Fragrantica.
The deepest fragrance record.
127K+ fragrances · 7.6K+ brands · 2.9K+ perfumers · 2.5K+ notes · 92 accords — every record sourced, normalised, cross-referenced. Reviews & news parquet datasets carry 4,643,851 reviews, 24,440 editorial articles, and 263,798 community comments in 23 languages. One auditable archive.
cryptocurrency only
The complete perfume
ledger — every entity
indexed and joined.
Five normalised tables — fragrances, brands, perfumers, notes, accords — plus two heavy parquet datasets for reviews and news. Cross-references resolve by perfume ID across every table. HTML-rich fields preserved as-is.
en
ru
es
fr
de
it
pt
nl
pl
tr
ar
zh
ja
ko
sv
no
da
fi
cs
el
he
uk
huThe complete perfume database, built for developers.
FragDB ships every fragrance Fragrantica has indexed, normalised across five relational CSV tables. Perfume IDs join brands, perfumers, notes, and accords without manual mapping. Every text field, every cross-reference, every rating — preserved as-is for reproducible analysis.
The Reviews & News bundle adds three parquet datasets — community reviews, editorial articles, and threaded comments — for sentiment work, recommender systems, semantic search, and LLM training corpora.
Built for retail engineers, ML researchers, recommender system developers, and academic teams who need an auditable corpus rather than a scraped snapshot.
4.6M reviews.
One columnar drop.
Four heavy datasets land in the same archive: comments.parquet, news.parquet, news_comments.parquet, plus the structured CSVs. Every text row joins back to the master fragrance by perfume ID.
Train collaborative filters on millions of user reviews with embedded note vectors.
Multilingual review corpus with star ratings — supervised, out of the box.
Index review and editorial text with vector stores; retrieve by smell, mood, occasion.
Augment retail catalogues with notes, accords, perfumer credits, similar-product graphs.
Olfactory science, consumer behaviour, perfume industry economics — citeable corpus.
Domain-specific data for fine-tuning models on fragrance, scent, and perfumery vocabulary.
Static archives
go stale.
Each month brings hundreds of new fragrance launches, thousands of fresh reviews, dozens of editorial articles. A one-shot purchase is a snapshot. A subscription is a living catalogue — your application stays current without you re-shipping data.
The catalogue ships as 5 normalised CSV tables plus 3 parquet datasets (reviews, news, news_comments), all bundled in a single ZIP archive. Joinable by perfume ID across every table.
After payment confirmation you receive an email with a signed download link. The link stays valid for 3 days and permits 6 downloads. Subscribers get a new link with every monthly drop.
Cryptocurrency only. No card processors, no third parties, no chargebacks. Email verification is required before purchase.
Annual subscribers receive 3 updates per month — 36 per year. Lifetime customers receive every update for the life of the project. One-time purchases (Core, Full bundle) do not include updates.
Yes. License covers commercial use — recommender systems, e-commerce, mobile apps, academic publication. Redistribution of the raw archive is not permitted.
Due to the digital nature of the product all sales are final. If you hit a technical issue with the download, contact support and we will resolve it.
Get Fragrantica.
Join developers and researchers using FragDB for their fragrance data needs. Pay in crypto. Receive a signed link within minutes — and start joining 4.6M reviews to 130K+ fragrances by perfume ID, across 23 languages.