The FTC's Fake Reviews Rule and Affiliate Sites: What It Bans, Why 'We Tested' on a Guide You Didn't Test Is a Legal Problem, and the Wording That's Safe
The FTC's rule on consumer reviews and testimonials has been in force since 21 October 2024, and most of the coverage was about sellers buying five-star reviews. Affiliate sites read it as someone else's problem, which is half right. The rule doesn't touch an honest comparison of products written from data.
It does touch the things sites add to make a comparison feel first-hand when it isn't: the quoted "owner" who doesn't exist, the reviewer persona with an invented background, the testimonials box, the "we tested all twelve" line on a guide nobody tested. Those were always deceptive. Now the FTC can fine them per instance without proving harm first.
This post covers what the rule prohibits in its own terms and what the penalties are, where an affiliate site is actually exposed (five things, all common on AI-assisted sites), where the method claim sits (older law the rule makes more dangerous), what the rule leaves alone, and the wording for a method section, an author line and a quoted owner comment that keeps a data-driven guide on the right side.
The reviewing without the product post covers the honest method; this is the legal reason it matters.
The short answer
Don't publish anything that reads as a consumer's review or testimonial unless a real consumer wrote it and you can show where. No invented owner quotes, no persona reviewers, no generated testimonials, no "readers rate this" that nobody rated.
Don't claim a method you didn't use. Compare products from specs, the review pattern and price, say that's what you did, attribute quoted owner comments to Amazon reviews by pattern not by invented name, and put a real author on the page.
What the rule prohibits
| Prohibited | What it means | Where an affiliate site meets it |
|---|---|---|
| Fake or false consumer reviews and testimonials | Reviews by people who don't exist (AI-generated ones included), by people who never used the product, or that misrepresent the writer's experience | Invented owner quotes; "reader reviews" written by the site; a generated testimonials box |
| Buying reviews with a required sentiment | Paying or incentivising for positive or negative reviews | Rare on affiliate sites; relevant if you solicit reader reviews with a reward |
| Insider reviews without disclosure | Owners, employees or relatives reviewing without a clear connection disclosure | A site owner reviewing their own product line; a brand partner's "review" without the relationship stated |
| Company-controlled review sites posing as independent | A review site owned by the seller presented as neutral | A brand's affiliate site reviewing only that brand without saying it's the brand's |
| Review suppression | Threats to stop a review, or showing only the positive ones while implying all are shown | A "reader reviews" widget that hides the negative submissions |
| Fake social media indicators | Buying bot followers, likes, views | Buying followers for the site's social accounts |
The penalty for a knowing violation is up to $51,744 per violation, and the rule doesn't pin down what a single violation is, so a page of twenty invented quotes is not safely one. The rule also authorises consumer redress. What was previously a case the FTC had to build under general deception law is now a rule with a number on it.
Where an affiliate site is exposed
- Invented owner quotes. "One owner told us the mount held up for two winters." Nobody told anyone; the generator wrote a plausible owner. That's a fabricated consumer review attributed to a person who doesn't exist, which is the first row of the table.
- The reviewer persona. "Reviewed by Sarah, a cycling coach with 15 years' experience" above a page nobody named Sarah wrote. A testimonial-style endorsement by a fictitious person, and a misrepresentation of the author's experience.
- The testimonials box. "What our readers say" with three glowing lines nobody submitted. Generated testimonials, by definition fake.
- The fake rating. A "reader score: 4.7" or a star widget with no readers behind it. A fabricated review signal.
- The method claim. "We tested all twelve for three weeks" on a guide written from listings. Not a consumer review, but a material misrepresentation under the FTC Act and the Endorsement Guides, and the first thing an investigator reads after the invented quotes.
All five are what a generator produces when the brief says "make it feel first-hand" or when the writer wants the page to look like a competitor's. Faking experience fails with Google's quality systems too; the FTC has fewer updates and larger fines.
What it leaves alone
- Editorial comparison. A guide that ranks ten products on specs, the review pattern and price, written by the site, is the site's opinion, not a consumer review. The rule isn't aimed at it.
