The Federal Trade Commission and 22 state attorneys general sued Amazon.com on August 31, 2026, alleging the company spent seven years quietly inflating the prices more than one million brands and sellers paid to advertise on its store. The complaint centers on a mechanism Amazon called a “soft reserve price” internally, added to its advertising auctions in 2019 without notice to the advertisers bidding in them.
According to the FTC, the effect was to convert an auction Amazon publicly described as a second-price auction into something that functioned as a first-price auction. The agency says the scheme has “likely extracted tens of billions of dollars” from advertising customers who did not know the rules had changed.
In short
- Who filed: the FTC plus attorneys general from 22 states, in the U.S. District Court for the Western District of Washington, on August 31, 2026. The Commission vote authorizing the complaint was 2-0.
- The mechanism: a hidden surcharge Amazon internally called a “soft reserve price,” introduced in 2019, which the complaint says pushed advertisers to pay close to their own bid rather than one cent above the next-highest bidder.
- The scale: the FTC alleges the conduct touched over one million brands and sellers, including more than 500,000 small and medium-sized businesses, and generated tens of billions of dollars in revenue. Reporting on the complaint puts the figure above $20 billion since 2019.
- The timing pattern: the complaint alleges surcharges were ramped hardest around Prime Day and Black Friday, with gradual increases beforehand to disguise the size of the move.
- Amazon’s position: the company calls the suit “misguided” and says it “fundamentally misunderstands how advertisers operate.” No trial date has been set.
What exactly does the FTC allege Amazon did?
The complaint describes a gap between how Amazon marketed its advertising auctions and how those auctions actually priced. Amazon sells placements for Sponsored Products ads, Sponsored Brands ads and display ads that appear alongside keyword search results on Amazon.com and its mobile app. Placements go to the highest-ranked bidder for each keyword.
For years, according to the FTC, Amazon told prospective advertisers it ran second-price auctions in which the winner would pay “one cent more than the next highest bidder” for each successful keyword bid. That representation appeared on Amazon’s website, in training videos, in other public-facing materials, and in presentations delivered by what the complaint describes as hundreds of sales personnel.
The agency alleges the reality diverged sharply. Beginning in 2019, the complaint says, Amazon changed its auction rules without notice by adding an undisclosed surcharge, with the result that advertisers paid substantially more than the price the second-price auction would have produced.
The structure of the claim will be familiar to anyone tracking the agency’s recent priorities. It targets an opaque pricing layer sitting between a stated rule and a charged amount, the same concern behind the FTC’s scrutiny of personalized pricing practices in consumer retail.
Why the auction format matters to a bidder
The distinction is not academic, and the FTC spends part of the complaint explaining why. In a sealed first-price auction, winners pay the full amount they bid, so bidders run a real risk of overbidding. In repeated auctions for the same inventory, they respond by shading their bids downward to find the minimum needed to keep winning.
Second-price auctions invert that incentive. Because the winner pays only the least amount required to win under the rules, bidders can safely bid closer to what a click is genuinely worth to them. Generalized second-price auctions, or GSP, have been an accepted standard in digital advertising for exactly this reason.
The FTC’s theory is that Amazon captured the bidding behavior of a second-price auction while charging the prices of a first-price one. Advertisers kept bidding high because they believed they were protected by GSP mechanics. The complaint quotes Amazon employees acknowledging that advertisers “are operating under the assumption that Amazon uses a GSP auction” and that “many advertisers bid far higher than what they are willing to pay … because they assume [the auction is] a GSP.”
What the “soft reserve price” actually did
A reserve price is an ordinary and generally disclosed feature of auctions: a floor below which the seller will not transact. The complaint’s objection is not to the existence of a floor but to its concealment and its calibration. Amazon’s internal term, according to the FTC, was “soft reserve price,” and one internal document described the auction pricing as having “a surcharge hidden in it.”
The complaint quotes the executive in charge of Amazon Ads explaining internally that the price advertisers pay “isn’t set by an actual bidder” but is instead a “proxy 2nd price that we calculate.” Another document cited in the filing has Amazon acknowledging that it uses an “invented auction participant” to increase prices. The FTC characterizes both the invented participant and the proxy second-price bids as functionally equivalent to shill bids.
How much money is actually at stake?
The FTC’s own language is deliberately broad: the scheme has “likely extracted tens of billions of dollars from its unwitting advertising customers.” Reporting on the complaint has been more specific, putting the figure at more than $20 billion drawn from roughly 1.2 million U.S. advertising customers since the 2019 rule change.
