Industry story
FTC sues Amazon over alleged secret ad auction surcharges
antitrust auction-transparency publisher-economics retail-media walled-gardens
The U.S. Federal Trade Commission and 22 state attorneys general filed suit against Amazon in August, alleging the company secretly applied surcharges in its advertising auctions that caused Sponsored Products advertisers — brands paying to appear prominently in Amazon search results — to overpay. The complaint contends advertisers were not adequately informed about how their winning prices were determined. Amazon disputes the allegations.
The regulatory scrutiny landed directly alongside Amazon's unBoxed product announcements. When Digiday asked whether the new tooling would disclose how individual Sponsored Ads winning prices are set — including any reserve or calculated-price mechanisms — Amazon declined to commit to that level of auction transparency and reiterated its disagreement with the FTC's claims.
Full analysis
The FTC and 22 state attorneys general sued Amazon in August over its ad auction, alleging Amazon quietly added surcharges that made brands running Sponsored Products (the paid slots at the top of Amazon search results) pay more than they thought. The question for every ad-tech operator: does this crack open the walled garden's auction, and does auction transparency finally become table stakes in retail media?
This is easy to undo for Amazon. Settle, tweak the documentation, keep the auction. It is hard to undo for the industry if a consent decree forces disclosure of how winning prices get built, because that becomes a precedent every retail media network gets measured against. What's actually being decided is not whether Amazon pays a fine. It's whether "show me how you built my price" becomes standard buy-side language in retail media. The deadline is soft: unBoxed already happened, discovery runs for a year-plus, and 2027 upfront and RFP cycles are when procurement teams either write transparency clauses in or don't.
The Skeptic. Every auction has a floor. Every platform optimizes its yield. "Secret surcharge" is a scary phrase for what most SSPs do every day when they set a reserve. The FTC has to prove Amazon extracted money in a way that harmed advertisers at scale, and "brands felt overcharged" is not that. Amazon settles, pays something that rounds to nothing against a roughly $47 billion ads business, edits three paragraphs of auction docs, and moves on. Twenty-two states sound fierce but usually land a consent decree with weak teeth. And there's no substitute for bottom-funnel Amazon intent, so nobody moves budget on principle. For the informed outsider: complaining that the casino has a house edge is not the same as proving the dealer cheated.
The Market Analyst. The market has already shrugged. Amazon's ad multiple hasn't moved, and for the fine math, that's correct. But there's a second-order effect the stock isn't pricing: if buyers believe they've been overcharged, even unproven, CPMs on Amazon face psychological resistance going into 2027 planning. The real risk sits in discovery. If documents surface that make programmatic buyers genuinely distrust the calculated-price mechanism, that's a low-odds, high-damage outcome no analyst is modeling. The narrow winners are already visible. Walmart Connect gets a clean "we show you the price" talking point. Any measurement or audit vendor that can verify Amazon spend, VideoAmp, iSpot, DoubleVerify if it builds retail media audit tooling, just got a new reason to be in the room. For the outsider: the fine won't hurt Amazon, but what the lawsuit's paperwork drags into daylight might.
The Operator. Pull your Amazon auction logs now and model what "calculated price" actually cost versus the bid you submitted. Do it before finance asks, because finance will ask within two quarters whether to reclassify Amazon spend as higher-risk. Procurement at the big brands starts writing auction-transparency clauses into RFPs, and by late 2027 that language is standard. Amazon resists it, but resisting is its own signal. The agency trading desks that never built an Amazon audit practice suddenly look asleep at the wheel. Here's what breaks first: you can't actually reconcile the logs, because Amazon doesn't expose the reserve mechanism, and when Digiday asked at unBoxed whether the new tooling would show how winning prices get set, Amazon declined to commit. For the outsider: the buyers can't check their own bill, and Amazon just said it won't hand them the receipt.
The Customer / End User. The advertiser here is a brand paying to sit at the top of Amazon search. Are they actually asking for transparency, or are we projecting? Mostly projecting, until now. Brands tolerated the black box because Amazon search converts. But tolerance and trust aren't the same thing, and a federal complaint gives every CMO cover to demand the receipts they always wanted and never asked for. The catch: even with full disclosure, most brands can't leave. There's no comparable inventory for someone searching a product with a credit card out. So the ask becomes loud, the leverage stays weak, and the outcome is more reporting. Spend holds. For the outsider: the customers will complain harder and pay anyway.
