How BeautyGuard evaluates cosmetics — a regulatory approach

Earlier versions of BeautyGuard displayed a 0–100 score. We removed it deliberately: a number suggests a precision that public regulatory data cannot deliver, and it blurs the line between "legal status" and "harmfulness rating". Today BeautyGuard describes products and ingredients strictly in regulatory categories.

What you see instead of points

• Banned — the ingredient appears in Annex II of Regulation (EC) No 1223/2009.
• Restricted — an Annex III ingredient: allowed under conditions (concentration, product category, warnings).
• Colourant / preservative / UV filter — an ingredient from the positive lists of Annexes IV–VI.
• Fragrance allergen — subject to mandatory label declaration.
• Safety Gate alerts — real notifications of products containing the ingredient.

How we match alerts

Matching an alert to a product requires an identical EAN code. For ingredient-level matches we apply a confidence threshold — matches below 0.7 are not shown — plus an INCI typo heuristic (edit distance 1), because Safety Gate entries are often typed carelessly.

What this method does not do

It cannot detect allergens hidden behind a collective "Parfum", does not assess concentrations manufacturers do not publish, and does not replace the safety assessment (CPSR) performed by a qualified assessor. The full list of limitations and the versioned history of the method live in our public methodology.

Legal basis and sources: Regulation (EC) No 1223/2009 (EUR-Lex) · the European Commission's CosIng database · EU Safety Gate.