Context helps here. Amazon’s advertising business generated roughly $68 billion in revenue in 2025 by external estimates, approaching $70 billion, which makes it the third-largest digital advertising business globally behind Google and Meta. A surcharge alleged to have run continuously since 2019 across the majority of Sponsored Products auctions is not a rounding error against that base.
The complaint also traces an escalation. It alleges the share of the time Sponsored Products advertisers paid the full amount of their own bid climbed steadily as the surcharges tightened.
The escalation the complaint describes
| Period | Share of Sponsored Products auctions where the advertiser paid its own bid | What the complaint attributes it to |
|---|---|---|
| 2021 | Between 30% and 40% | Early operation of the 2019 surcharge |
| 2022 | Approximately 70% | Tightening of surcharge settings |
| 2024 | Approximately 80% | Continued escalation to meet revenue targets |
At roughly 80%, the practical difference between Amazon’s described auction and a straightforward first-price auction narrows to very little. That figure is the numeric heart of the FTC’s deception count: an auction described as second-price that charges the winner’s own bid four times in five is, in the agency’s framing, not the product advertisers were sold.
Stated model against alleged practice
| Element | What Amazon represented, per the complaint | What the FTC alleges happened |
|---|---|---|
| Auction type | Generalized second-price (GSP) | Functionally converted to first-price |
| Price paid by winner | One cent above the next-highest bid | The winner’s own bid, roughly 80% of the time by 2024 |
| Reserve mechanics | Not disclosed as a pricing lever | Undisclosed “soft reserve price” added in 2019 |
| Competing bid | Set by other advertisers | Partly an “invented auction participant” and a calculated “proxy 2nd price” |
| Peak shopping events | Same rules as ordinary days | “Far greater increases” during Prime Day and Black Friday |
| Advertiser inquiries | Answered accurately | Answered with “false and misleading answers,” per the complaint |
What do the internal documents say?
The FTC’s complaint leans heavily on Amazon’s own records, which is typical of a Section 5 deception case built around concealment. The quoted material is unusually direct.
The agency says Amazon made the change because it was unhappy with how much revenue its advertising auctions were generating. One employee is quoted saying the surcharges obtain prices “beyond what [can] be achieved through advertiser competition.” Another said the hidden surcharges are “good for Amazon” because “advertisers must pay more for the same advertising,” adding that “the benefit to Amazon comes at the cost of advertisers.”
The most quotable passage in the filing comes from notes on a 2024 discussion between senior executives, including the head of Amazon Ads and Amazon’s Chief Digital Economist. Those notes describe the arrangement as a “clever non-transparent way to charge first price” and call it an “incredibly effective way to drive revenue.”
The concealment allegations
The FTC alleges Amazon understood what disclosure would cost it. Internal documents quoted in the complaint warn that revealing the surcharges would cause “irrevocable damage to advertiser trust” and set off a “downward spiral” of advertisers lowering bids, with dramatic revenue losses following.
That fear, the agency argues, produced active concealment rather than passive omission. The complaint alleges Amazon and senior executives gave false and misleading answers to advertisers who asked directly whether the auction format had changed, specifically so those advertisers would keep bidding as though nothing had.
The filing also describes a deliberate ramp pattern. Ahead of high-volume shopping days, the complaint says, Amazon carefully increased surcharges in stages to disguise the fact that it was inflating prices, letting the increases blend into ordinary seasonal cost-per-click volatility. Advertisers did notice on occasion: reporting on the complaint describes nearly two dozen advertiser complaints after cost-per-click spikes in the 2021 holiday period, and further complaints when Prime Day 2023 surcharges more than doubled CPCs for some accounts.
How does Amazon respond?
Amazon has rejected the case in strong terms. The company called the lawsuit “misguided” and said the complaint “fundamentally misunderstands how advertisers operate,” publishing its position rather than confining it to a litigation filing.
The substance of Amazon’s defense is behavioral. In a statement quoted in trade coverage, the company argued that “advertisers adjust bids based on real-world performance, not descriptions of auction mechanics.” The argument is that sophisticated advertisers optimize against outcomes such as cost per acquisition and return on ad spend, so what they were told about auction internals did not drive their bidding.
Amazon has also argued that price variation across placements and formats is a natural product of its auction system, and that advertisers receive appropriate information about pricing. Expect that framing to become a materiality fight: the FTC must show the misrepresentation was material to advertiser decisions, and Amazon’s own internal documents about bid shading and “downward spiral” risk cut against its public position.
Which states joined, and what can they recover?
Twenty-two state attorneys general joined the FTC: Alaska, Arizona, California, Colorado, Florida, Idaho, Illinois, Indiana, Iowa, Kentucky, Louisiana, Maryland, Nebraska, New Jersey, New York, North Carolina, Oklahoma, Pennsylvania, Rhode Island, South Carolina, Vermont and Washington. The bipartisan spread matters procedurally as much as politically.
