How we
test
A review is only worth reading if you can see the ruler. This page is the ruler: six weighted criteria, a panel protocol, a list of things that fail a product outright, and an honest statement of what we do when we have not been able to test something ourselves.
The weights below were set before any product was scored and have not been adjusted to suit a result. If they change, the change is logged with a revision number and every affected review is re-scored.
The scoring model
- 01
EVIDENCE FOR THE ACTIVE
30%Does the published literature support this molecule at this concentration for this claim? Graded A–C, same scale as the evidence register.
- 02
FORMULATION INTEGRITY
20%Working pH inside the active's window, a vehicle that permits penetration, and packaging that does not oxidise the contents.
- 03
DISCLOSURE
15%Are concentrations published? Is the full INCI available before purchase? Are claims quantified or is it all 'up to'?
- 04
TOLERANCE IN USE
15%Panel-reported irritation, stinging and flaking across the trial window, against a vehicle-only control arm.
- 05
TEXTURE & WEAR
10%Absorption, residue, pilling under sunscreen or make-up. The reason people stop using a product that works.
- 06
COST PER APPLICATION
10%Price divided by realistic applications per unit at the label dose — not price per millilitre.
When we have not tested it
Some products we assess from published information alone — because they are prescription-only, because the formula is compounded per patient, or because we have not yet run a panel. We mark those DESK REVIEW and we do not pretend otherwise.
On a desk review, the two criteria that need the product in hand — tolerance in use and texture and wear — are left unscored rather than estimated. The total is then calculated across the criteria we could actually assess, and the coverage percentage is printed beside it. A 47 out of 100 on 75% of the model is a different claim from a 47 out of 100, and it is labelled as one.
We will not take a brand’s own customer satisfaction survey and convert it into a tolerance score. An uncontrolled, unblinded, self-reported survey of people who paid for a product measures satisfaction, not effect.
Outright failures
These are not deductions. Any one of them caps a product’s score in the relevant criterion at or near zero, however good the rest of the formula is.
- 01UNDISCLOSED CONCENTRATIONSIf the percentage is not published, the claim cannot be assessed. Scored zero on disclosure.
- 02JAR PACKAGING FOR OXIDISERSRetinoids and ascorbic acid in an open jar degrade before the trial window closes.
- 03pH OUTSIDE THE ACTIVE'S WINDOWA 20% vitamin C at pH 4.2 is not a 20% vitamin C. Formulation integrity fails outright.
- 04“CLINICALLY PROVEN” WITHOUT A METHODNo n, no duration, no design, no score. We ask once and publish the answer or the silence.
- 05BEFORE / AFTER PHOTOGRAPHY AS EVIDENCELighting is a more powerful active than anything on the market. It is never accepted as data.
- 06PAID PLACEMENTNo product enters the register in exchange for payment. See the disclosure document.
How we get the products
Our default. We pay full price with our own money and the receipt is filed against the review.
Declared permanently at the top of the review. Never carries an obligation to publish, or to publish by a date.
Labelled as advertising, excluded from the register, given no score. See the disclosure document.
- Panel sizes of 12–20 detect large effects, not small ones. We do not report differences this method cannot resolve.
- We are not a laboratory. We do not assay what is actually in a tube; we work from published formulas and published INCI.
- Blinding is imperfect. A tester can often guess a strong retinoid from how it behaves.
- Reformulations happen silently. A review describes the formula on the date printed on it.