On August 31, 2026, the Federal Trade Commission and attorneys general from 22 states sued Amazon in the U.S. District Court for the Western District of Washington. The allegation, in one sentence: for seven years, the Sponsored Products auction you have been bidding into was not the auction Amazon told you it was.
You have read the headlines by now. What you probably have not done is work out what the complaint means for the numbers sitting in your own account this morning.
That is what this post is for. No verdict, no outrage cycle — just what is alleged, what Amazon says back, why the two sets of numbers do not actually contradict each other, and the four things worth pulling from your own reports before you touch a single bid.
Table of Contents
What the complaint actually alleges
Amazon has historically run Sponsored Products through a generalized second-price (GSP) auction. The pitch is simple, and it is the reason most of us bid the way we do: you set a max bid, and if you win, you pay one cent more than the next-highest bid. Your max is a ceiling, not a price. Bid your true value and the auction protects you from overpaying.
The FTC alleges that in 2019, Amazon quietly inserted a charge inside that mechanism — described in internal documents as a “soft reserve price” — that caused advertisers to pay substantially more than the second-price outcome would suggest. The complaint’s specific claims:
- The practice ran for over seven years and touched more than one million brands, across Sponsored Products, Sponsored Brands, and Display.
- The share of Sponsored Products advertisers paying their own bid rose from 30–40% in 2021, to 70% in 2022, to roughly 80% by 2024. In other words: the ceiling quietly became the price.
- Amazon internally referenced “invented auction participants” and “proxy 2nd price” bids — synthetic competition inserted to lift the clearing price.
- Prices allegedly rose during normal periods and rose far more during Prime Day and Black Friday, with increases ramped in gradually beforehand so the size of the jump was harder to see.
- When advertisers asked directly whether the auction format had changed, the complaint says Amazon gave “false and misleading answers.”
- Internal documents allegedly acknowledge the practice risked “irrevocable damage to advertiser trust.”
The FTC puts the revenue extracted at tens of billions of dollars over the period, against an Amazon advertising business that approached $70 billion in 2025.
What Amazon says back
Amazon’s public response is worth reading in full, because it is not a non-denial. It is a direct argument with numbers attached:
- Average cost-per-click was flat from 2019 to 2024 once adjusted for inflation.
- Winning bids fell 50% from 2019 to 2025.
- Roughly 92% of placed ads were not the highest bid in their auction.
- Conversion rates rose 24% from 2021 to 2025, and Amazon says its relevancy-weighted auction saved advertisers over $8 billion between 2021 and 2025.
- Soft and hard reserve prices are “common across the industry.”
- The FTC reviewed 1.5 million pages and “cherry-picked” a handful — including internal training material with, by Amazon’s count, 23 total enrollments.
Nothing here has been tested in court. Amazon has not been found liable of anything, and a complaint is an allegation.
The part nobody is saying out loud: both sets of numbers can be true
Here is the thing that should interest you more than the litigation.
Look closely at the two headline stats. The FTC says roughly 80% of advertisers were paying their own bid. Amazon says 92% of placed ads were not the highest bid.
Those are not the same measurement, and neither one refutes the other.
“Not the highest bid” is about who wins — it describes a relevancy-weighted auction where a lower bid with better predicted conversion can beat a higher one. “Paying your own bid” is about what you are charged after you win. You can absolutely win an auction with the fourth-highest bid and still be charged your full maximum. The two statistics describe different halves of the transaction, and each side picked the half that flatters it.
Same with “winning bids fell 50%.” If clearing prices moved closer to advertisers’ maximums, rational advertisers respond by lowering their maximums. Falling bids are consistent with a healthier auction and consistent with advertisers getting squeezed and pulling their ceilings down in response. The stat does not distinguish between them.
This is the practical takeaway, and it holds regardless of how the case resolves: you cannot infer your auction mechanics from Amazon’s aggregate statistics, and you cannot infer them from the FTC’s either. You can only infer them from your own account.
Four checks to run this week
None of these prove or disprove anything about the lawsuit. All of them tell you something useful about how your money is actually being spent.
1. Calculate your bid-to-CPC ratio.
Pull a Targeting report for the last 30 days. For every keyword with meaningful click volume, divide average CPC by your current max bid.
- Ratios clustering at 0.90–1.00 mean you are effectively in a first-price auction. Your ceiling is your price, and every cent of bid headroom you leave is a cent you spend.
- Ratios in the 0.50–0.75 range mean second-price behavior is still doing work for you.
Do not average this across the account — the average hides everything. Segment it by match type, by placement, and by how competitive the term is. Most accounts have both patterns running side by side.
