As of November 7, 2026, the Federal Trade Commission and 22 state attorneys general have made public the previously redacted portions of their advertising suit against Amazon, filed in the Western District of Washington. The unsealed passages allege that Amazon charged advertisers as much as 50 percent above the price its own search auction produced on ordinary days, and as much as 250 percent on peak days such as Prime Day and Black Friday. The figures are allegations, not findings, and Amazon denies wrongdoing. For anyone setting promotional budgets, the practical question is narrower than the legal one: if peak-week click prices can move for reasons a campaign cannot see, then last year's peak-week benchmark is not a target, it is an assumption.
What the unsealed filing actually adds
Three details in the release are worth separating. The first is the mechanism: the complaint describes a soft reserve price, a floor that can sit above what competing bids alone would have produced, which is a different claim from collusion or from a rigged score. The second is the money: 87 million dollars of extra worldwide revenue in the 2022 holiday season, and an internal 2023 discussion about finding at least 40 million dollars from reserves to close a gap to plan. The third is what advertisers were told. The complaint quotes an account manager answering a question about whether the second-price auction had changed with a denial that there was "a bug, or a change in auction style".
Why a hidden surcharge lands inside a promotional plan
Advertising cost per order is the number a promotional budget is built on, and it is calculated from clicks, click-to-order rate and the price paid per click. A surcharge moves the third input without touching the first two, so a campaign can hit its traffic target, hold its conversion rate and still miss its contribution margin, because the cost of a click rose while every dashboard kept reporting the same structure. That is why the allegation matters far beyond the litigation: it separates the price a team believes it bid from the price the auction actually charged.
The distortion compounds at peak. Discounts are funded from a planned margin, and the margin was computed against a cost per order that the team measured in a normal week and then carried into Black Friday. If the peak-week click price carried a peak-only markup, the discount was funded from a number that did not exist in the week it was spent. The campaign did not fail because the offer was weak; it failed because the cost line under the offer was too optimistic from the start.
| Line in the plan | What a surcharge does to it | What to watch in the report |
|---|---|---|
| Cost per click | Rises above the auction outcome | Price charged against the maximum bid |
| Click-to-order rate | Unchanged, because the shopper never sees it | Conversion rate by campaign |
| Cost per order | Rises with the click price | Cost per order versus the quiet-week baseline |
| Contribution margin | Falls faster than the discount was funded for | Margin per order on peak dates |
How to audit peak-week pricing without waiting for the court
Rebuild the baseline from quiet weeks
Start from weeks with no event and no sale, and build the cost per order there. A promotional calendar priced off a quiet-week cost, with the peak uplift entered as an explicit input, is auditable. A calendar priced off last year's peak, which may already contain a surcharge, is a forecast of the same mistake. The difference is small in the spreadsheet and large in the result, because the second version treats an inflated price as the normal cost of doing business in November.
Log the gap by campaign, not by month
Monthly averages hide exactly the events that hurt. A campaign-level log that pairs the maximum bid with the price actually charged, for the same date and the same placement, produces the only evidence that can support a recovery claim or an internal challenge. Without it, a team arguing that it overpaid has a feeling; with it, the team has a series of charge-versus-bid observations that a lawyer, an auditor or a procurement review can use.
What the defence settles and what it leaves open
Amazon's response is that advertisers never pay more than their maximum bid, that its reserve prices reflect the value of a placement, and that average winning bids fell over the period. Read carefully, that defence answers a different question from the one the complaint asks. A cap on bids does not rule out a price set above what the auction alone would have produced, because the two statements can both be true at once: the advertiser paid less than the ceiling and more than the auction. The case will turn on that gap, and until a court rules, the only defensible position for a media team is to measure it rather than assume it away.
The broader point for retail and marketplace advertisers is that platform pricing is now a supply-chain risk, not a line item to be negotiated once a year. A brand that monitors the price of its own products across channels, down to the coupon and the shelf, but never monitors the price it pays for the traffic that sells them, is watching half the cost base. The unsealed filing is a reminder that the other half can move quietly, in a week when everyone is looking at discounts.
FAQ
What exactly did the FTC unseal on November 5, 2026?
Previously redacted portions of an advertising lawsuit filed in August 2026 in the Western District of Washington, which allege that Amazon charged advertisers above the price its search auction produced, with markups reaching 50 percent on ordinary days and 250 percent on peak days. All of it remains allegation.
Does the case mean advertisers are owed money?
No. A government action does not automatically return money to any advertiser, and no court has ruled on the allegations. A separate claim would rest on a brand's own billing records and its advertising terms, which is why campaign-level bid-versus-charge data is the practical asset to preserve now.
Why rebuild the baseline from quiet weeks?
Because peak-week benchmarks from 2021 to 2023 may already include an alleged surcharge. Pricing a 2026 promotion off an inflated peak cost turns the overpayment into a target, while a quiet-week baseline with an explicit peak uplift keeps the assumption visible and testable.
What should a media team log first?
Billing and campaign exports for every peak period, then the maximum bid and the price actually charged for the same date and placement. The gap between the two is the number that can be audited, escalated or defended.
