Methodology

How ratings work

Formulary was built with one goal: to help you tell the difference between a product marketed as good for your skin and one that actually is.

We review the research and facts behind every ingredient and give it a rating from Best to Worst.

Ratings reflect how effective an ingredient is at doing its intended skincare job, not how suitable it is for every skin type. For example, vitamin E is an effective moisturizing ingredient, but it may not be ideal for someone with oily or acne-prone skin.

543

ingredients in Formulary’s data

156

ratings reviewed by a person

studies and reviews cited

EU regulation entries cited

Counted from Formulary’s data on November 6, 2026.

What the ratings mean

Here’s how to read the ratings and the criteria behind them. The criteria are the guidelines we use in most cases. They aren’t rigid rules or a perfectly mechanical formula, because judging an ingredient takes some judgment.

Best

Great for skin, with a high likelihood of meaningful short- or long-term benefits.

Large effect: studies show a large benefit compared with a control group, a difference most people would notice, or a clearly meaningful improvement in a relevant measure (for example, acne lesions cut by about a third or more). Or there’s no qualifying study yet, but dermatologists widely recognize it as an effective active. We hold a high bar for this.
And safe: no safety flags or concerns in the EU.

Good

Does a helpful job on the skin, or has a reasonable likelihood of some short- or long-term benefit.

Modest effect: studies show a real difference, but a smaller or subtler one. Or it does a known job on the skin, like cleansing, without supporting research.
And safe: no safety flags or concerns in the EU.

Neutral/Average

A solid, safe ingredient, but don’t expect a major improvement from it.

Minimal effect: studies show a tiny effect or none at all. Or it’s formula-functional: it helps the product work, for example as an emulsifier or a preservative.

Bad

A real risk of irritating the skin or causing an allergic reaction for some people. Worth avoiding, especially if your skin is sensitive.

Irritation or sensitization: research shows this ingredient, or its category, has some risk of irritating or sensitizing the skin, with strong or moderate evidence. A mild concern, or one with only weak evidence, is shown as a note instead and doesn’t change the rating. Or
EU-listed fragrance allergen: one of the 82 fragrance allergens EU law says must be named on the label (above 0.001% in leave-on and 0.01% in rinse-off products). This applies even when other studies suggest it rarely causes allergy, as with benzyl alcohol.
A concern always outranks a benefit: an ingredient that meets any of these is Bad even if it also has proven benefits.

Worst

Unsafe, banned, or one of the most common causes of skin reactions. We recommend finding a better product.

Banned or unsafe: banned in the EU (Annex II), such as butylphenyl methylpropional (Lilial), or the EU’s safety committee (SCCS) has concluded it isn’t safe in cosmetics. Or
A common allergen: reactions in more than about 3% of patients patch-tested in Europe, such as methylisothiazolinone or fragrance (“parfum”). Or
Becomes an allergen over time: harmless when fresh, but reacts with air to form strong allergens, such as linalool and limonene once oxidized. Or
Irritation or sensitization: research shows this ingredient, or its category, has some risk of irritating or sensitizing the skin. A mild concern, or one with only weak evidence, is shown as a note instead and doesn’t change the rating. Most fragrances fall in this category.

Unknown

We haven’t rated it or reviewed the research yet, or
it isn’t in our database, or
we can’t find enough information to recommend a rating.

A note on safety and irritation. Generally, if an ingredient has a safety concern, is an allergen, is banned in the EU, or is a known skin sensitizer, it gets a Bad or Worst rating. Not everyone is allergic to every allergen, but most allergens can sensitize skin, so we’ve chosen to be conservative. Our philosophy: there are a lot of great ingredients out there, so why put these in your skincare?

Definitions, labels and notes

Term or label	What it means
Skin-functional	An ingredient meant to do a job on the skin: cleanse, moisturize, soften or protect.
Formula-functional	An ingredient that helps the product itself work, like a thickener or an emulsifier. These aren’t bad: they’re important for making the product work. They just usually aren’t meant to change your skin.
Effectiveness	Is there science-backed evidence that the ingredient benefits the skin? Science-backed means at least one study that passes our quality bar (see Which studies count). We measure how big the benefit is, compared with a control group: Large: a difference most people would notice, or a clearly meaningful improvement in a relevant measure (for example, acne lesions cut by about a third or more) Modest: real, but smaller or subtle Minimal: a tiny effect or none at all
Evidence level	Best · Strong evidenceBest · Moderate evidenceGood · Limited evidence Strong evidence: a systematic review, or at least two agreeing controlled trials, at least one of them independent. Moderate evidence: at least one good independent controlled trial (blinded, about 20 or more people, at least 4 weeks). Limited evidence: everything else that passes our quality bar, for example only company-funded trials, or a single trial that’s small, short or not blinded.
Allowed to a limit	Allowed only up to a limit or under conditions (EU Annexes III, V and VI). Shown as a tag, “safe to a limit” or “safe with conditions”, with the actual limit or condition in the description. Other tags: “UV filter limit in the EU”, “partly banned in the EU” or “may contain a banned substance”.
Mild concern	A mild concern from research, such as “can irritate”, or any concern with only weak evidence. It’s shown, but it doesn’t change the rating.
Human reviewed	A person has reviewed the rating and the evidence behind it.
Not human reviewed yet	A custom agent generated the rating using Formulary’s criteria and methodology. We plan to review it by hand soon.
AI estimate	The ingredient isn’t in our database yet, so Claude, an AI model, gave an estimated rating.

Which studies count as strong and moderate evidence

A study counts toward a benefit only if it passes every check below. Studies that fail are left out entirely.

On human skin, applied topically. Not swallowed, and not only on cells or animals.
Isolates the ingredient. The same formula with and without it, or the ingredient on its own. Whole-product tests and combinations don’t count.
Controlled. Compared against a group that didn’t get the ingredient.
Peer-reviewed and not retracted. Single case reports don’t count.
Realistic concentration. A strength a cosmetic product would actually use.

What counts as independent: a trial not paid for by a company that sells the ingredient. Either the paper says so, or it doesn’t mention funding and every author works at a university or hospital, checked against the full author list. Unknown funding with any company author, or with authors not yet checked, isn’t independent.

Not medical advice. Ratings are general information about ingredients. Ask a doctor or dermatologist about allergies, reactions or skin conditions.

See a rating you think is wrong? Tell us, and include the ingredient’s name.