How We Review Our Products

What we are, and what we are not

We review published evidence: peer-reviewed studies, clinical trial records, ingredient and formulation databases, regulatory filings and dermatology literature. Every recommendation we make traces back to a source you can open and read for yourself.

We do not run a testing lab, and we do not claim to. We do not conduct our own clinical trials, take before-and-after photographs, or measure devices on a bench. Where a product’s effect has only been established by the manufacturer’s own unpublished testing, we say that plainly rather than repeating the claim as if it were settled.

That distinction matters more than it sounds. A great deal of beauty coverage presents a personal impression as evidence. We would rather tell you what the literature can and cannot support than tell you how something felt on one face.

How an article gets made

1. We start from the claim, not the product

   Every piece begins with the specific assertion under examination — “retinol thins the skin,” “this wavelength stimulates collagen,” “hyaluronic acid hydrates” — because that is the thing that can be checked. Product names come later.

2. We find where the claim originates

   Marketing claims usually trace back to a single study, and frequently to a study that does not say what the marketing says. We follow the chain back to the primary source and read it, rather than citing the article that cited the article.

3. We weigh the evidence by what it can actually show

   Study design decides how much weight a finding carries. See the hierarchy below. A well-run randomised controlled trial and an in-vitro result on cultured cells are not interchangeable, and we do not present them as though they were.

4. We write the uncertainty in, not out

   Where the evidence is thin, contested, or limited to a population that may not include you, that is part of the finding and it appears in the text. A confident sentence over weak evidence is the failure mode this publication exists to avoid.

5. We publish the sources

   Every article carries its full source list at the foot of the page. If a statement matters, you can click through and check it. If we could not source a claim, it does not run.

How we weigh evidence

Not all findings are equal. We rank the evidence behind a claim roughly in this order, and we tell you which tier a claim rests on when it matters:

1. Systematic reviews and meta-analyses — multiple trials pooled and assessed together. The strongest signal available to us.
2. Randomised controlled trials in humans — especially vehicle-controlled and blinded, on a population and timescale relevant to ordinary use.
3. Controlled human studies without randomisation, including split-face and cohort designs.
4. Ex-vivo and in-vitro work — results on skin samples or cultured cells. Useful for mechanism, weak for predicting what happens on a living face.
5. Manufacturer testing and consumer-perception studies — reported as what they are. “89% of participants agreed their skin looked brighter” is a survey result, not a measurement.

We also read the parts most summaries skip: sample size, trial length, funding source, whether the control was a true vehicle, and whether the measured outcome is the outcome the marketing is selling.

What disqualifies a product or a claim

• No traceable evidence. If the only support for a claim is the brand asserting it, the claim does not get repeated as fact.
• The study does not test the product. A trial on a 10% concentration says nothing useful about a serum containing 0.1% of the same ingredient.
• Undisclosed or unquantified actives. A formula that will not say how much of the ingredient it contains cannot be assessed against the research on that ingredient.
• Safety signals. Credible reports of harm, regulatory action, or device output that exceeds what the supporting research used.
• Claims that are not physically possible. Topicals that claim to do what only a procedure can do get named as such.

Corrections and updates

Research moves and we get things wrong. When an article is materially incorrect, we correct the article itself rather than quietly deleting it, and we note what changed. If you think something here is wrong — especially if you have the source that shows it — tell us on the contact page. Corrections are the fastest route to changing what this site says.

Commercial relationships

Some outbound links on this site are affiliate links, and we may earn a commission when you buy through one. That is disclosed on the articles that carry them and in our terms.

No brand pays for coverage, for a position, or for a rating on this site, and no brand sees an article before it publishes. The commercial link is attached after the assessment is written, never before — which is why plenty of what we publish concludes that a popular product is not worth buying.

Who writes this

Articles are published under the byline of Olivia Morgan, Research Editor, which is where you will find how this publication is produced.