How To Read Supplement Reviews: A Practical Guide

Start with the shape, not the headline.

An average compresses every opinion into one figure, and two very different products produce the same one. A supplement where almost everybody is mildly positive and a supplement where most people are delighted and a sizeable minority are angry can both average 4.7. The second is a much riskier purchase and the average will never say so.

So the first question to ask of any review page is whether it publishes the distribution. Five numbers, one per star level, adding to a hundred. If a page shows you a headline score and no distribution, the distribution is the thing it has decided not to show you.

The second question is the sample. A 4.7 from 20 reviews and a 4.7 from 2,000 are different claims, and neither is meaningful without knowing how many people bought and did not write. That last figure is almost never published, and the people least likely to write a review are the people who quietly stopped in week two.

Vendors in this category frequently supply summary text rather than transcripts, and rendering it as quotation is where review pages become dishonest without anybody deciding to lie.

The test takes ten seconds. Count the distinct review texts and compare with the number of named reviewers. If twelve people carry four sentences between them, those four sentences are categories a system filed reviews under, not things twelve people said.

A page that prints them as twelve quote cards is showing you the same words three times in a column, which is the most recognisable signature of a fabricated review section. A page that rewrites them into twelve different-sounding quotes has manufactured testimonials, which is worse. The honest third option is to print the sentence once with the count attached.

The FTC endorsement guides set out what a seller may and may not do with a testimonial, and the question-and-answer version is the readable one. The consumer guidance on evaluating reviews is the other half of the same subject.

Some patterns that should raise an eyebrow, none of which is proof of anything on its own.

• No distribution. A headline average with no breakdown beneath it.
• No low scores at all. Every product in this category has people who wanted more, and a page with none has filtered them.
• Reviews that name a measured outcome. Very few buyers measure anything, and a page full of specific numbers is a page where somebody supplied them.
• Identical sentence structure across reviewers. Real people write differently; summary categories do not.
• No dates or day counts anywhere. The most informative field, absent.
• A review count that never moves. Worth checking twice a few weeks apart.
• Outcome claims in the reviews that the label does not make. A label is constrained by regulation. A testimonial is the usual route around that, and the FTC has a consumer page on exactly that practice.

The last one is the most consequential. If a product page carefully says “supports” and its review section says something much stronger, the review section is doing work the label was not allowed to do.