How We Rate Skincare Evidence: The Scale, and What Disqualifies an Ingredient

Every ingredient page on this site carries a grade before it carries a description, because "studies show" is doing an enormous amount of unexamined work in skincare writing. A study exists for nearly everything. What differs is what kind of study, on how many people, funded by whom, and whether the finding has ever been repeated by someone with nothing at stake.

This page is the method. It exists so the grades can be checked rather than trusted, and so I cannot quietly loosen the scale as the ingredient list grows.

How we rate skincare evidence: the hierarchy it borrows

The ranking used here is the standard one from medicine, not something invented for this site. StatPearls describes the principle behind the evidence pyramid: different types of evidence do not carry the same weight. And the shape of it: In its traditional form, animal and translational studies, case reports, and expert opinion sit near the bottom; observational studies occupy the middle; and randomized controlled trials, systematic reviews, and meta-analyses appear near the top.

One more line from the same chapter matters more in cosmetics than anywhere else, because it is the gap most product claims live in: Even a strong trial may not translate neatly if the study population differs from the patient in age, comorbidities, baseline risk, or clinical setting.

A well-run trial in 40 people with moderate acne is a real result about 40 people with moderate acne. Whether it says anything about your dry, sensitive, non-acneic face is a separate question that the trial did not ask.

The four grades

Grade	What it means	What earns it
Strong	Repeated human evidence, independently	Two or more human randomized trials agreeing, with replication beyond a single group
Moderate	Real human evidence, thin in some way	Human trials that are single, small, manufacturer-funded, or unreplicated
Limited	Human data, weak design	Open-label studies, questionnaires, consumer self-assessment
In vitro only	Not tested on people	Cells in a dish, or animal work alone

When one page needs two grades. Some names cover more than one thing. "Retinoid" is a category containing prescription molecules with first-line trial support and cosmetic ingredients without it, so a single grade would average away the distinction a reader needs. In that case the page carries both, labeled separately, and says which applies to what. Retinol versus retinoid is the first of these. The frontmatter records it as Split, which is a pointer to two grades in the text rather than a fifth grade.

And the rule that does the real work: an ingredient with no human data at all does not get a page. Not a low grade. No page.

That is deliberate, because a low grade still puts the name in front of you, and a name in front of you is most of what marketing needs. The absence is the honest treatment.

What "in vitro only" actually means

It means an effect was observed in cells, or in a laboratory model, and it has not been shown to happen on a face.

This grade exists because it covers a large share of the mechanisms quoted in skincare copy. A molecule that increases collagen synthesis in fibroblast culture has done something real and measurable. Whether the same molecule, in a cosmetic formulation, at the concentration used, penetrating a living stratum corneum, produces a visible change in a person is a chain of four additional questions, none of which the cell study answered.

The evidence hierarchy places translational work near the bottom for exactly this reason. It is where good ideas start, and it is not where claims should be made.

How we rate skincare evidence in practice, step by step

The procedure, which is the same for every ingredient and recorded on each page:

1. Search the human literature, not the brand's summary of it: PMC for randomized trials, the Cosmetic Ingredient Review's assessment, and the American Academy of Dermatology's material.
2. Apply the table above, taking the highest tier the evidence genuinely reaches rather than the highest tier it is described as reaching.
3. Record the sources and the date checked in the page's own metadata, so the call can be re-examined.
4. Re-audit five earlier pages for every twenty new ones, because the failure mode of any scale is drift, and drift is invisible from the inside. Salicylic acid is the first page graded this way, with its reasoning shown rather than summarized, and benzoyl peroxide is the first to reach Strong.

Step 4 is there because grading gets easier as you go, and easier grading is usually looser grading.

Who funds cosmetic ingredient safety reviews

Worth stating outright, since the Cosmetic Ingredient Review is one of the sources used here and its structure deserves disclosure rather than a footnote.

CIR describes its own origin: it was established in 1976 by the industry trade association, now the Personal Care Products Council, with support from the FDA and the Consumer Federation of America. On its independence, it states that Although funded by the Council, CIR, the Expert Panel for Cosmetic Ingredient Safety, and the review process are independent from the Council and the cosmetics industry.

Industry-funded, with a claimed firewall and outside representation on its steering committee. That is not a reason to discard its assessments, which are detailed and public, and it is a reason to say where they come from every time they are used. Independence claimed is not the same as independence demonstrated, and you are entitled to weigh that yourself.

The same disclosure logic applies to manufacturer-funded trials, which is why funding is one of the things that holds an ingredient at Moderate rather than lifting it to Strong.

How to judge a skincare study yourself

Five questions, in the order that eliminates the most claims fastest:

1. Was it done on people? If not, the ceiling is In vitro only regardless of how strong the mechanism sounds.
2. How many people, and for how long? Twelve participants over four weeks is a pilot, described honestly as a pilot by researchers and as proof by marketers.
3. Who paid for it? Manufacturer funding does not make a result false. It makes replication necessary.
4. Was the outcome the thing you care about? Improvement measured by participant questionnaire is a different claim from improvement measured by a blinded assessor. In pore research, for instance, the measurement problem is severe enough that the review literature says so outright, as large pores on nose covers.
5. Has anyone independent repeated it? This is the question that separates Strong from Moderate, and the one almost never mentioned in copy. Running these five questions on a real product claim shows how quickly they narrow it.

What this scale cannot tell you

Two limits worth being explicit about.

Does this ingredient have evidence? is not the same question as will it work on me. A grade describes the literature, not your face. Strong evidence for an ingredient in a population does not predict how you will respond, and the applicability point from the evidence hierarchy above is the reason.

A grade is not a safety verdict. Evidence of efficacy and evidence of safety are separate literatures with separate sources, and an ingredient can be well supported for one and thinly documented for the other. Where the boundary sits for topicals in general is collected in what skincare can and cannot fix.

Neither of those is a reason to skip the grade. Both are reasons not to treat it as a purchase instruction.

FAQ

How do you rate skincare evidence on this site? On a four-step scale: Strong for replicated human randomized trials, Moderate for human trials that are single, small, unreplicated or manufacturer-funded, Limited for open-label or self-reported data, and In vitro only for cell and animal work. Ingredients with no human data at all are not given a page.

What does "in vitro only" mean? That the effect has been shown in cells or a laboratory model rather than on people. The evidence hierarchy places translational work near the bottom because the step from a dish to a face involves formulation, concentration and penetration, none of which a cell study addresses.

What counts as strong evidence for a skincare ingredient? Two or more human randomized controlled trials that agree, with replication beyond a single research group. Systematic reviews and meta-analyses of such trials sit at the top of the standard hierarchy.

Who funds cosmetic ingredient safety reviews? The Cosmetic Ingredient Review was established by the industry trade association and is funded by it, while stating that the review process is independent of the Council and the cosmetics industry. Its assessments are used here with that disclosure attached.

Does a strong grade mean an ingredient will work for me? No. A grade describes the strength of the published evidence in the populations studied. Even a strong trial may not translate if the study population differs from you, which is a limitation stated in the evidence-based medicine literature itself.

Why not include ingredients with no human data? Because a low grade still puts the name in front of a reader, and a name in front of a reader is most of what a marketing claim needs. Leaving it out is the more honest signal.

---

How we rate skincare evidence exists so that the sentence "there are studies on this" stops being an argument and becomes a question with an answer. If a grade here ever looks generous compared to the sources listed under it, the grade is wrong, and the whole method is published precisely so that can be said out loud.