How we build it, and why that matters

Three kinds of source, and we keep them distinct because their licences are distinct.

Our own work. Products entered and verified by BeautySignal, the plain-English ingredient explanations, and all the normalisation and reconciliation that turns messy records into a usable catalogue.

Regulatory reference. The European Commission's cosmetic ingredient reference and the Cosmetics Regulation itself, which are reusable under CC-BY 4.0 with attribution. This is where the ingredient knowledge base comes from.

Open community data. Open Beauty Facts, whose database is under the Open Database License. Useful, credited, and deliberately kept separable — see below.

The Open Database License carries a share-alike obligation that attaches to a derivative database. If we folded that material indistinguishably into our own, we would risk the whole catalogue inheriting that obligation.

So every record carries the licence class of the source it came from, for the life of the record. Share-alike-derived material can be identified and separated in a single query. That is a technical property of the schema, not a promise in a policy document.

We don't scrape sites because it is technically possible. We don't exceed a source's published rate limits or work around its terms. We don't copy product photography we have no licence to use — which is why you will see a designed placeholder rather than somebody else's image.

811 products across 390 brands, 810 barcodes and 185 ingredients. Early, and growing. We would rather publish the real number than a rounded-up one.

Every scan that finds nothing, and every search that comes back empty, goes onto the list of what to add next. That is the main thing that decides what gets added.

Where a product record draws on a source that requires attribution, that source is named on the product page with its licence. The full list of sources is public, including the ones we decided not to import from and why.