You’re looking at a joint supplement. It shows 4.7 stars across 3,200 reviews. The top one runs 400 words, names a specific knee, and ends with a photo from a hiking trail.
That review may be completely sincere. As evidence that the product works, it’s close to empty.
Since October 2024, a federal rule has made a lot of review manipulation directly punishable. It helps. It does not fix the deeper problem, which is that a pile of personal stories is a weak way to learn whether a health product does anything at all.
This is the number that reframes the category. A 2022 study in the Proceedings of the National Academy of Sciences tracked the private social media groups where sellers shop for fake reviews. On a typical day the researchers identified about 23 active groups, each averaging roughly 16,000 members, with about 568 fake-review requests posted per day in each group.
What the federal rule bans
The Federal Trade Commission’s Rule on the Use of Consumer Reviews and Testimonials lives at 16 CFR Part 465. It took effect on October 21, 2024, and it authorizes courts to impose civil penalties for knowing violations. That detail carries weight. Civil penalties mean a company can face a dollar judgment for conduct the rule names, rather than only a cease-and-desist order.
The rule is short, and it is organized around a handful of things a business may not do.
First, a distinction that trips everyone up
The rule treats reviews and testimonials as two different objects, and the difference changes who is responsible for what.
A consumer review, as the FTC defines it, is an evaluation published on a site or platform dedicated at least in part to displaying such evaluations. The customer rating area on a retailer’s product page counts. So does a standalone review site. A rating with no text still counts.
A testimonial is an advertising message that consumers would likely believe reflects a real person’s opinions or experiences. The quote on the brand’s home page is a testimonial. So is the before-and-after clip in a video ad.
Most reviews are not testimonials and most testimonials are not reviews. The FTC’s guidance flags one group that is both: incentivized reviews. If a company gave someone a free bottle and that person posted a rating, the post is a review and an advertising message at the same time, and both sets of obligations attach.
One more classification matters for health content. The agency’s guidance treats a paid influencer’s post as a celebrity testimonial rather than a consumer review, which is why a sponsored video and a product-page rating get analyzed under different parts of the rulebook.
Writing, selling or buying fake reviews
Section 465.2 covers reviews and testimonials that misrepresent that the reviewer exists, that the reviewer used the product, or what the experience was. Writing or selling one is prohibited flat out, with no knowledge requirement attached. Buying one or spreading it around is prohibited when the business knew or should have known it was false.
Paying for a particular opinion
Section 465.4 bans compensation or incentives conditioned on a review expressing a particular sentiment, positive or negative. Read that carefully, because the common reading is wrong. Incentives themselves are not banned. A company may hand every customer a coupon for leaving a review. What it may not do is make the coupon depend on the review being favorable.
Insider reviews with no disclosure
Section 465.5 covers reviews and testimonials written by a company’s officers, managers, employees, agents and their immediate relatives. Those relationships have to be disclosed clearly and conspicuously. The rule’s own definition of that phrase is strict: the disclosure must be unavoidable, so tucking it behind a click or a hover does not count.
Pretending a site you own is independent
Section 465.6 prohibits a business from misrepresenting that an entity it controls supplies independent reviews or opinions about a product category that includes its own products. In its question-and-answer guidance on the rule, the FTC says this reaches past review websites. A company that invents a certification seal and awards it to its own products has the same problem wearing a different costume.
Suppressing reviews and faking influence
Section 465.7 covers review suppression. That includes unfounded legal threats, intimidation and knowingly false public accusations used to get a review taken down. It also covers claiming that the reviews on display represent most or all of what was submitted when negative ones were filtered out. Section 465.8 covers buying and selling fake followers and bot accounts.
Three things the rule does not do
This is where consumer coverage of the rule usually goes wrong, so it is worth being precise.
Ordinary consumers cannot be liable. The FTC’s guidance says this plainly. If you write a glowing review of a product you never opened, you are not violating this rule. Even a consumer who accepts cash for a five-star review is not liable under section 465.4. The rule aims at businesses, the people who sell fake reviews as a service, and the agencies and reputation-management firms working on a brand’s behalf.
The rule does not govern influencer disclosure. This one gets conflated constantly. Whether a creator has to tell you she was paid is a question for the FTC Act and the agency’s Endorsement Guides, which were revised in 2023. The reviews rule touches influencers in only one narrow spot: a person in the business of posting testimonials can be liable under 465.2(a) for lying about having used the product or about the experience. When and how a brand relationship gets disclosed is a separate body of law.
