Skip to main content

FTC Sues Amazon Over $20 Billion in Alleged Ad Surcharges: What Advertisers Can Control

The FTC and 22 states say Amazon quietly rewrote its ad auction in 2019 and never told buyers. The transferable lesson is not about Amazon: you cannot audit an auction from inside the platform that runs it.

Mauricio Valdivia

Mauricio Valdivia

·11 min

FTC Sues Amazon Over $20 Billion in Alleged Ad Surcharges: What Advertisers Can Control

Nothing in the dashboard looked wrong

A seller watching Sponsored Products in 2019 would have seen cost-per-click drift up a little. Then a little more the following quarter. There was no notice, no changelog entry, no line item to point at. Just a number that was slightly worse than last month, in a market where numbers are slightly worse than last month all the time.

On August 31, 2026, the Federal Trade Commission and 22 state attorneys general filed suit saying that drift was not the market.

The case is FTC et al. v. Amazon.com, Inc., No. 2:26-cv-03097, filed in the U.S. District Court for the Western District of Washington. The FTC's own summary of it is that the agency and 22 state attorneys general "allege Amazon concealed unfair charges in its digital advertising auction pricing." The Commission vote authorizing staff to file was 2-0.

None of it has been decided. Amazon calls the suit misguided and disputes the merits in detail, and every number below is an allegation in active litigation. So the part worth your time is not the verdict, which does not exist yet. It is the mechanism the complaint describes, and the uncomfortable question it raises for anyone who buys media anywhere: if this happened on the platform you spend on, how would you have known?

What the FTC and 22 states filed on August 31

The filing is long, but its spine is short. A pricing rule changed, buyers were not told, and the reporting they had could not have shown them.

The case, the venue and the scale

The complaint runs 181 pages and names the FTC alongside the attorneys general of 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. That is 22, and it is worth counting rather than trusting a headline.

The agency states the scale in three numbers, and each one is doing work:

  • Duration. The alleged conduct ran "for over seven years." Note that span, because some coverage compressed it.
  • Reach. Amazon "covertly and substantially increased the prices that more than one million brands and sellers were required to pay to advertise on its platform."
  • Who absorbed it. Among them, per the agency, are "over 500,000 small- and medium-sized businesses that participated in auctions for advertising placements."

Two documents, two different numbers

This is the part most coverage flattens, and it is the one detail to get right if you repeat any of this.

The FTC's press release does not say twenty billion dollars. It says the complaint alleges the scheme "has likely extracted tens of billions of dollars from its unwitting advertising customers." The specific figure lives one document over. Paragraph 1 of the filed complaint pleads that Amazon "has secretly and systematically overcharged its approximately 1.2 million advertising customers" and that the scheme "has likely illegally extracted over 20 billion dollars."

DocumentWhat it says about the amount
FTC press release"tens of billions of dollars"
Complaint, paragraph 1"over 20 billion dollars"

So the honest sentence is: the FTC's complaint alleges more than $20 billion. Not "the FTC announced $20 billion," and not a number a journalist rounded up. It also comes with a caveat you should carry: the average surcharge rates that support the estimate are redacted in the public version of the filing, so the topline is an allegation whose arithmetic is sealed.

What is being asked for, and what will not happen soon

The FTC brings the action for alleged violations of Section 5(a) of the FTC Act and, in the complaint's words, "seeks a permanent injunction and other relief pursuant to Section 13(b) of the FTC Act." The state attorneys general go further, seeking "injunctive relief, recission or reformation of contracts, restitution, the refund of monies paid, disgorgement of ill-gotten gains, civil penalties, attorneys' fees, expenses, and costs."

Translate that into a planning assumption rather than a hope:

  • There is no refund program to sign up for and no claims portal to register with.
  • There is no date. A contested federal case of this size, against a company of this size, is measured in years rather than quarters.
  • The likeliest near-term outcome for a working advertiser is a change in how auction pricing gets disclosed, not money arriving.

Budget as though nothing here will affect your account this year, because it very likely will not.

The mechanism, in the FTC's words

The alleged instrument has a name. According to the complaint, "beginning in 2019, Amazon changed its auction rules without notice by adding an undisclosed surcharge that Amazon referred to internally as a 'soft reserve price.'" One internal document quoted by the agency says the auction pricing had "a surcharge hidden in it."

A UGC creator filming a short product video at home with a phone on a small tripod
Novoads · UGC video ads with AI, ready in minutes.
Try now

How a second-price auction becomes a first-price one

To see why the alleged change matters, you need one paragraph of auction mechanics. It is genuinely simple, and it is the whole case.

