The FTC Says Amazon Charged Your Full Bid 80 Percent of the Time. Test It on Your Own Account This Week.

On August 31, 2026, the Federal Trade Commission and the attorneys general of 22 states sued Amazon in the Western District of Washington. The allegation is not antitrust. It is deception, and it concerns a single piece of pricing logic inside the Sponsored Products auction.

Most of the coverage is about the damages number. Skip that. The number is not actionable, the case will run for years, and no seller should be building a plan around a hypothetical refund.

The recommendation: spend two hours this week running a bid-sensitivity test on a slice of your keywords, before Prime Big Deal Days. Not because the FTC says so, but because if the complaint is directionally right about how the auction prices clicks, then the way most sellers set max bids has been wrong for several years, and the fix costs nothing to test.

What the complaint actually alleges

Amazon has told advertisers for years that it runs a generalized second-price auction, where the winner pays one cent more than the next highest bidder. That representation appeared on Amazon’s website, in training videos, in Ads Academy courses, and in presentations from its sales organization.

The FTC alleges that beginning in 2019 Amazon added an undisclosed surcharge on top of the auction result, referred to internally as a “soft reserve price.” The complaint characterizes it as a second calculation applied after the winner and runner-up are already determined, with the winning bid itself as the only ceiling.

The number that matters operationally is the share of clicks charged at the advertiser’s own bid rather than at anything resembling a runner-up price. Per the FTC, for Sponsored Products that share ran between 30 and 40 percent in 2021, reached 70 percent in 2022, and hit approximately 80 percent in 2024.

The complaint quotes Amazon’s own people describing the mechanism. The executive in charge of Amazon Ads is quoted saying the price paid “isn’t set by an actual bidder” but is a “proxy 2nd price that we calculate.” Another document refers to an “invented auction participant.” Notes from a 2024 discussion among senior executives describe a “clever non-transparent way to charge first price” that has been an “incredibly effective way to drive revenue.”

The FTC also alleges the concealment was deliberate and for a specific reason: that advertisers bid higher than they otherwise would because they believe the auction is second-price, and that disclosing the change would trigger advertisers lowering their bids. The complaint quotes internal documents describing that outcome as “irrevocable damage to advertiser trust” and a “downward spiral.”

The Commission vote authorizing the filing was 2-0.

The part that is specifically a Q4 problem

One paragraph in the FTC release is the reason this is a September article rather than a general-interest one.

The complaint alleges Amazon increased prices on ordinary shopping days and applied “far greater increases to prices on high-volume shopping days like Prime Day and Black Friday,” and that ahead of those events Amazon “carefully ramps up surcharges” in a way that is hard to detect because seasonal demand is moving at the same time.

That is a specific, testable claim about the calendar you are about to enter. Prime Big Deal Days is roughly five weeks out. Black Friday and Cyber Monday are about twelve.

It also explains why so many sellers have had the same annual experience of watching CPCs jump in peak and attributing all of it to competitor demand. Some of it certainly is competitor demand. The complaint alleges some of it is not, and that the overlap is the point.

What Amazon says back

Amazon rejects the case, and the rebuttal deserves to be stated rather than waved at, because parts of it are relevant to how you should act.

Amazon says no advertiser was overcharged, that average Sponsored Products cost-per-click was flat in inflation-adjusted terms between 2019 and 2024, and that roughly 92 percent of Sponsored Products ads selected in 2024 were not the highest bid, because ad rank weighs relevance alongside price. It says conversion rates improved for individual advertisers over the period, that relevance-weighted ranking produced materially better results than a bid-only auction would have, and that advertisers saved billions as a result. It characterizes the FTC’s evidence as outdated training materials and stray internal emails, notes that reserve pricing is common across the ad industry, and states that no advertiser ever paid more than the maximum bid they authorized.

That last point is worth holding onto, because it is not in dispute and it is the whole basis of the practical response below. Nobody alleges Amazon charged above your bid. The allegation is that it charged your bid, far more often than you were told to expect. Those are very different problems, and only one of them has a fix you control.

These are allegations in a complaint and statements from a defendant. Neither has been tested in court.