The FTC is seeking an injunction and monetary relief under Sections 5 and 13(b) of the FTC Act. Reporting on the complaint describes five counts, covering misrepresentation, deceptive auction manipulation, concealment, unfair omissions and unfair billing practices.
The states bring a different toolkit. Under their own consumer protection statutes they are pursuing restitution, disgorgement and civil penalties, remedies that in several states carry per-violation penalties. With over a million affected advertising accounts, per-violation math is the reason state participation raises Amazon’s exposure ceiling well beyond what the federal counts alone would support.
Why does this land differently from Amazon’s other cases?
Amazon is not short of regulatory matters, but this one has a distinct shape. Its previous high-profile FTC consumer protection case concerned Prime enrollment and cancellation, where the agency targeted a cancellation flow Amazon internally called “Iliad.” That case settled in September 2025, three days into trial in Seattle federal court, for $2.5 billion: a $1 billion civil penalty plus $1.5 billion in refunds to an estimated 35 million customers.
The separate 2023 monopolization case, which alleges Amazon illegally suppressed marketplace competition, survived partial dismissal and remains scheduled for trial in early 2027. The new advertising complaint is the third distinct FTC action against the company in recent years.
What separates the surcharge case is the identity of the injured party. Prime was about consumers. The 2023 antitrust case is about rivals and market structure. This complaint is about Amazon’s paying business customers, the sellers and brands who fund the advertising machine, and it arrives while the company is separately defending pricing conduct in state court after beating California’s bid for a preliminary injunction in its price-fixing case, which is now set for trial in January 2027.
Amazon’s three FTC matters compared
| Matter | Filed | Theory | Injured party | Status |
|---|---|---|---|---|
| Prime enrollment (“Iliad”) | 2023 | Deceptive enrollment and cancellation | Consumers | Settled September 2025 for $2.5bn |
| Marketplace monopolization | 2023 | Antitrust, suppression of competition | Rivals and sellers | Trial scheduled early 2027 |
| Ad auction surcharges | August 31, 2026 | Deception, concealment, unfair billing | Advertisers (brands and sellers) | Newly filed, no trial date |
The pattern regulators are drawing is consistent: enforcement aimed at the gap between what a platform says a mechanism does and what the mechanism does. Each of the three matters attacks a different surface, but the common thread is an automated system whose behavior the counterparty cannot independently observe.
What does this mean for advertisers right now?
Nothing about the filing changes today’s auction. Amazon has not been ordered to alter its pricing, and no preliminary injunction has been granted or, on the current record, sought. Advertisers should plan for the mechanics to stay as they are through at least the 2026 holiday season.
The practical implication is analytical rather than operational. If the complaint’s account is accurate, the assumption that a higher bid is protected by second-price mechanics has been wrong for years, and bid strategies calibrated on that assumption have been systematically overpaying.
What to check in your own account data
The complaint gives advertisers a testable hypothesis. If a soft reserve pushes paid price toward bid, the ratio of average cost-per-click to maximum bid should sit close to one and should move upward around peak events.
Pull cost-per-click against your bid caps for Sponsored Products across ordinary weeks and compare it with Prime Day and Black Friday windows in 2023, 2024 and 2025. A stable ratio near your bid ceiling is consistent with first-price behavior. A ratio that jumps specifically in event windows, beyond what your own competitive density explains, matches the ramp the complaint describes.
That analysis is worth running before the coming holiday period regardless of how the litigation resolves, because it feeds directly into bid ceilings. Advertisers who reduce maximum bids toward true value get the bid-shading protection a first-price auction requires, which is precisely the “downward spiral” Amazon’s internal documents reportedly feared.
Where budgets may move
Amazon’s position in retail media is dominant enough that a pricing dispute does not easily translate into an exodus. The company accounted for more than three quarters of U.S. retail media ad spending in 2025, a function of its roughly 40% share of U.S. e-commerce sales.
Forecasts put Amazon’s 2026 U.S. retail media spend at $56.71 billion against $5.99 billion for second-place Walmart.
That gap is the reason substitution pressure is limited in the near term. Still, marginal budget does move, and rival surfaces have been actively courting it, including through the expansion of in-store retail media inventory and through changes in how search platforms allocate shopping traffic. Google’s decision to make local inventory ads the default in Shopping campaigns from August 31, 2026, arrived the same week and pushes in the same direction.