2. Run a bid elasticity test on a stable cohort.
Pick 15–20 keywords with steady click volume and no recent changes. Cut bids by 15%. Wait 14 days. Check what happened to CPC.
- In a genuine second-price auction, a 15% ceiling cut barely moves your CPC until you drop below the clearing price — then impressions fall off a cliff.
- If your CPC tracks your bid almost one-for-one while impressions hold, you were paying your own bid, and the ceiling was doing all the pricing.
This is the single most informative test in this post, and it costs you nothing but two weeks of patience.
3. Baseline your CPCs before Q4 — starting now.
The complaint alleges prices were ramped up gradually in the weeks before peak events, not on the day of. If that is happening in your categories, the only way to see it is to have a clean baseline from before the ramp begins.
Export weekly CPC by campaign for the last four weeks and save it somewhere that is not Seller Central. Then keep exporting weekly through Black Friday and Cyber Monday. In December you will be able to answer “when did my CPC actually start climbing” with a chart instead of a feeling. Most sellers cannot answer that question today, which is precisely the problem.
4. Recheck your break-even ACoS against real CPCs, not assumed ones.
If you built your target ACoS on the assumption that second-price mechanics would keep your realized CPC comfortably under your max bid, and check #1 shows you sitting at 95% of your ceiling, then your break-even math has been quietly wrong for a while. Rebuild it from your actual trailing-30-day CPC.
This is the check that changes money this month. The other three change how well you understand your account.
| The question | What the FTC alleges | What Amazon says |
|---|---|---|
| How the auction priced your clicks | A hidden charge — internally a “soft reserve price” — was inserted inside the second-price auction in 2019. | Soft and hard reserve prices are “common across the industry” and standard practice. |
| What you actually paid | Share of advertisers paying their own bid rose from 30–40% (2021) to 70% (2022) to roughly 80% (2024). | Around 92% of placed ads were not the highest bid in their auction. |
| Where CPCs went | Prices rose in normal periods and far more before Prime Day and Black Friday, ramped in gradually to obscure the jump. | Average CPC was flat 2019–2024 inflation-adjusted; winning bids fell 50% from 2019 to 2025. |
| What it cost advertisers | Tens of billions of dollars over seven years, across more than 1 million brands. | Conversion rates rose 24% (2021–2025); relevancy-weighted auctions saved advertisers over $8B. |
| The internal documents | References to “invented auction participants” and “proxy 2nd price” bids, and to “irrevocable damage to advertiser trust.” | Cherry-picked from 1.5 million pages, including a training course with 23 total enrollments. |
UNPROVEN Filed August 31, 2026 in the U.S. District Court for the Western District of Washington. None of these allegations has been tested in court.
What this does not mean
Do not budget for a settlement. There is no claims process, no damages figure, and no timeline. Litigation of this size runs for years. Treat any recovery as zero until it is not.
Do not pull spend as a protest. If Sponsored Products is profitable at your real CPCs, it is profitable regardless of how the clearing price is calculated. An unfair auction, if proven, is an argument for tighter bid discipline — not for handing the channel to competitors who will keep bidding.
Do not assume this is over. The complaint alleges the practice was still running as of 2024, and nothing in Amazon’s response says it has stopped. Amazon’s position is that reserve prices are standard and appropriate, which is a defense of the practice, not a retirement of it.
What we still do not know
- Whether soft reserve prices are still active in the auction today, and at what level.
- Whether the mechanics differ by category, by marketplace, or by ad type — the complaint covers Sponsored Products, Sponsored Brands, and Display together, but those auctions are not identical.
- Whether Sponsored Prompts inventory, which became billable on March 25, 2026 and shares your existing campaign bid, is priced the same way as standard placements. There is no separate bid line and no separate budget for prompts, which makes that one genuinely unauditable from the advertiser side right now.
- What, if anything, Amazon changes voluntarily while the case is pending.
We will update this post as the docket moves.
The bottom line
The lawsuit is not your problem to solve. Your bid-to-CPC ratio is.
If it turns out the auction has been charging you your ceiling for years, the fix was always the same fix: know your break-even, set your maximum where you would genuinely be willing to pay it, and stop treating the max bid as a theoretical number the auction will politely round down for you. The advertisers who have been managing bids as if every auction were first-price have been right by accident since about 2022.
Pull your Targeting report. Divide CPC by bid. Sort descending. Then decide.
Ad Badger’s bid automation adjusts your maximums against your actual break-even ACoS, not against an assumption about how the auction prices them. See how it works — or book a one-on-one with Michael.