Why this changes how you set a max bid

In a true second-price auction, your bid should be close to your genuine maximum willingness to pay. That is the entire design. You are protected from overpaying by the rule, so bidding your true value is rational, and it is what Amazon’s guidance has encouraged for years.

In a first-price auction, that logic inverts. If you pay what you bid, then bidding your maximum willingness to pay guarantees you capture none of the surplus. The correct behavior is bid shading: bid down toward the minimum needed to win, discover it iteratively, and treat your true valuation as a ceiling you never actually submit.

Most Amazon sellers we see are still bidding the first way. The max bid is set as “I would tolerate paying $2.40 for this click,” on the reasonable assumption that actual CPC lands well below it. If the auction charges your bid most of the time, that seller has been paying $2.40 for clicks they could have won at $1.90.

Compounding this: dynamic bidding can raise your submitted bid substantially above what you set in the interface, which means the number being charged may not be the number you think you are testing. If you are running up-and-down bidding, your effective bid is not your set bid, and that needs to be controlled for before any of this is measurable.

The diagnostic, and it takes about a week

You do not need to accept or reject the FTC’s characterization to find out how your own account behaves. Amazon does not itemize charges per click, but auction behavior still shows up in the aggregate if you construct the test properly.

Pick a clean test set. Twenty to forty keywords in exact match, with enough click volume to move out of noise inside seven days, where you have been winning placement comfortably. Comfortable winners are the point: in a genuine second-price auction, a keyword you win with room to spare should show a CPC that barely moves when you trim the bid, because you are paying the runner-up price either way.

Control for dynamic bidding first. Set the test campaigns to fixed bids or down-only for the duration. Otherwise the platform is adjusting your bid underneath the experiment and you are measuring nothing.

Cut bids 20 percent on half the set, hold the other half flat. Matched pairs by volume and CPC where you can. Run seven to ten days.

Read the CPC, not the ACOS. If average CPC on the cut group falls by something close to 20 percent, your effective price is tracking your bid, which is first-price behavior. If CPC barely moves while impression share holds, you are being priced off a runner-up, which is second-price behavior. Watch impression share alongside it, because a bid cut that loses you the placement entirely tells you something different from one that keeps the placement at a lower price.

Then decide per keyword, not per account. The realistic answer is mixed. Some of your keywords will look first-price and some will not, and the ones that look first-price are where there is money sitting on the table.

Run it now, in the first half of September. Two reasons. It is the last clean read you will get before demand starts moving on its own, and if the complaint’s claim about peak-period surcharge ramping is accurate, the pre-peak window is exactly when you want your bids already corrected rather than mid-experiment.

What not to do

Do not budget for a settlement. This is a complaint filed last week. Restitution, if it ever arrives, is years out and may not be money. It is not a receivable.

Do not cut bids across the account as a reaction. The FTC’s own framing is that advertisers lowering bids en masse is the outcome Amazon feared. That does not make an untested 30 percent haircut on your whole catalog a good idea five weeks before peak. You will lose placement on terms that were fine.

Do not assume today’s behavior is permanent. An auction under this much scrutiny may not price the same way in November as it did in July. Re-run the diagnostic after peak rather than treating one week’s result as a standing fact.

Do not stop advertising on principle. Sponsored Products remains the highest-intent paid traffic in US eCommerce. A pricing dispute about the auction rule is not a reason to leave the auction.

The honest summary

The legal case will take years and its outcome is not something you can plan around. What is available now is narrower and more useful: a credible, document-heavy allegation that the auction you have been bidding into does not price the way you were told, published five weeks before the highest-spend quarter of the year, along with Amazon’s own statement that it never charges above your authorized maximum.

Put those two together and the action is obvious. Your max bid is doing more work than you assumed. Find out how much, on your own keywords, with your own data, in the quiet part of September.

That is a small test with an asymmetric payoff. If your CPCs move with your bids, you have found margin that was always yours. If they do not, you have spent a week and confirmed your bidding was already right, which is worth knowing before you scale spend into Q4.

Structuring that test properly, controlling for dynamic bidding, and reading the result without fooling yourself is the kind of work we run for the brands we operate. If you want a second set of eyes on your bid structure before peak, book a strategy assessment and we will go through your account with you.