What advertisers were told they were buying

Amazon, the FTC says, "has represented to prospective advertisers for years that Amazon runs 'second price' auctions where the winner of the auction would only pay 'one cent more than the next highest bidder'." That format is the long-standing default in search advertising, and it exists for a reason: the winner pays just enough to beat the runner-up, not the number they typed.

That carries two promises a buyer relies on without thinking about them:

  • Your maximum bid is a ceiling, not a price. You are quoting the most you would pay, not the amount you expect to be charged.
  • Honesty is the optimal strategy. Because you never pay your own number, there is no reason to shade it down, which is why the format became the search-advertising standard in the first place.

Take either promise away quietly and every bid in the account starts behaving differently than the person who set it intended.

What the complaint says changed

Per the FTC, the surcharge was computed after the auction had already selected a winner, which raised the winner's price above the level the auction produced. The agency's conclusion is blunt: Amazon "has charged its Sponsored Products advertisers their own winning bid close to 80% of the time and has effectively converted its nominally 'second price' auction into a first price auction."

The number that tracks the shift

The complaint puts a trend line under that claim. It alleges the share of the time Sponsored Products advertisers paid their own full bid moved like this:

YearAdvertisers paying their own bid
202130% to 40%
202270%
2024about 80%

Read that as a shape rather than a scandal. Even if a court eventually finds the pricing was lawful and adequately described, an account whose bids clear at their own maximum four times in five is a materially different account from one that clears at second price. The bid you typed stopped meaning what it used to mean, and nothing in the interface announced it.

Amazon's answer, in its own words

A one-sided reading of a complaint is a bad way to learn anything, and Amazon published a detailed response the same day. It is worth reading in full, because the areas of agreement are more interesting than the areas of dispute.

What Amazon disputes

Amazon's opening line is that "The FTC today filed a misguided lawsuit claiming Amazon misled advertisers about its Sponsored Ads pricing and auction." Its rebuttal rests on three distinct arguments, and they are worth separating because they do not stand or fall together:

  • The model is wrong about behavior. Amazon says the FTC's theory assumes advertisers do not adjust their bids in response to results, which it calls a flawed premise. Anyone who has managed a search account will recognize the objection.
  • The outcome went the other way. Working from the FTC's own premise anyway, Amazon says "we estimate they saved over $8 billion from 2021 to 2025 as a result of Amazon prioritizing ad relevancy over selecting ads on bid price alone."
  • The price did not actually rise. Amazon states that "From 2019 through 2024, the average cost-per-click for Amazon's Sponsored Products search ads remained flat adjusted for inflation, while conversion rates grew 24% from 2021 to 2025."

What Amazon does not dispute

Here is the sentence that makes this story useful rather than merely loud. Amazon does not deny the feature exists. In its own words, "we began to test a concept called 'soft reserve prices,' a real-time minimum value that seeks to better reflect what each placement is actually worth."

So both sides describe the same machinery. The FTC calls it an undisclosed surcharge bolted onto a stated second price auction. Amazon calls it an estimate of a placement's true market value. Strip the adjectives and the dispute narrows to two lines:

  • Not in dispute: a reserve price mechanism exists, and it sits between the auction result and the price an advertiser pays.
  • In dispute: whether it was adequately disclosed, and whether it left advertisers worse off.

A court will decide which characterization the law prefers. Reasonable people can hold different views on that while agreeing on the fact that matters to a buyer: a pricing layer existed, and it was not in the interface.

A grid of vertical ad variations from one product, arranged for testing
Novoads · UGC video ads with AI, ready in minutes.
Try now

Why a change like this would not show up in a dashboard

This is the part that generalizes, and the part I would put in front of anyone who buys media on any platform.

You only ever see the output

Amazon's response describes a genuinely rich reporting suite. It says advertisers "get real-time reporting on actual cost-per-click, return on ad spend, click volume, purchase data, and sales attribution." All of that is true and none of it answers the question at issue.

Every metric in that list is downstream of the price. They tell you what you were charged and what it produced. They do not tell you how the charge was computed, because the clearing logic is not a reported field on any self-serve platform. A cost-per-click of $1.40 looks exactly the same whether it came from beating a rival by a cent or from a floor applied after the fact.

