Frequently Asked Questions
Everything you need to know about fragrance clones, notes, maceration, and verified Indian sellers.
We act in steps, and every step can be appealed.
- A review we can't stand behind stays on the page, marked "not counted", and stops affecting any score.
- An account whose reviews we find were manipulated goes back to new-reviewer standing for 180 days.
- A seller we find behind manipulated reviews loses trust-score points. The penalty shrinks a little every day and is gone after a year.
- In serious or repeated cases, the seller is removed from the Bureau.
Anyone affected is told what happened and why. You can ask us to look again, and if you disagree with that answer, you can ask once more; the second answer is final. If we reverse a decision, the scores correct themselves overnight.
A fragrance is built in three layers. The Bureau compares the clone's notes against the original's, layer by layer, and weights each by how much of the wearing experience it accounts for.
| Layer | Weight | Why |
|---|---|---|
| Top notes | 10% | The first ten minutes. Easy to imitate, and gone quickly |
| Heart notes | 30% | The body of the fragrance, through the first few hours |
| Base notes | 50% | Weighted heaviest. What is still on skin after an hour — the part you live with, and the part a clone most often gets wrong |
| Overall accords | 10% | The character of the whole, independent of individual notes |
Where a seller's claimed match and wearers' ratings disagree by more than 15 points, the listing shows an Accuracy Disputed notice. The Bureau does not quietly average the difference away.
A note comparison is chemistry on paper. It cannot tell you whether the clone smells right on skin, so people who bought it rate accuracy from 1 to 10 — and their answer takes over from the paper as it accumulates.
Score = 40% note match + 60% community side
The community side starts at 50% note match and 50% the seller's own claim. Every review moves it toward what wearers actually reported — more reviews, and more trustworthy ones, move it further. With no reviews at all the score is 70% note match and 30% the seller's claim.
It never jumps at a fixed number of reviews: two ratings nudge a score, they do not rewrite it, because an average of two people is a measurement of two people. The seller's figure is only ever part of the starting point, it fades as reviews arrive, and it is labelled claimed wherever it appears on its own. How much any single review weighs is not published — that is what keeps it from being gamed.
10 components, each measured from a record. A component the Bureau has no evidence for is left out entirely and its weight shared among the rest — it never quietly counts as full marks. 6 of them come from buyers rather than from the Bureau's own measurements, and carry 66% of the score between them.
| Component | Weight | Measured from |
|---|---|---|
| Authenticity | 18% | How far the seller's claimed match sits from what wearers actually reported |
| Order Accuracy | 14% | Did what arrived match what was ordered — reported by confirmed buyers |
| Packaging Quality | 12% | How well orders were packed, reported by confirmed buyers |
| Delivery Speed | 12% | How quickly orders arrived, reported by confirmed buyers |
| Communication | 10% | How the seller answered questions, reported by confirmed buyers |
| Issue Resolution | 10% | What happened when something went wrong. Buyers who never had a problem skip this rather than guess |
| Reputation | 8% | Average star rating across the seller's listings and their shop feedback |
| Price Competitiveness | 6% | Per-millilitre position against the market median, compared within the same size band. Decants are excluded |
| Behaviour Compliance | 6% | Accuracy flags raised against the seller's listings, and bans |
| Listing Maintenance | 4% | 60% how recently the listings were edited, 40% how recently they were reviewed. A listing never observed being edited counts as unmeasured, not as neglected |
Below 3 measured components a seller reads UNRATED — not a low score, not a provisional one. The reason is arithmetic: the fewer things you measure, the more each survivor dominates. Scored on one component alone, every unflagged seller would read 100 out of 100. Few components and high confidence are exactly backwards, so the composite is withheld until it rests on something. Today 8 of 9 sellers correctly read UNRATED.
A shop with only a few reviews sits close to the Bureau average, and moves to its own score as more people report on it.
Reviewing a fragrance
Only someone the Bureau recorded going to buy that exact listing. The verified-buyer mark is set on the server from that record — a reviewer cannot claim it for themselves. There is no waiting period: the click is what verifies, and waiting a week never verified anything.
Reviewing a shop
Stricter. Only someone who confirmed they actually completed a purchase from that seller. Feedback about a shop carries further than a review of one bottle, so it opens later. One review per person per listing, editable and removable by whoever wrote it — and an edited review is marked as edited.
Most sites show you what they know. These are the places where there isn't enough evidence to publish a number — counted from the index itself rather than written by hand, so the list cannot go stale. Catalogue published 20 Oct 2026.
- 0 of 129 listings carry a review from a buyer who confirmed the purchase — 0%. The rest rely on the note comparison and the seller's own claim.
- 129 listings have no wearer rating yet, so their accuracy rests on the note comparison and the seller's claim alone.
- 8 of 9 sellers cannot be scored yet and read UNRATED.
- 6 of 27 originals have no indexed alternative.
- 1 review sitewide. Every threshold above waits on this number.
⚠️ These numbers are supposed to be bad. If they ever look good without the underlying evidence having grown, something is computing wrong. The correct reading is "the Bureau has work to do", never "the Bureau is failing".
It is not a guarantee of any individual order, and it is not a statement about a fragrance — only about the shop. The Bureau takes no payment and no commission on your purchase, holds no stock, and is not party to the transaction. If something goes wrong with an order, the Bureau can record it; it cannot resolve it.
A shop with no verification record on file is not marked, and no seller can pay for the mark. Which checks were run, and on what date, are shown on the seller's own page — browse the directory.