- Reporting the review pattern. "Reviewers commonly mention the mount loosening" is a summary of real reviews on Amazon, attributed to where they are. It isn't a fabricated review as long as it's true and not dressed as a named individual.
- AI assistance in writing. The rule covers what the text claims to be, not how it was drafted. An AI-assisted guide that's honest about its basis is editorial content.
- Quoting real reviews with attribution, within fair-use limits and Amazon's rules on reproducing its content; a paraphrased pattern is safer than a verbatim quote for both reasons.
- Affiliate links with disclosure. The Endorsement Guides already required the disclosure; the legal pages post covers it. The new rule adds nothing there.
The wording that's safe
- Method section: "We compared these products on their specifications, the patterns in owner reviews on Amazon, and current price. We did not test them ourselves." The how we picked post has the fuller template; this is the sentence that matters legally.
- Owner comments: "Among Amazon reviewers, the most common complaint is..." Not "one owner, Mark, told us...". Pattern, attributed to its source; never a named individual you invented.
- Author line: a real person who edited the page, with a real bio. If the site is one person using tools, say that. No persona.
- Reader input, if you collect it: show what was submitted, including the negative, or don't show it. Say how it was collected.
- Disclosure of connections: if you review your own products, a partner's, or a brand that paid you, say so above the review in plain words.
- Nothing styled as a review that isn't one: no star widgets without raters, no quote boxes without sources, no "verified buyer" badges you can't verify.
Doing it with AffSERP
Guides written by AffSERP's X Best Review tool report the owner review pattern from Amazon's data as a pattern ("reviewers commonly mention"), never as an invented individual, and the method section states the basis as specifications, review patterns and price without a testing claim, which is the wording above by default.
The tool doesn't generate testimonials, reader ratings or reviewer personas, and the author on the page is whoever you set in WordPress. That leaves the five exposures as things a site would have to add by hand, and the shortest compliance advice is not to.
FAQ
What does the FTC fake reviews rule prohibit?
The Trade Regulation Rule on the Use of Consumer Reviews and Testimonials, in force since 21 October 2024, bans creating, buying or publishing consumer reviews or testimonials that are fake or false: written by people who don't exist (including AI-generated ones presented as real), by people who haven't used the product, or that misrepresent the writer's experience. It also bans paying for reviews with a required sentiment, insider reviews without a clear disclosure of the connection, company-run review sites posing as independent, suppressing negative reviews by threat or by burying them, and buying fake social media indicators. Knowing violations carry civil penalties of up to $51,744 each.
Does the FTC rule apply to affiliate marketing websites?
Where the site publishes consumer reviews or testimonials, yes. An affiliate site's own editorial comparison isn't a consumer review, so a guide that compares products on specs and the review pattern isn't the rule's target. The exposure is in what sites add to look more convincing: quoted 'owners' or 'readers' who don't exist, a fake persona as the reviewer, testimonials generated to order, a 'user rating' nobody submitted, or a 'what our readers say' box written by the writer. Each is a fake consumer review or testimonial in the rule's terms, and each is the kind of thing a generator produces when asked to make a page feel first-hand.
Is saying 'we tested' when we didn't a violation?
It's deception under the FTC Act and the Endorsement Guides whether or not the new rule names it, and the rule makes the environment stricter. A false claim about how a recommendation was reached is a material misrepresentation to the reader, and the FTC has pursued review sites for exactly that. The rule itself bites hardest on fabricated consumer reviews and testimonials; the method claim is the older, broader problem the rule sits on top of. A site that compared products from data and says so is in the clear on both; a site that invents a testing programme is exposed on both.
Can I use AI to write product reviews under the FTC rule?
Yes, for editorial content that's honest about what it is. The rule doesn't ban AI-written text; it bans fake consumer reviews and testimonials, which includes AI-generated ones that pretend to be from real customers. An AI-assisted buyer guide that compares products on specifications, owner review patterns and price, says so in its method section, carries a real author and attributes any quoted owner comments to their source, is editorial content, not a fake review. The same tool asked to produce 'reviews from happy customers' for a page produces the thing the rule bans.
AffSERP reports owner reviews as patterns from Amazon's data, states the method as specs, reviews and price, and generates no testimonials, ratings or personas.
Start with free words