A clearing price arrives without a receipt

The complaint makes this explicit. In paragraph 51 it alleges that Amazon controlled and limited the auction data available to customers, internally described its own aggregated price reporting as opaque, and that as a result customers had no way to know whether the price they were charged was the minimum needed to beat the second-place bid or an amount set by Amazon. An internal document quoted later in the filing calls the auctions a black box to advertisers, on the grounds that advertisers only see the amount they get charged.

The FTC also quotes an Amazon employee saying "the company's hidden surcharges are 'good for Amazon' because 'advertisers must pay more for the same advertising,' and 'the benefit to Amazon comes at the cost of advertisers.'" That is the agency's evidence of intent. It is also, incidentally, a clean description of why the reporting gap is worth caring about.

The asymmetry is not an Amazon feature

Nothing in this case says anything about Meta, TikTok, Google or anyone else, and it would be irresponsible to imply otherwise. But the structure is not special. On every auction-based ad platform, the same company:

  • Owns the inventory being sold.
  • Writes and changes the auction rules.
  • Computes the price you are charged.
  • Defines and publishes the metrics you judge it by.

You are a participant in a market whose books you cannot open. It is the same asymmetry behind a high click-through rate that looks like success and means very little, and behind the workarounds buyers build for problems they cannot inspect directly, like cutting invalid clicks without waiting for the platform to act.

Split what a platform report can and cannot answer, and the practical boundary is clear.

What the report answers well:

  • What you spent, at what unit price, over any window you choose.
  • What that spend produced in clicks, orders and attributed revenue.
  • Whether the ratio between the two moved, and roughly when.

What no self-serve report can answer:

  • Whether the price you paid is the price the auction produced.
  • Whether a pricing rule changed, when it changed, or who it changed for.
  • Whether last quarter's efficiency drop was competition, seasonality or a decision made on the other side of the glass.

That last gap is what makes this case hard for a buyer to feel confident about in either direction. If the allegations are false, no advertiser could have proven it. If they are true, no advertiser could have caught it.

What you still control when you cannot audit the auction

The wrong response to this story is to stop buying ads. The right one is to stop building plans on the assumption that platform-side efficiency is a stable input, and to move weight onto inputs that are yours.

Keep the ledger outside the platform

If your only record of performance lives in the platform's reporting, a pricing change and a market change look identical. Keep spend, orders and contribution margin in your own system, reconciled monthly. The point is not distrust for its own sake. It is that a number computed by the counterparty cannot be the number you use to detect a problem with the counterparty. That habit belongs in your creative operations routine alongside naming conventions and asset versioning.

Judge on outcomes, not on platform efficiency alone

Cost-per-click and reported ROAS are the two metrics most exposed to a clearing-price change, because both have the price in the denominator or the numerator. Orders, revenue and margin do not. Building your creative analytics around outcomes you can count independently means an unexplained shift shows up as a real result, not as an unfalsifiable efficiency story.

MetricMoves if the clearing price moves
Cost-per-clickDirectly
Reported ROASYes, through spend
OrdersNo
Contribution marginOnly via spend
Cost to produce a creativeNo
An ecommerce marketer reviewing finished vertical video ads on a laptop
Novoads · UGC video ads with AI, ready in minutes.
Try now

Volume of creative is the lever that does not move

Here is the arithmetic that survives any auction change. Take a short AI clip at roughly $2 of credits and price the thing you actually want, which is one winning ad:

  • One winner in ten tries: about $20 of production to find it.
  • One in five: $10.
  • One in twenty: $40.

Now change the auction under all three cases. Your unit price rises, your break-even ROAS rises with it, and every one of those production numbers stays exactly the same. Creative cost is the one line in a media plan the platform does not price. That is precisely why how many creatives you need is a more useful planning question than what your CPC will be next quarter, and why structured creative testing beats bid tinkering when the unit price gets worse.

A better ad lowers your effective cost of a customer by winning more auctions at the same bid. That works no matter what the reserve is doing.

The short version, if you want something to act on this week:

  1. Export your own spend and revenue for the last twelve months and keep it somewhere the platform does not control.
  2. Pick one outcome metric you can count independently, usually orders or contribution margin, and make that the number the team argues about.
  3. Set a creative floor, a minimum count of genuinely different concepts live per audience, and treat dropping below it as an incident rather than a slow quarter.
  4. Re-read your bidding assumptions once a year. A stated maximum bid is a ceiling only for as long as the platform says it is.

How Novoads helps when the auction is the part you cannot fix

You cannot audit a clearing price. You can decide how many different ads meet it. Novoads turns a product photo and a script into UGC-style video ads with AI actors, so the number of angles you can put in front of an audience stops being limited by a shoot calendar. Short clips start at around $2 of credits, which is what makes testing ten variants a normal week rather than a budget request. If you sell on Amazon specifically, the same pipeline covers product video for listings and Sponsored Brands.

