On August 31, 2026, the FTC and 22 state attorneys general sued Amazon, alleging it hid ad-auction surcharges from advertisers for more than seven years. If you manage paid media on any platform — not just Amazon — this case is worth reading closely, because the allegation isn't really about Amazon. It's about what happens when the same company sells you an ad auction and then tells you how that auction went. Below is what the case actually alleges, why the same structural problem exists everywhere you spend, and a concrete habit for checking platform-reported numbers against something you control instead of taking them at face value.

The Lawsuit That Just Proved What Marketers Suspected

According to the FTC's own press release, Amazon told advertisers it ran standard "second price" auctions, where the winner pays roughly one cent more than the next-highest bid. The FTC alleges that starting in 2019, Amazon added an undisclosed surcharge on top of that — and that the gap between what a second-price auction should have charged and what Amazon actually charged grew over time: from surcharges affecting 30-40% of bids in 2021, to about 70% in 2022, to roughly 80% by 2024, at which point most winning advertisers were effectively paying close to their own full bid rather than the discounted second-price amount they believed they'd agreed to. The FTC puts the affected group at over 1 million brands and sellers, including more than 500,000 small and medium-sized businesses, with "tens of billions of dollars" extracted in total. Separately, Marketing Dive's reporting on the complaint describes internal Amazon documents as acknowledging the practice risked "irrevocable damage to advertiser trust," and states the surcharge was ramped more aggressively ahead of high-volume events like Prime Day and Black Friday — that specific timing detail comes from Marketing Dive's account of the complaint rather than the FTC's press release directly, so it's flagged here as secondary reporting rather than confirmed straight from the primary document.

Amazon disputes the framing. Its stated defense is that inflation-adjusted cost-per-click stayed flat between 2019 and 2024 while conversion rates rose 24% — in other words, advertisers got more for their money even if the auction mechanics worked differently than they assumed. The case is contested and unresolved as of this writing.

None of that changes the part that's already useful to you: a platform's own report and an independently verified number are not automatically the same thing, and right now there's a live federal case built on exactly that gap.

Why Every Platform Has the Same Incentive Problem

The Amazon case is one instance of a structural pattern, not a one-off. Any platform that both sells you an ad auction and reports how that auction performed is grading its own work. That doesn't make every platform dishonest — but it does mean the incentive to shade a number in its own favor exists everywhere, not just where a lawsuit happens to be filed.

Gartner analyst Eric Schmitt, quoted in Marketing Dive, points to a specific version of this showing up in automated bidding tools right now: platform-generated recommendations that skew toward telling you to spend more, regardless of whether that's actually the efficient move. Gartner projects roughly 80% of US ad spend will flow through AI-driven platforms by 2028, and Google, Meta, and Amazon already account for more than half of US paid media spend between them, a share that's trending toward 60% globally. Schmitt's recommendation is blunt: don't give an ad platform unfettered access to your budget without a human checking its work, and don't rely solely on platform-generated performance reports to judge that work.

The practical takeaway isn't to distrust every number a platform shows you. It's to stop assuming a platform's report is neutral by default — it's coming from the same entity that benefits when you spend more, so it needs the same scrutiny you'd give a vendor's own case study.

What the "Black Box" Actually Looks Like Day to Day

For anyone running Performance Max, Advantage+, or a comparable auto-bidding system, this isn't an abstract concern. One performance marketer described it plainly: with these systems, you don't get to ask where your money went or which audiences it touched — you feed in budget and creative and see outcomes, with no visibility into the allocation in between. An agency Google Ads strategist described the same experience from the buyer's side: the system amounts to handing over your budget and trusting it to spend "wherever it wants," with no way to see detailed reports or control where the money goes. Another practitioner put the underlying economics more bluntly, calling auto-bidding "the most expensive way to manage campaigns and the best way to maximise Google's revenue" — a specific, if pointed, take rather than an industry consensus.

These are individual practitioner accounts, not a formal study, so treat them as descriptions of a common experience rather than a proven industry-wide statistic. But if you've ever gotten an auto-bid recommendation to raise your budget with no explanation of why, or watched a campaign's reported ROAS shift after a platform quietly changed its measurement window, this is what that experience looks like from the inside — and it's the same underlying problem the Amazon case surfaces, just without a lawsuit attached.

