English
What Ambarium is, and how the verification works
Every "best dupes" list on the internet is one person's opinion. Ambarium is an attempt at something different: aggregate what the fragrance community has already concluded across thousands of discussions, and show the receipts.
How a dupe gets verified
A pair like Creed Aventus → Armaf Club de Nuit Intense Man only appears on Ambarium after clearing a cross-validation pipeline:
- Collect claims from four independent source types: Reddit posts, Reddit comment threads, clonespreadsheet's community-rated similarity sheet, and brands' own "inspired by" product pages.
- Extract and normalize each "X is a dupe of Y" claim (an LLM pass over the raw threads; the corpus is tens of megabytes of discussion text), keeping the original quote and link for every claim.
- Cross-validate. A pair is only "verified" when multiple independent source types agree. One viral comment is not verification; agreement between strangers in different places is.
- Show the receipts. Every dupe page displays the actual quotes, with source and score, that back the pair, chosen relevance-first (a quote must actually discuss the dupe, not just be popular).
What the numbers mean
The similarity % on a pair is clonespreadsheet's community rating: aggregated perception, not lab analysis. "Verified by N sources" counts independent source types, not raw mentions. Critical quotes are shown too: if the community says a clone is only average, the page says so. That's the point.
The ranked lists
The men's and women's lists are built from per-perfume community vote breakdowns (ratings, seasons, longevity, sillage, price-value) published on Fragrantica, re-ranked with a Bayesian weighting so a perfume with 40 votes can't outrank one with 4,000 on a lucky average. Nothing on any list is sponsored, and there is no way to pay for placement.
Independence
Ambarium is built and run by one independent developer. No fragrance house, retailer, or clone brand is involved; there are no affiliate links, no sponsorships, and no accounts or tracking beyond anonymous page analytics. Community data is credited to where it came from (Fragrantica's vote breakdowns, Reddit's discussions, clonespreadsheet's ratings), and every dupe page links back to its sources.
Honest limitations
- Clone houses reformulate; batches vary. A pair that was 1:1 last year may drift.
- Skin chemistry is real. Community consensus is a strong prior, not a guarantee.
- The data refreshes periodically, not live. Very new releases can lag.
Think the consensus on a pair is wrong? The fastest way to improve the data is to say so in the communities it comes from, and the next refresh will pick it up.