Amazon published a detailed response on August 31, 2026, the day the FTC and 22 states filed their complaint. The points below are Amazon’s assertions and estimates, not court findings or independent audit results. Amazon
Relevance, not the largest raw-dollar bid
Amazon says Sponsored Ads winners are selected heavily on expected relevance to the shopper. It reports that the largest nominal bid did not win in approximately 92% of placements and that the winner’s bid ranked around 128th by dollar amount on average. Amazon uses those figures to argue that a simple comparison between the two largest dollar bids misdescribes its auction.
Relevance-adjusted ranking could complicate a simplified second-price illustration. It does not, by itself, resolve the separate government allegation that Amazon added an internally selected price after calculating a GSP CPC.
A bid is a maximum CPC
Amazon says Campaign Builder informed advertisers that a bid was the maximum amount they could pay for a click. It argues advertisers therefore understood the price could reach their bid. The complaint does not allege charges above those maximums.
The government’s competing theory is that a maximum-price disclosure did not reveal the mechanism used to select the actual charge. No court has decided whether Amazon’s disclosures were adequate.
Reserve pricing is common
Amazon rejects the FTC’s “invented participant” or shill-bid framing. It describes reserve pricing as an ordinary auction practice used to reflect inventory value. The central dispute is thus not whether auctions can ever have reserves, but how the challenged reserve operated and whether advertisers received material information about it.
Amazon’s economic counterarguments
| Amazon’s assertion | Period | What it is offered to show |
|---|---|---|
| Inflation-adjusted average Sponsored Products search CPC was approximately flat | 2019–2024 | Amazon disputes an inference of systematic advertiser harm. |
| Average winning Sponsored Products bids fell about 50% | 2019–2025 | Advertiser bidding behavior changed rather than remaining fixed. |
| Sponsored Products conversion rates improved approximately 24% | 2021–2025 | Advertisers received increased performance value. |
| Amazon’s relevance-oriented system saved advertisers more than $8 billion under its analysis | 2021–2025 | The FTC’s preferred counterfactual would have left advertisers worse off, Amazon says. |
Those figures do not necessarily measure the same counterfactual as the alleged difference between a GSP CPC and a final CPC. Auction-level evidence and damages methodology remain unresolved.
How Amazon describes advertiser behavior
Amazon says sophisticated advertisers adjust bids and budgets using sales, conversion, return on ad spend, and other observed performance—not a fixed “true value” derived from auction theory. That argument may bear on whether different disclosures would have changed bids and on any damages model.
Training materials and internal documents
Amazon says the FTC gave undue weight to outdated or low-reach training content. It reports small enrollment, completion, or viewing figures for several cited materials and says it corrected or removed outdated content when identified.
Amazon also says investigators received approximately 1.5 million pages over six years and that the complaint cherry-picks a small number of informal discussions and brainstorming comments. The government, in turn, cites communications, webpages, training, sales materials, and advertiser presentations as evidence of both expectations and intent. U.S. District Court filing via FTC
What the court has ruled
Nothing on the merits. No court has accepted Amazon’s defenses or the government’s allegations, calculated advertiser loss, or ruled on liability.