Amazon submitted its response on August 31, 2026, to the lawsuit filed by the U.S. Federal Trade Commission (FTC) concerning Sponsored Ads, rejecting the basis of allegations related to harming consumers or advertisers. The company says the complaint provides no evidence that product prices increased because of advertising, and that it focuses on a limited number of materials and simplified communications to portray a misleading institutional practice, after reviewing approximately 1.5 million pages of documents and messages over six years.
This article represents Amazon’s position in a legal dispute and is not an independent judgment on the validity of either party’s claims. Its argument centers on the idea that the advertising system does not select the winner based solely on the highest monetary bid, but balances the value of the bid with the ad’s relevance to the user’s search.
How Does Amazon Describe the Sponsored Products System?
The company says its ads began in 2006 through generalized second-price auctions, in which the advertiser sets the maximum amount they are willing to pay per click and may pay less than that amount. As Amazon’s store evolved and expanded to hundreds of millions of products across more than 35 categories, the company began testing machine-learning models in 2014 that predicted the likelihood that an ad would benefit a shopper, and deployed these models across store ads by 2019.
According to Amazon’s explanation, the algorithm initially narrows the list of eligible ads from hundreds of thousands of potential ads to fewer than 1,000 relevant ads, then assigns each ad a score based on relevance and the monetary bid. The company uses a “hard” reserve price representing the minimum for entering the auction, and a “soft” reserve price intended to reflect the estimated value of the ad placement. If the winning bid exceeds both prices, the advertiser pays the soft reserve price; if it exceeds the hard threshold but does not reach the soft one, the advertiser may pay its bid amount. Amazon emphasizes that in no case does the advertiser pay more than the maximum amount it set.
The Figures on Which the Company Relies
Amazon claims that the average cost per click for Sponsored Products ads, adjusted for inflation, remained stable between 2019 and 2024, while conversion rates increased by more than 24% for individual advertisers between 2021 and 2025. It also says that average winning bids fell by 50% between 2019 and 2025, and that approximately 92% of ads selected in 2024 were not the highest-priced ads; instead, the winning bid was on average close to 128th place by monetary value.
The company estimates that advertisers saved more than $8 billion between 2021 and 2025 as a result of giving relevance greater weight than the bid alone. Amazon also expects sales in 2026 to be at least 58% higher, with a 46% better return on ad spend, compared with a model based solely on the highest monetary bid. These figures remain estimates and claims presented by the company in its response, and are not independent findings presented in the article.
The Issue of Explanation and Transparency
Amazon says it updated its Sponsored Ads materials in 2018 to clarify that emphasizing relevance could result in charging the advertiser an amount up to its maximum bid, and that the Ad Console Campaign Builder has stated since that year that the bid represents the maximum amount that may be paid. The company acknowledges that some training segments and educational materials retained older or simplified examples of auctions, but describes them as having limited reach and says it later removed or updated them.
Amazon says that three training courses cited in the FTC’s filing recorded a total of 1,849 registrations and 779 completions, while one of the videos had no more than 928 viewers over two and a half years—approximately 0.09% of Sponsored Ads advertisers in the United States, according to the company’s estimate.
What Changes in Practice?
The technical significance of this dispute is not limited to the question of whether an advertiser pays more than its bid; it also concerns how understandable a system is when it combines machine learning, bids, and reserve prices determined in real time. Amazon says advertisers and automated bidding tools make their decisions based on actual outcomes such as cost per click, sales, and return rate, rather than on the theoretical description of the auction structure.
The company adds that 80% of bid changes in active campaigns between 2019 and 2024 occurred within one day of a previous change in Sponsored Products ads that received clicks. It also cites an experiment in which competitors’ bids on specific keywords fell by approximately 83%, from $15.96 to $2.74, after an advertiser stopped bidding on them.
In practice, this response leaves two questions open: Were the disclosures provided to advertisers clear and sufficient as the system shifted to more complex models and reserve prices? And are the performance data presented by Amazon enough to establish the absence of harm, or does the court need an independent test of the relationship between auction design, prices, and competition? What is clear from the article is that the company updated its help instructions and began periodic reviews and internal training, but it does not resolve the outcome of the lawsuit or the FTC’s assessment of the evidence.