How we score

Every product gets a score from 0 to 10, to one decimal place. Here's how that number is built and where its limits are.

Note: We don't lab-test products, and we don't personally try every product we score. Scores are editorial opinions based on published ingredient lists and specifications, verified creator reviews, community consensus and value.

Four rubrics

A serum, a lipstick, a perfume and a hair dryer succeed in different ways, so each product is scored with the rubric for its kind. Every rubric has five factors, each scored 0 to 10.

Beauty score: skincare, hair and body

Beauty score: makeup

Scent score: fragrance

Tool score: hot tools and devices

How we weigh the evidence

• Agreement counts most. We don't test everything ourselves, so results and performance scores rest on independent sources agreeing. If a brand promises a brighter, more even tone in four weeks, do creators who used it for a month show it? Does a "24-hour" foundation still look intact in end-of-day wear tests? Where sources conflict, the score comes down and the page says why.
• Ingredients are judged by evidence, not buzz. Formula credit goes to actives with good human evidence, used at disclosed, meaningful strengths, in packaging that protects them, such as opaque or airless containers for vitamin C and retinoids. Fragrance and other common irritants count against a formula for sensitive skin, and the page says so.
• Scores are relative to the category. A 9.0 drugstore moisturizer is an excellent moisturizer, not a rival to a 9.0 LED mask.
• Value uses prices we can cite. Price per ounce, per milliliter or per use comes from a dated price from the brand or a retailer other than Amazon.
• Specific beats vague. A brand that discloses percentages, shade undertones, UV filters, wavelengths or clinical study details earns more credit than one that says "clinically proven" without showing the study.
• No medical claims. We score how well a cosmetic does its cosmetic job. We don't score claims to treat acne, eczema, rosacea or hair loss; we report them as the brand's claims. See our editorial standards.

From factors to one number

The overall score is the weighted average of the factors, calculated automatically so nobody can type in a headline number. A factor we can't judge fairly is left blank and the others' weights scale up to fill the gap. If the scored factors cover less than half the rubric's weight, the product gets no overall score until we have more evidence.

Score bands

We use the whole scale and publish low scores, because knowing what to skip is useful too. Badges such as Editor's pick, Best value or Trending are editorial labels and never change a score.

Community ratings

Members rate products from 1 to 10 and log what a product did for them on the Benefit Wheel, from hydration and oil control to long wear and frizz control. That community rating sits beside ours and is never averaged into the editorial score, though member sentiment feeds Community love.

The community rating is Bayesian-smoothed, so one enthusiastic rating can't top the charts. Every product starts as if four members had rated it 7.4, and real ratings take over as they add up: a single 10 shows as 7.9, while twenty ratings averaging 9.0 show as 8.7. Logged benefits appear as a heat map on each product's wheel, separate from the brand's claims.

How we verify videos

• We check every YouTube and TikTok video with the platform's official oEmbed endpoint, which confirms it's public and embeddable and supplies the title and creator name we show.
• An editor confirms it's a review, try-on, wear test or first impressions of that exact product. Titles that don't name the brand or product are flagged for a closer look.
• We prefer independent creators and videos from about the last three years. If only a brand's own video exists, we say so, and paid partnerships are labeled "Sponsored by the brand."
• Each video is labeled Positive, Mixed, Critical or Overview, based on what the creator actually says.
• We re-check videos whenever we refresh the catalog and remove any that go private or disappear. Players load only when you press play.

How we choose quotes

Quotes come from Reddit, beauty forums, blogs and the press. Each one is verbatim (trimmed with an ellipsis, never paraphrased), 60 words or fewer, and linked to the original. We never quote Amazon customer reviews, and we include the complaints that keep coming up, not just the praise.

How often we update

Every product page shows when it was last reviewed. We revisit a product when something changes: a reformulation, new shades, a price shift, reliability reports, new creator reviews or a reader correction. Otherwise we aim to re-review each product at least once a year. Mistakes are handled under our corrections policy.

What we don't do

• We don't lab-test products, measure SPF, verify ingredient percentages or run clinical trials ourselves, and we don't personally try every product we score.
• We don't sell reviews, scores, badges or placement, and commission rates are never an input. See our affiliate disclosure.
• We don't show Amazon prices or star ratings.
• We don't score medical or health claims; we report them as the brand's claims.

Questions about a score? Ask us.