You can try it for $1 for 3 days, which then continues at $49 a month, and cancel whenever you want.

The auction is rented. The creative is yours.

Nothing in this case is settled. Amazon may well persuade a court that a soft reserve price is an honest estimate of what a placement is worth, and the FTC may not prove a dollar of the twenty billion it has pleaded. Treat every number here as an allegation, because that is all it is today.

What is already settled is the shape of the relationship. You rent the auction, the targeting and the reporting from a company that also sells the inventory, and you own exactly one input in the whole system. Spend your attention accordingly: the platform decides what a click costs, and you decide what shows up when someone sees it.

Frequently Asked Questions

What is the FTC's lawsuit against Amazon about?

On August 31, 2026 the Federal Trade Commission and 22 state attorneys general filed suit against Amazon in the U.S. District Court for the Western District of Washington, case number 2:26-cv-03097. The complaint alleges that Amazon concealed unfair charges in its digital advertising auction pricing, specifically that beginning in 2019 it changed its auction rules without notice by adding an undisclosed surcharge it referred to internally as a soft reserve price. The allegations are untested, Amazon denies the merits, and nothing has been decided by any court.

Where does the $20 billion figure actually come from?

From the complaint, not the press release, and the difference matters if you are citing it. The FTC's public press release says the alleged scheme has likely extracted tens of billions of dollars from its unwitting advertising customers. The specific number lives in paragraph 1 of the filed complaint, which pleads that the scheme has likely illegally extracted over 20 billion dollars from approximately 1.2 million advertising customers since 2019. It is an FTC estimate in active litigation, and the average surcharge rates that support it are redacted in the public version of the filing.

What is a soft reserve price?

In the FTC's telling it is a price floor applied after the auction has already picked a winner, which pushes the winner's price above what the auction itself would have set. Amazon describes the same feature differently. In its public response it says it began to test a concept called soft reserve prices, a real-time minimum value that seeks to better reflect what each placement is actually worth. Both sides agree the mechanism exists. What is disputed is whether it was adequately disclosed and whether it harmed advertisers.

Does this affect me if I only buy ads on Meta or TikTok?

Not legally. The case is about Amazon's Sponsored Products, Sponsored Brands and Sponsored Display auctions, and it makes no allegation about any other platform. Structurally, though, it is a useful reminder: every self-serve ad platform reports what you were charged, and none of them expose the clearing logic that produced that number. That asymmetry is the same everywhere, which is why your plan should rest on inputs you control rather than on efficiency metrics the seller computes.

How long does Amazon's alleged conduct go back?

The complaint dates the overcharging to 2019 and the FTC frames the span as over seven years. Be careful with secondary coverage here. Some outlets compressed it to a shorter window, and Amazon's own defense discusses a 2021 to 2025 period for its savings estimate, which is a different window than the one the FTC pleads.

What should an advertiser actually do about this?

Nothing panicked, and nothing that assumes an outcome. Keep your own record of spend and revenue outside the platform so a pricing change shows up in your numbers rather than only in theirs. Judge campaigns on outcomes you can verify, such as orders and contribution margin, rather than on platform-computed efficiency alone. And keep enough creative in rotation that a rising unit price meets a better ad rather than a bigger bid, because creative is the input that does not move when the auction does.

Key Takeaways

  • On August 31, 2026 the FTC and 22 state attorneys general sued Amazon in the Western District of Washington, case 2:26-cv-03097, alleging it concealed unfair charges in its digital advertising auction pricing.
  • Read the right document for the right number: the FTC press release says only tens of billions of dollars, while the filed complaint is where the specific over 20 billion dollars figure appears, in paragraph 1.
  • The alleged mechanism is an undisclosed surcharge added in 2019 that Amazon called a soft reserve price internally, which the FTC says turned a nominal second price auction into an effectively first price one.
  • Amazon denies the merits, calls the suit misguided, does not dispute that soft reserve prices exist, and says advertisers saved over $8 billion from 2021 to 2025.
  • Nothing here is proven. The durable point for buyers is structural: platform reports tell you what you paid, never how that number was produced, so build your plan on the levers you own instead.
Mauricio Valdivia

Mauricio Valdivia

Founder of Novoads

Mauricio is the founder of Novoads, where he works to democratize video advertising with AI for brands in Latin America.