About

Search and you will find plenty of writing that says something is good. What is rare is writing that says on what basis.

So we inverted the order. First what trial there was — its number of people, its length, its control arm, who paid for it — and then what that trial did not look at. We do not write the conclusion first and attach the evidence afterwards.

• ingredients — what you apply or take
• medicines — what comes with a prescription or from a pharmacy
• procedures — injections and devices
• devices — the device-delivered procedures, filtered out
• surgery — operations
• skin concerns — where to start reading
• products — examples, not a ranking

There is a reason for the separation. Retinol and tretinoin are in the same family, but one is a cosmetic and the other a drug. A procedure and an operation differ in how far they can be undone. Put them in one place and that difference disappears.

There are four badges — strong, moderate, early, not established.

'Not established' does not mean 'it does not work'. It usually means 'this design cannot tell' — there was no control group, or the numbers were small, or several things went in at once, or nobody measured it.

If a concentration says 'unreported', the paper did not state it, and we do not fill the gap by estimating.

Read the number of people, the length, the control arm and the funding next to every figure. Those four change what the figure means.

• No prices, no clinics, no treatment testimonials.
• No rankings or star ratings on products.
• No telling you to buy or undergo anything.
• No diagnosis and no prescribing. Medicines and procedures are settled in a clinic.

The detailed rules are on the editorial policy page.

Korean, English, Japanese, Chinese, Spanish and Russian. Not run through a translator and pasted in — the same evidence written in each language.

And if text goes missing in one language, the build fails: the types enforce it. It is there to stop one language quietly keeping an older version.