There is no private right of action. You cannot sue a company under this rule. Enforcement belongs to the FTC. Your realistic move as a consumer is to file a complaint with the agency, not to look for a lawyer.
One more boundary worth knowing. A retailer that merely hosts reviews it did not write or buy is generally not on the hook for fake ones, under the exception in 465.2(d). That shelter disappears when the retailer pulls a review into its own marketing, because a review featured in an ad is being disseminated rather than hosted.
There is also an older law in the background. The Consumer Review Fairness Act, which the FTC explains in guidance dated February 2017, makes it illegal for a company to put a term in its customer contract barring you from reviewing it, penalizing you for a review, or seizing the copyright in what you wrote. Companies can still remove reviews that are libelous, obscene, off-topic or clearly false. A review they simply disagree with does not qualify.
Why honest reviews still mislead
Suppose every review on a page is real, unpaid and written in good faith. The page is still not evidence.
Reviews are a volunteer sample, and volunteers cluster at the extremes. People write reviews when something delighted them or enraged them. The much larger group in the middle, the people for whom the product did roughly nothing, mostly says nothing. The result is a rating distribution with heavy ends and a thin middle, which no amount of honesty corrects.
There is no control group. A clinical trial exists to answer one question: what would have happened to these same people without the product? A review set cannot answer that, because nobody writing reviews was randomly assigned to a sugar pill. You are reading outcomes with no comparison, which is the exact condition under which humans are worst at judging cause.
Nobody reviews the thing they stopped taking. A person who bought a supplement, took it for nine days, felt nothing and dropped it rarely comes back to post about it. Dropouts are invisible in a review set, and they are usually the most informative group.
Regression to the mean, explained without the math
This is the single most underrated reason testimonials fool smart people.
Most symptoms that drive supplement purchases swing around. Joint pain, fatigue, bloating, sleep quality, headaches and low mood all move up and down on their own. You do not go shopping on an average day. You go shopping on a bad week, when the symptom is at its worst.
From a bad week, the most likely direction is improvement, with or without the capsule. Statisticians call that regression to the mean: an extreme measurement tends to be followed by a less extreme one. The purchase happens at the bottom of the swing. The improvement arrives a few days later. The review writes itself.
This is not a trick, and the reviewer is not lying. The timing simply guarantees a flattering coincidence for anything bought at a low point.
Placebo response, and why symptom products are the worst case
Placebo effects are real and measurable, and they are largest exactly where supplement marketing concentrates: subjective symptoms.
A 2021 systematic review in the journal Pain compared placebo treatments against no treatment at all in people with nonspecific low back pain. Across 19 studies of chronic back pain, placebo beat doing nothing for short-term pain intensity, with moderate-quality evidence. The authors put the size of the effect at roughly 8 points on a 0-to-100 pain scale and said plainly that it probably is not clinically meaningful. The effect also faded by medium-term follow-up.
Read that twice. Something inert, compared against nothing, produced a genuine short-term improvement that a person would absolutely feel and would absolutely write a review about. That improvement had nothing to do with any ingredient.
Now stack it on regression to the mean, on the expectation created by a $45 purchase, and on the fact that you took the product during a season when you were also sleeping differently and moving differently. A five-star review is the expected output of that system, whether or not the ingredient does anything.
Why “verified purchase” proves less than you think
Here is the common belief that needs correcting. A verified-purchase badge tells you money changed hands. It does not tell you the review was unpaid.
The 2022 PNAS study documented the standard mechanism directly. A seller posts a product in a private group. A reviewer buys it at full retail price, which triggers the verified badge. After the five-star review goes up, the seller refunds the purchase price through a payment app, often covering the transaction fee and sales tax and sometimes adding a commission on top. The purchase was real. The review was bought.
The researchers also noted that across their entire observation window, they saw sellers buying positive reviews only. Buying fake negative reviews about a competitor costs more and pays off less.
Their detection work points at something you can use. The strongest signal was not the wording of reviews or the photos. It was network structure: products buying fake reviews share reviewers with each other, because the pool of people selling reviews is small. In one large category of about 65,000 products and 11 million reviews, two clusters holding just 3.4% of products contained roughly 70% of the products their model flagged as fake-review buyers.
You cannot run that analysis at home. But it explains why a page of individually plausible reviews can still be rotten: the fraud lives in the pattern across products, not in any one paragraph.
How to read a review set instead of a review
Reviews are not worthless. They are just useful for a narrower set of questions than people think. Here is a working method.
