Why does a placebo work on skin?

Some of it is not the placebo at all. Skin conditions come and go, people apply a base cream that moisturises, and everyone in a trial is watched, measured and reminded. A vehicle arm captures all of that at once, which is why it is the only fair thing to measure a treatment against.

The same cream, gel or shampoo without the active ingredient in it. It is the strictest comparator a topical product can have, because both arms feel identical to the person using them.

Is a bigger margin always better?

No. A margin is only comparable inside one trial. Two treatments measured in different trials, on different people, with different outcome definitions cannot be ranked by their margins.

Free tool · no signup

How much of the result was the treatment?

Pick a treatment. See what it did — and what the arm with nothing in it did.

Choose a treatment

Compared against

22 treatments with a published two-arm result · 14 against a dummy

Reached clear or almost clear at week 12

On the treatment37.9%Adapalene + benzoyl peroxide

On the comparator17.9%Gel vehicle

The margin 20 percentage points

This was a dummy: the same formula with the active taken out. The gap is the ingredient.

Trial detail

The active: Adapalene 0.1% + benzoyl peroxide 2.5%
The trial: NCT00421993 — randomised, double-blind, four-arm, results posted.
In the trial’s own words: In the same trial: adapalene alone 21.8%, benzoyl peroxide alone 26.7%. Total lesions −57.7% versus −35.2% on vehicle.
What it means: The combination beat either ingredient alone by a clear margin — the whole is genuinely more than the parts, which is unusual. The FDA approved it for over-the-counter sale on 22 May 2026 (NDA 220736), the first OTC dual-active acne product; in Canada it remains prescription-only.
Source: NCT00421993 Open the record ↗

See the evidence-ranked options for each concern All tools

Sources & limitsWhere the answer comes from and what it cannot prove

What the two numbers are

Every row is one registered trial with results posted. The first number is the treated arm. The second is the arm it was compared against, exactly as the trial defined it — a vehicle, a placebo, another treatment, or nothing.

The gap between them is the margin. It is the part of the result the treatment is responsible for; the rest happened anyway.

Three kinds of comparator, kept apart

A dummy — a vehicle, placebo or sham. This measures the ingredient, because the two arms are otherwise identical.
Another treatment — this measures a choice between two real options. It says nothing about how much of either result was the placebo response.
Nothing — no cream, no sheeting, wait and see. Useful, but the two arms differ by more than the ingredient.

Rows are never mixed across these kinds, because the margins do not mean the same thing.

What this can’t tell you

A margin belongs to one trial: its people, its dose, its outcome definition and its length. Two treatments from two different trials cannot be ranked against each other here, and a result measured at week 12 says nothing about week 52.

For the full record behind each row, including who was studied, see the evidence index.