid requests, data availability or auction treatment can alter which demand sources compete and how publishers perceive the performance of an alternative provider. A nominally available integration can have little commercial value if it performs worse at scale.

The monitor therefore has a practical scorecard: how AdX bids perform through competing ad servers, whether Prebid receives equivalent information and response treatment, how often publishers route auctions through non-Google infrastructure, whether publishers migrate configurations and data successfully, and whether those changes affect fees and publisher yield.

The behavioral remedies are expected to be implemented within 15 months. The final judgment will take effect 60 days after entry, while the monitor and technical committee process begins immediately. The order applies globally.

Google says it plans to appeal the underlying liability ruling involving Google Ad Manager. The Justice Department and Google also have to submit proposed final judgment language reflecting the remedy.

The Streaming Wars Take

Rival ad servers have wanted access to AdX demand because publishers were reluctant to adopt infrastructure that could weaken access to Google’s buyers. Once that access becomes available on equivalent terms, those competitors have to show that publishers can change servers without sacrificing yield, speed or operating reliability.

Publishers gain another source of bargaining power. A company that can route demand across multiple systems, move its data and compare auction results has more credible alternatives when negotiating fees, deciding where to route inventory or evaluating whether one provider deserves to control several pieces of the transaction.

The liability markets in the case concern open-web display advertising, rather than the full streaming and CTV advertising market. The competitive mechanism extends well beyond one format: ownership of demand, infrastructure and inventory becomes more valuable when customers have to take those pieces together.

That pressure is already shaping streaming advertising. Independent ad tech is being valued against companies that own more of the audience, data, inventory and transaction, while media owners increasingly connect their inventory with several buying systems to increase competition for impressions.

The Google remedy creates a live test of whether mandatory interoperability can reduce the advantage of vertical integration without separating the assets themselves.

If publishers begin moving meaningful volume through competing servers and Prebid while preserving access to AdX demand, DFP will have to compete more directly on the quality and economics of ad serving. Rival exchanges will get more opportunities to prove incremental value inside auctions previously organized around Google’s workflow. Publishers will gain evidence they can use when deciding how much of their monetization stack should remain with one company.

If routing patterns barely move, Google will retain most of the commercial benefit of its installed base even after the technical tie has been weakened.

Six years of monitoring gives the market enough time to answer the question with transaction data. The relevant measure isn’t whether Google built the required connections. It’s whether publishers use them to move inventory, demand and revenue through companies Google doesn’t control.