Start by asking what a reviewer could possibly know. A customer can tell you the capsules are enormous, the bottle arrived crushed, the powder tastes like chalk, the subscription was hard to cancel, and the company never answered an email. Those are observable facts and reviews are genuinely good evidence for them. A customer cannot tell you whether an ingredient lowered inflammation. That requires a comparison they never ran.
Read the one-star reviews first, and look for repetition. A single angry review is noise. Eleven reviews describing the same specific failure, such as pills that arrive stuck together or a flavor change after a reformulation, is a signal. Specificity plus repetition beats volume.
Check the shape of the distribution, not the average. A product with an unusually thin middle, almost all five-star and one-star and little in between, is worth more skepticism than its average suggests.
Check recency against the product. Supplement formulas change. Reviews from three years ago may describe a different product with the same name on the bottle.
Treat any health outcome claim as unverified. When a review says a product fixed a diagnosed condition, that is the claim most likely to be wrong, whether or not it was paid for.
Notice what the seller does with negative reviews. The FTC’s guidance for review platforms, published in December 2021, says the basic obligation is simple: “Don’t display reviews in a misleading way.” A page where the company has replied to criticism without hostility is a better sign than a page with no criticism at all.
The FTC’s companion guidance for marketers, published in January 2022, tells businesses not to solicit reviews from “only customers you think will leave positive ones.” A seller who only emails its happiest buyers can produce a page of honest reviews that still misrepresents the product.
Where reviews cannot substitute for a clinician
Some review-driven decisions are low stakes. If a protein powder has a texture people hate, you are out $30.
Other situations call for a professional instead. Reviews are not the right input when you are trying to self-treat chest pain or pressure, shortness of breath, unexplained weight loss, blood in your stool or urine, a new lump, a severe headache unlike any you have had, or a symptom that has lasted more than a few weeks without improving. Those deserve an appointment, not a star rating.
Be especially careful with reviews that describe stopping a prescription. A review saying someone replaced a blood pressure medication or a thyroid medication with a supplement is describing a decision that can land a person in an emergency room, and it is being made by a stranger with no access to your labs.
Remember the regulatory baseline too. The National Institutes of Health Office of Dietary Supplements explains that supplements do not need FDA approval before they go on sale, and that the FTC is the agency that polices advertising claims. Nobody checked the product before it got its reviews.
The most useful takeaway
Change what you ask reviews to do.
Use them for logistics: shipping, packaging, pill size, taste, customer service, how hard cancellation was. Reviews are good at that, and no other source is better.
Do not use them for biology. For that question, a review set is the wrong instrument no matter how many stars it has, because it has no control group, no record of the people who quit, and no way to separate the ingredient from the week.
The new federal rule removed some of the worst fraud from the page. It did not give the page the one thing it never had, which is a comparison.
Sources
- Federal Trade Commission. “The Consumer Reviews and Testimonials Rule: Questions and Answers.” ftc.gov
- National Archives, Electronic Code of Federal Regulations. “16 CFR Part 465, Rule on the Use of Consumer Reviews and Testimonials.” ecfr.gov
- Federal Trade Commission. “Endorsements, Influencers, and Reviews.” ftc.gov
- Federal Trade Commission. “Consumer Review Fairness Act: What Businesses Need to Know.” ftc.gov
- Federal Trade Commission. “Featuring Online Customer Reviews: A Guide for Platforms.” ftc.gov
- Federal Trade Commission. “Soliciting and Paying for Online Reviews: A Guide for Marketers.” ftc.gov
- Federal Trade Commission. “FTC's Endorsement Guides: What People Are Asking.” ftc.gov
- National Archives, Electronic Code of Federal Regulations. “16 CFR Part 255, Guides Concerning the Use of Endorsements and Testimonials in Advertising.” ecfr.gov
- National Institutes of Health, Office of Dietary Supplements. “Dietary Supplements: What You Need to Know.” ods.od.nih.gov
- PubMed. “He S, Hollenbeck B, Overgoor G, Proserpio D, Tosyali A. "Detecting fake-review buyers using network structure: Direct evidence from Amazon." Proceedings of the National Academy of Sciences, 2022. DOI: 10.1073/pnas.2211932119.” doi.org
- PubMed. “Strijkers RHW, Schreijenberg M, Gerger H, Koes BW, Chiarotto A. "Effectiveness of placebo interventions for patients with nonspecific low back pain: a systematic review and meta-analysis." Pain, 2021. DOI: 10.1097/j.pain.0000000000002272.” doi.org