The Strategist. This reshapes the retail media moat over 24 months. Amazon's edge in sponsored search rested on an auction nobody could interrogate. Force disclosure of reserve or calculated-price mechanics, and Walmart Connect plus the Criteo-powered retailer networks get a structural opening, not because they run a better auction, but because "here's how your price is built" becomes a feature. The deeper shift is that open-web norms, where DSP and SSP contracts already spell out fees and floors, finally push into walled-garden retail media. The moat doesn't vanish. It narrows. For the outsider: the thing that made Amazon ads hard to argue with was that you couldn't see inside; regulators are trying to put a window in.
The tensions
Does discovery matter, or is it theater? The Skeptic says settle-and-forget, fine is rounding error, no substitute exists so nobody leaves. The Market Analyst agrees on the fine but flags the one thing that isn't priced: what internal documents reveal about the auction. Same event, and the disagreement is entirely about whether the lawsuit produces paper that changes buyer behavior. That's the real fork.
Loud ask, weak leverage. The Operator says transparency clauses become standard RFP language. The Customer says brands can't actually walk, so the demand has no teeth. Both are right, and that's the uncomfortable part: procurement writes the clause, Amazon strikes it, and everyone signs anyway because the inventory converts. A demand nobody can enforce is a press release.
Does the moat actually narrow? The Strategist sees open-web transparency norms migrating into retail media. The Skeptic sees a company that edits three paragraphs and keeps the black box. The difference is whether a consent decree forces mechanism disclosure or just clearer boilerplate.
What this hinges on
Three beliefs. One, whether the FTC can prove harm tied to a specific mechanism, not just that advertisers felt squeezed. Two, whether discovery surfaces documents that shift how buyers see the auction. Three, whether any consent decree forces disclosure of reserve and calculated-price mechanics, or just tidier language. The council leans toward the Skeptic on the money and the Market Analyst on the risk: the fine is noise, and the only thing that moves the industry is what gets dragged into the record. What to de-risk before spending another quarter on faith: run your own auction-log reconciliation now, so you know your exposure whether or not Amazon ever tells you.
The most durable move here isn't Amazon's. It's that when Amazon declined at unBoxed to commit to showing how winning prices are set, it told you what it will fight to protect. The auction opacity is the product. Amazon will spend legal dollars to keep it before it spends a dime settling on it.
Prediction: Any FTC settlement or consent decree Amazon reaches over its Sponsored Products auction will not require Amazon to disclose per-auction reserve prices or the calculated-price mechanism to advertisers, and Amazon's Sponsored Products auction will remain a black box through the resolution of the case.
Confidence: Medium. Amazon already refused to commit to disclosure at unBoxed, and opacity is the asset it defends.
Why: When Digiday asked at unBoxed whether Amazon's new tooling would show how individual winning prices are set, Amazon declined to commit and reiterated its disagreement with the FTC. That refusal, made voluntarily during a friendly product launch, signals Amazon values the auction's opacity more than the goodwill of transparency, because the opacity is what lets it optimize yield without buyers pushing back. Consent decrees historically extract behavioral promises and fines. They rarely force a company to surrender the core competitive mechanism that makes its auction more profitable than the transparent ones next door, and Amazon's legal team knows disclosure of reserve mechanics would hand Walmart Connect and Criteo-powered networks a differentiator worth more than any fine. The opposite outcome, a decree forcing mechanism-level disclosure, is less likely because the FTC would have to win or extract far more than settlements typically deliver, and Amazon would fight that harder than the money.
Revisit by 2027-09-30: We're right if any settlement, consent decree, or Amazon product change through that date leaves the Sponsored Products auction without published per-auction reserve or calculated-price disclosure to advertisers. We're wrong if Amazon, by settlement or on its own, publishes the reserve or calculated-price mechanism at the auction level to Sponsored Products advertisers.
This is the gap between what regulators want and what the incentive protects. The FTC can force a fine. It is much harder to force a company to dismantle the thing that makes its auction more profitable than the transparent ones next door.
Also covered this issue
-
CMO Median Tenure Down 35% Since 2010, Now 2.6 Years
digiday
CMO tenures now average 2.6 years, forcing ad-tech vendors to rebuild sales motions faster than contracts renew.
-
Apple iOS 27 Blocks The Trade Desk From Serving Ads on Safari
adexchanger
Apple's iOS 27 blocklist removes The Trade Desk's ad-serving domain from Safari while Google's remains untouched, forcing budget migration toward DV360 regardless of whether the block gets reversed.
Comments