Methodology

How AllerNote works

No black box. This page explains exactly what happens when you scan a product — where our allergen data comes from, how we match ingredient names, what each verdict means, and where the limits are.

What happens when you scan a product

Four steps, usually about four seconds.

  1. We read the label

    You take a photo of the ingredient list or paste it as text. Our AI extracts every ingredient — including labels in multiple languages and the tiny, dense INCI print that's nearly impossible to read by eye.

  2. We identify each ingredient

    Every extracted name is matched against our allergen database in tiers: first an exact name match, then the ingredient's known synonyms (one allergen can appear on labels under a dozen names), and finally a semantic match for spelling variants and unusual phrasings — which is verified before it's used. Matches are cached, so repeat ingredients resolve instantly.

  3. We check them against your profile

    Each identified ingredient is compared to your saved allergens — patch-test results, known triggers, and any custom ingredients you've added. We also look for umbrella terms like "parfum" or "fragrance", which can legally hide individual fragrance allergens without listing them.

  4. You get a verdict with reasons

    Every scan ends in Safe, Warning, or Avoid — and shows exactly which ingredient triggered it and under what name it appeared on the label. A "Safe" verdict can still carry a note, for example when "parfum" is present and you have a fragrance allergy we can't rule out.

Where the allergen data comes from

Clinical contact-allergy sources, not "clean beauty" opinion.

Clinical contact allergens

The database is built around the allergens dermatologists actually test for: the EU's 26 labelled fragrance allergens, MI/MCI and other preservatives, formaldehyde-releasers, lanolin, metals like nickel and cobalt, and allergens highlighted by the American Contact Dermatitis Society — including its Allergen of the Year selections.

Standard patch-test panels

The Patch Test Translator recognises the standardised series your report is likely based on — the TRUE Test, the ACDS Core Allergen Series, and the European Baseline Series — and maps each result to the ingredient names that appear on real product labels.

INCI naming & synonyms

Cosmetic labels use INCI nomenclature, and one allergen can hide behind many names — a formaldehyde allergy means watching for DMDM hydantoin, quaternium-15, imidazolidinyl urea, and more. Each allergen in our database carries its list of synonyms and label spellings, and that list grows as we find new ones.

Cross-reactor mapping

Some allergens travel in groups — react to one and you may react to its chemical relatives. The Cross-Reactor Checker maps these documented relationships so a single patch-test positive doesn't leave you blind to the rest of its family.

How we research and maintain the written content on this site — the ingredient encyclopedia and guides — is covered separately in our editorial policy.

What a verdict means — and what it doesn't

A verdict is a label-reading result, not a medical judgement. Avoid means the product lists an ingredient that matches something in your profile. Warning means there's a likely or partial match worth a closer look. Safe means we found no match between the listed ingredients and your saved allergens — it is not a guarantee the product can't irritate you, because skin can react to things outside any database, and no scan can rule that out.

That's also why AllerNote never gives products a universal score. A product that's perfectly fine for most people is still an Avoid for you if it contains your trigger — and a product other apps rate poorly may be completely safe for your skin. The only comparison that matters is the label against your list.

Where the limits are

Honest about what a scanner can't do.

  • We don't diagnose. AllerNote works with allergens you already know — from a patch test, a clinician, or a clear pattern of reactions. Identifying new allergies is your dermatologist's job, not ours.
  • A scan is only as good as the label. Blurry photos, glare, and worn packaging reduce extraction accuracy. If an ingredient looks wrong in your scan result, retake the photo or paste the list as text.
  • Umbrella terms hide things. "Parfum", "fragrance", and "aroma" can legally stand in for dozens of individual substances. When we see them, we say so — but no scanner can see inside them.
  • Formulations change. Brands reformulate without warning, and regional versions of the same product can differ. The label in your hand is always the final authority over any database, including ours.
  • No database is complete. If your allergen isn't in ours, add it as a custom ingredient and every future scan will check for it. Spotted a missing synonym? Tell us — reader corrections make the database better for everyone.

Questions people ask about our methodology

How accurate is the ingredient matching?

Matching runs in tiers: exact name, then known synonyms, then a verified semantic match for spelling variants. Regulated allergen groups like the EU 26 fragrance allergens and formaldehyde-releasers get extra checks so they are never silently missed. When we are not sure, you see a Warning with the reason rather than a silent pass.

Which allergen lists and panels does AllerNote cover?

The database centres on clinically tested contact allergens: the EU 26 labelled fragrance allergens, MI/MCI and other preservatives, formaldehyde-releasers, lanolin, nickel and other metals, and allergens recognised by the American Contact Dermatitis Society. The Patch Test Translator recognises the TRUE Test, the ACDS Core Allergen Series, and the European Baseline Series.

Why did AllerNote flag a product that another app rates as safe?

Other apps rate products against a generic standard; AllerNote checks the label against your personal allergen list. A product can be well formulated by any general measure and still contain the one ingredient your patch test flagged. The reverse also happens: a product rated poorly elsewhere may contain nothing you react to.

Can a product marked Safe still cause a reaction?

Yes. Safe means no listed ingredient matched your saved allergens. It cannot account for allergens you have not saved, undisclosed components inside terms like parfum, contamination, or irritation that is not allergy at all. If a Safe product reacts, log it — the Reaction Detective looks for the common ingredient across your reactions.

Does AllerNote use AI, and can it get things wrong?

AI does the reading — extracting ingredients from photos and resolving unusual name variants — and a database of clinical allergens does the judging. Extraction can misread a blurry label, which is why every scan shows you the ingredient list it read, so you can verify it against the package.

Is AllerNote medical advice?

No. AllerNote is an educational label-reading tool. It helps you apply the allergen list you already have; diagnosis and treatment stay with your dermatologist.

See it work on a real label

Scan any product free, no account needed.

Try the scanner