Retail media scale, for context
| Platform | 2026 U.S. retail media ad spend (forecast) | Position |
|---|---|---|
| Amazon | $56.71bn | Over 75% of U.S. retail media spend in 2025 |
| Walmart | $5.99bn | Second place, roughly one ninth of Amazon |
| All other retail media | Remainder | Highly fragmented |
Figures are external forecasts rather than company disclosures, and they measure U.S. retail media specifically rather than total advertising revenue. Amazon’s global advertising revenue across all formats was roughly $68 billion in 2025 by the same class of estimate.
How did this run for seven years without detection?
The obvious question about a surcharge operating since 2019 is why a market full of sophisticated, well-resourced advertisers did not identify it. The complaint’s answer is that the surcharge was designed to be invisible to exactly the measurement tools advertisers use.
Amazon does not publish the losing bids in its auctions. An advertiser sees what it bid and what it paid, but it cannot see the second-highest bid that supposedly set the price. Without that reference point, a payment equal to the full bid is indistinguishable from a genuine second-price outcome in a tightly contested auction.
Why rising costs looked like competition
Cost-per-click in Amazon search has risen for years, driven by real and well-documented factors: more sellers, more categories opening to advertising, and higher advertising intensity in crowded niches. Any surcharge layered on top of that trend inherits a ready-made explanation.
The complaint alleges Amazon exploited precisely this ambiguity by ramping surcharges hardest during peak events, when CPC volatility is highest and least likely to prompt structural questions. An advertiser seeing costs double on Prime Day has an intuitive explanation available that does not involve auction mechanics.
The filing indicates the disguise was imperfect. Nearly two dozen advertisers complained after cost-per-click spikes in the 2021 holiday period, according to reporting on the complaint, and further complaints followed when Prime Day 2023 surcharges more than doubled CPCs for some accounts. The FTC alleges those inquiries were met with false and misleading answers.
The measurement blind spot
Most advertiser optimization runs on outcome metrics: return on ad spend, advertising cost of sale, cost per acquisition. Those metrics tell an advertiser whether a campaign is profitable. They do not reveal whether the price paid matched the stated auction rule.
An advertiser paying above the second-price benchmark simply sees a lower return on ad spend and responds by lowering bids or reallocating budget. The system absorbs the overcharge as apparent inefficiency rather than surfacing it as a pricing discrepancy. That is why Amazon’s public argument, that advertisers optimize on performance rather than on auction descriptions, is double-edged: performance optimization is exactly what would mask the conduct the FTC alleges.
This asymmetry is also why the case matters beyond Amazon. Every closed-auction advertising marketplace shares the same structural property, in that the seller runs the auction, sets the rules, observes all the bids and reports the outcome. Auditability depends on the platform’s own disclosure, which is the condition the complaint says Amazon exploited.
How strong is the FTC’s case?
Deception cases under Section 5 turn on three elements: a representation or omission, materiality, and the likelihood of misleading a consumer acting reasonably. On the first element the complaint looks well supplied, since Amazon’s second-price representations were public, repeated and documented across its own materials.
Materiality is where Amazon will fight, and its “advertisers adjust bids based on real-world performance” argument is aimed squarely at it. The counterweight is the internal record: documents predicting a “downward spiral” if the truth emerged are difficult to reconcile with a claim that the representation did not affect bidding.
The concealment allegations, if supported at trial, carry independent weight. Giving false answers to advertisers who asked directly whether the format had changed is harder to characterize as an omission in a complicated technical product, and it is the kind of conduct that tends to affect both liability and remedy.
The open questions
Damages methodology is the largest unresolved issue. Calculating what advertisers would have paid in a counterfactual GSP auction requires modeling how bids would have adapted, and Amazon will argue that advertisers would simply have bid higher in a disclosed first-price world, leaving clearing prices similar.
Jurisdictional scope is a second question. The complaint is framed around U.S. advertising customers, but Amazon runs the same advertising infrastructure across its international marketplaces, which raises the question of whether other regulators open parallel files. The theory transfers cleanly to the EU and UK, where deceptive commercial practice rules and platform transparency obligations cover much the same ground.
A third question is whether the case migrates from deception into the algorithmic accountability space that regulators have been steadily building out. The FTC’s interest in how automated systems steer commercial outcomes is well established, and an auction that inserts an “invented auction participant” sits close to that boundary.
What happens next?
No trial date has been set. The immediate procedural steps are Amazon’s response to the complaint, which will almost certainly include a motion to dismiss, and the parallel question of whether the state claims proceed on the same schedule as the federal ones.
Amazon’s motion to dismiss is likely to argue that its representations were accurate at a level of generality advertisers understood, that auction pricing disclosures were adequate, and that the FTC’s damages theory is speculative. On the record quoted in the complaint, a full dismissal looks unlikely, though narrowing of individual counts is plausible.
