How we grade evidence

Every active ingredient carries an overall grade, and every claim on its entry carries its own. The overall grade reflects the best-supported claim, not an average. The word carries the meaning; the bar is secondary; the system reads in greyscale.

The grades

Strong — Multiple randomised controlled trials in humans with reasonably consistent results, including at least one independent replication. Regulatory assessments and systematic reviews drawing on that base count in favour.

Moderate — Some controlled trials, or one good trial plus consistent mechanistic and observational data. Most well-known actives sit here honestly.

Limited — Small or uncontrolled human studies; a single unreplicated trial; or good evidence only for a related form (a 20 % prescription drug when the cosmetic version is 5 %).

Emerging — In-vitro, animal, or very preliminary human data. Not a euphemism for "doesn't work"; it means we don't know yet.

Functional ingredients

Functional — Not a grade. Thickeners, emulsifiers, surfactants, solvents (including water), pH adjusters, chelators, preservatives, propellants, colourants, fillers, fragrance and product stabilisers are in a formula to make it work, not to change your skin. Grading them for skin benefits would be meaningless, so their entries carry the label Functional ingredient, not graded for skin benefits, describe what the ingredient does in the formula, and list any regulatory limits. Their function tags come from a fixed lookup, not from an automated guess.

What we weigh

• Trial design (randomised, controlled, blinded) and size
• Whether the tested concentration and vehicle resemble what's sold
• Independence of replication and funding
• Consistency with mechanism
• Regulatory findings, which we treat as evidence about safety rather than efficacy

What moves a grade

Replication moves it up. Failed replication, retraction, or a regulatory finding that undercuts the mechanism moves it down. Every change is logged in the entry's public revision history with a one-line reason.

Review

A grade is only as good as the person who checked it. Each entry shows whether a named member of staff has reviewed it and when; entries that have not been reviewed say "Not yet reviewed" and are excluded from search indexes until they are.

Regional limits

Safety statements list the permitted concentration and product type per region — EU, UK, Canada, US, and where relevant Korea, Japan, Australia and China — each with a source. Where a regulator has no limit, we say so rather than leave a blank.

Concentration and pH

Typical concentration ranges reflect what the cited trials used, not marketing claims. Where pH materially affects activity (acids, L-ascorbic acid) we state the range the evidence supports.