If regulators are right, Amazon sold its advertisers a second-price auction and charged many of them a first-price bill. This is not a story about reserve prices. Reserve prices are as old as the auction block, and Amazon says such floors are common in the industry 22. The story is about disclosure. Federal and state regulators allege that the pricing mechanism Amazon marketed and the pricing mechanism it ran were not the same machine 11 — and that advertisers had no way to see the difference.
I come from a century of retail advertising, and the rule has not changed: the question is not whether it works, but how you know it works. On the record described here, advertisers could not know. That creates undetected risk — not only for the advertisers, but for one of Amazon's most profitable growth engines.
What the Complaint Alleges
The Mechanism
The FTC and 22 state attorneys general allege that Amazon told advertisers for years that auction winners would pay one cent more than the next-highest bid, then — beginning in 2019 — added an undisclosed "soft reserve price" after the winner had already been selected 13. That reserve, the complaint says, could function as an "invented auction participant": a bidder who does not exist, raising the clearing price above the actual competing bid 3.
The distinction matters. A floor price, stated plainly, is honest commerce; a catalog house that refuses to sell below cost is running a reserve, and says so in the catalog. The allegation here is concealment against a stated mechanism. Internal documents sharpen it: one reportedly described the surcharge as delivering prices "beyond what can be achieved through advertiser competition" 23. Read plainly, that describes a floor set above the market — a substitute for competition, not a term of it.
The Scale and the Trend
The FTC alleges that by 2024, roughly 80% of Sponsored Products winners paid their entire winning bid, up from approximately 30–40% in 2021 15. Other accounts place the intervention rate at 70–80% of ad auctions in recent years 9. If those figures are substantiated, my inference is straightforward: effective pricing moved materially toward first-price behavior even as the system was marketed as second-price.
The alleged exposure is broad. More than 1.2 million advertisers were affected, the FTC states, including more than 500,000 small and medium-sized businesses 3,16. The case carries the institutional weight of a politically diverse coalition of 22 states 9. And the timing allegations cut where sellers feel it most: the complaint alleges higher surcharges on Prime Day, Black Friday, and other high-volume periods 11 — precisely the events around which sellers plan their acquisition spending.
Amazon's Defense — and Why No One Can Audit It
The Company's Case
Amazon contests the theory of harm. It says its advertising policies are designed to show shoppers the most relevant ads 1,2,14. It points to performance: an average winning-bid decline of 50% and a conversion-rate increase of more than 24% from 2021 through 2025, offered as evidence that advertisers and customers both benefited 4,10,12,18. It has also updated its help materials to explain its use of reserve prices 10.
These are not empty numbers. A full-bid win that converts materially better is a different economics problem. Rising conversion can offset a higher cost per click, and cost-per-acquisition integrity depends on both sides of that ratio.
The Gap
But none of it can be checked from outside. Advertisers reportedly lack access to auction logs, which limits their ability to verify auction mechanics or to test either side's account 20. Amazon retains meaningful defenses and its commercial position is resilient; without auction-level auditability, however, neither the regulators' narrative nor the company's can be independently validated 12,17,20.
In the catalog business, we counted coupon returns. The coupon was the log. When the merchant holds the only ledger and the customer cannot see it, every number on both sides is a claim, not a measurement. That is the attribution collapse at the center of this case.
Consequences: Remedial Risk, Not a Damages Headline
The complaint does not specify a maximum exposure figure 23, and litigation of this kind could be prolonged; the FTC's recent record against large technology companies has been mixed 8,22. A damages number, then, is not where the risk lives. The identified primary risk is an injunction that forces an auction redesign: abandoning the soft-reserve mechanism could lower ad monetization per query and inject execution risk into Amazon's marketplace advertising operations 9,21. More broadly, mandated transparency or a redesigned auction could pressure the growth engine more than a damages award alone 5,6,19, touching monetization, operational execution, and advertiser confidence at once.
There is a governance dimension as well. Alleged concealed surcharges bear directly on the weight investors assign to business-conduct transparency 7. For a company whose marketplace runs on advertiser trust, an allegation of concealment is the costliest kind.
The alleged peak-period surcharges noted above matter beyond the courtroom, too: peak events anchor sellers' acquisition economics, which is where bid inflation would bite hardest.
The Lock-In That Makes Trust Expensive
Amazon's retail-media share and its Prime ecosystem contribute to advertiser lock-in, and advertisers are described as having limited alternatives despite greater scrutiny 22. Dependence of that kind monetizes well in the short run. It also raises the price of a transparency failure: a business cannot easily discipline a platform by switching when the platform is its primary route to high-intent shoppers.
One boundary should be drawn carefully. The broader record on seller fees, fulfillment incentives, and parity requirements is largely separate from this auction complaint. It is commercial context — evidence of limited negotiating leverage — but it is not corroboration of the auction allegations, which remain unproven. Conflating the two is exactly the kind of accounting error this case is about.
What Rigor Would Require
Sort what is known from what is inferred. Known: the allegations and their claimed scale; Amazon's stated metrics; the absence of advertiser access to auction logs. Inferred: any estimate of the true waste fraction, the direction of effective pricing, or the net effect of higher bids against better conversion. A more rigorous approach would close that gap in three moves:
- Open the auction ledger. Log-level access for affected advertisers, or an independent auditor, so that claims about the mechanism can be tested rather than asserted.
- Run transparent incrementality tests. The advertiser, not the platform, should hold the counterfactual, and conversion gains should be forced to reconcile with bid costs in one auditable frame.
- Separate disclosure from design. A stated reserve is defensible. A stated mechanism that differs from the running mechanism is the failure, whatever the reserve level.
The Question That Remains
The history of advertising is a history of unmeasured waste. Every merchant who ever bought space in a mail-order catalog asked the same quiet question: did I pay the going rate, or the rate the house set after the bidding closed? The FTC says Amazon's advertisers paid the second. Amazon says the numbers show they did fine. Neither claim can be proven from outside the logs. Until they open, ask yourself: if you could not see the reserve, what exactly were you bidding against?