Building an Independent Ad-Spend Audit Habit

None of this means walking away from automated bidding. It means treating a platform's report as a claim to verify, not a fact to relay. Here's a habit that does that without requiring an enterprise measurement budget.

Set a cadence, and actually keep it. A useful baseline, drawn from paid-media budget-review guidance, is a monthly review for standard operations — enough to recalculate blended performance and catch drift before it compounds. Campaigns spending more than roughly $50K a month can justify a lighter weekly check limited to spend pacing and tracking integrity, without turning that into a full monthly-style re-audit every week. Quarterly is for the board-level strategic view, not the working audit.

Pull data at the campaign level, not the account-summary level. Dashboard summaries are where a lot of the ambiguity in "what actually happened" gets smoothed over. Working from campaign-level exports gives you something you can actually reconcile line by line.

Reconcile against a number you control — not the platform's own claim. One reconciliation approach matches platform-claimed conversions against your own CRM, order system, or finance data rather than accepting a platform's attributed total. That specific method is written for ecommerce contexts, but the underlying principle generalizes: whatever revenue or pipeline system you already trust internally is a better baseline than any single ad platform's own dashboard, because it's the one number in the chain that platform isn't grading itself.

Watch specifically for gradual cost creep around your highest-volume periods. This is the direct, practical lesson from the Amazon case: the alleged surcharge increases were reportedly ramped ahead of Prime Day and Black Friday precisely because a gradual change is harder to isolate from normal seasonal cost movement. If your cost-per-result quietly climbs heading into your biggest campaigns of the year, that's exactly the pattern worth a closer look, not just a shrug at "seasonality."

When something doesn't reconcile, don't try to resolve it with more platform data. Schmitt's advice — bring in independent, external measurement rather than leaning further on the platform's own reporting — applies here directly. If your internal numbers and a platform's claimed numbers disagree after a proper campaign-level reconciliation, that's a signal to verify externally, not to keep re-pulling the same dashboard.

This is a starting framework, not a finished audit program — treat the cadence and thresholds above as a baseline to adapt, and set your own targets based on your own historical data rather than importing someone else's benchmark wholesale.

When the Automation Is Actually Working

Building an audit habit isn't the same as assuming automation is always working against you. One agency's account is a useful counterpoint here: after resisting Performance Max for years over exactly the visibility concerns above, the agency eventually saw it outperform their old manually-managed Local Search Ads campaigns — but only after they switched their optimization target from raw conversion count to conversion value, effectively telling the system that a phone call was worth more than a directions request. That's one agency's self-reported experience on one account type, not a general finding you should expect to reproduce automatically. But it's a useful reminder that the fix for a black box isn't necessarily to abandon it — it's often to check what you're optimizing it for, verify the result independently, and go from there.

FAQ

Can you trust Google Ads reported spend?
Treat it as a starting number, not a final one. Google Ads reporting reflects Google's own measurement methodology and attribution windows, which can change without materially changing your underlying results. The habit above — reconciling campaign-level data against your own CRM or revenue system on a regular cadence — is how you check whether what the dashboard shows matches what actually happened.

What is the Amazon FTC ad surcharge case about?
The FTC and 22 state attorneys general allege that starting in 2019, Amazon added an undisclosed surcharge on top of what it told advertisers were standard second-price auctions — growing from affecting roughly 30-40% of bids in 2021 to about 80% by 2024, at which point most winning advertisers were effectively paying close to their own full bid rather than the discounted second-price amount they believed they'd agreed to. The FTC puts the affected group at over 1 million brands and sellers, with "tens of billions of dollars" extracted in total. Amazon disputes the FTC's characterization and says advertiser cost-per-click stayed flat, adjusted for inflation, over the period in question. The case is contested and unresolved as of this writing.

How often should I audit ad spend across platforms?
Monthly is a reasonable baseline for most advertisers, based on paid-media review guidance and cross-platform reconciliation practices. Higher-spend campaigns (roughly $50K/month or more) can add a lighter weekly check focused on pacing and tracking integrity, while a full strategic review is better suited to a quarterly cycle.