Watch also for follow-on private litigation. A government complaint quoting internal documents this directly is an invitation to class actions from advertisers, and those cases typically arrive within weeks rather than months. Any such filings would concentrate in the same Western District of Washington venue.
Dates worth marking
Amazon’s responsive pleading deadline will be the first hard date on the docket. Beyond that, the early 2027 monopolization trial and the January 2027 California price-fixing trial mean Amazon faces a concentrated litigation calendar in which the advertising case will compete for the same internal witnesses.
For advertisers, the operative calendar is commercial rather than judicial. Prime Day and Black Friday 2026 are the next windows in which, on the complaint’s account, surcharge behavior would be most pronounced, and they are the cleanest natural experiment available for testing bid-to-CPC ratios in your own data.
Frequently asked questions
What is a “soft reserve price” and is it illegal?
A soft reserve price, in the FTC’s description of Amazon’s internal terminology, is an undisclosed floor added to auction pricing that pushes the amount charged above what the second-price mechanism would produce. Reserve prices are legal and common in auctions. The FTC’s claim is not that the reserve existed but that it was concealed while Amazon continued describing the auction as second-price.
How much does the FTC say Amazon collected?
The FTC’s complaint says the scheme “likely extracted tens of billions of dollars” from advertising customers. Reporting on the filing puts the figure above $20 billion collected from roughly 1.2 million U.S. advertising customers since the 2019 auction change. Amazon disputes the characterization of the conduct.
Which advertisers were affected?
The complaint covers advertisers bidding for Sponsored Products, Sponsored Brands and display placements on Amazon.com and its mobile app. The FTC says more than one million brands and sellers participated, including over 500,000 small and medium-sized businesses. The most specific escalation figures in the complaint relate to Sponsored Products.
Does this change how Amazon ads are priced today?
No. The filing of a complaint does not alter Amazon’s auction mechanics, and no injunction has been entered. Any change to pricing would follow either a court order, a settlement, or a voluntary decision by Amazon.
Can advertisers get refunds?
Not at this stage. The FTC is seeking monetary relief and the participating states are pursuing restitution, disgorgement and civil penalties, but any advertiser recovery would depend on the outcome of the litigation or a settlement. The Prime case offers a partial precedent, since that 2025 settlement included $1.5 billion in consumer refunds, though it involved consumers rather than business customers.
What should advertisers do now?
Audit the relationship between your maximum bids and realized cost-per-click for Sponsored Products, comparing ordinary periods against Prime Day and Black Friday windows. If realized CPC sits consistently near your bid ceiling, your bidding should be calibrated for first-price conditions, which generally means bidding true value rather than relying on second-price protection.
Why did 22 states join a federal case?
State consumer protection statutes provide remedies the FTC Act does not, notably civil penalties assessed per violation and restitution under state law. With over a million affected accounts, those per-violation remedies materially increase potential exposure. The participating states span both parties, which reduces the likelihood the case is characterized as politically driven.
How does this relate to the 2023 FTC antitrust case against Amazon?
They are separate matters with different legal theories. The 2023 case alleges monopolization of online marketplaces and is scheduled for trial in early 2027. The August 2026 complaint is a consumer protection case about deception and concealment in advertising auctions, and it does not require the FTC to prove monopoly power.
Could regulators outside the United States open similar cases?
It is plausible but not announced. Amazon operates comparable advertising infrastructure across its international marketplaces, and EU and UK rules on misleading commercial practices and platform transparency address similar conduct. No non-U.S. regulator has publicly opened a parallel investigation into the auction surcharges as of this writing.
The bottom line
The FTC’s case against Amazon’s advertising auctions is narrower than an antitrust action and, on the documents quoted so far, potentially harder to defend. It does not require proving market power or consumer harm across a market. It requires proving that Amazon said one thing about how prices were set and did another, and that advertisers changed their behavior because of it.
Amazon’s public defense, that advertisers optimize on results rather than on auction descriptions, is a real argument and may narrow the damages even if liability attaches. But the internal documents the FTC has chosen to quote, including a characterization of the system as a “clever non-transparent way to charge first price,” were selected precisely because they undercut that defense.
For the brands and sellers funding roughly $68 billion of annual advertising revenue, the practical takeaway does not depend on who wins. If the paid price tracks the bid roughly 80% of the time, the auction behaves as first-price, and bids should be set accordingly. That adjustment is available now, without waiting for a court.
The full complaint and the agency’s summary of the allegations are available from the Federal Trade Commission.