What the scan does

Twice a week, Monday and Thursday morning, an automated scan works through your catalog and checks products for signs a house is retiring them: whether the house still lists the product, whether the major retailers still carry it, and what the fragrance community is saying in forums, resale listings and social posts. Anything suspicious arrives here with the reasoning and the sources behind it.

Confirmed against rumored

Rumored is early chatter worth watching. Confirmed means several independent sources agree. Once an item is confirmed, the scan also checks what it is reselling for elsewhere, so you can read the premium before deciding whether to buy in.

How the priority score is built

Every flag carries a score out of 100 so the list arrives in the order worth working. It is arithmetic, not opinion, and every flag will show you its own breakdown under "why this rank".

• Scan status, up to 30. Confirmed scores well above rumored.
• Independent sources, up to 16. Four points each, counted to four.
• Kinds of signal, up to 10. A web signal plus a pricing signal is stronger than two of the same kind.
• Resale demand, up to 14. Whether the market check found it selling on, and whether that check reads as rising demand.
• Your own clearance pricing, 5. A compare at price sitting above your current price often means the item is already being run down.
• Value of the bottle, up to 25. Once most of the list is confirmed, certainty stops telling two items apart and money starts, so this carries real weight.

Your decision, separate from the scan

The pill on the right is what the scan concluded. The four buttons underneath are yours, and they answer two different questions.

• Accept and Reject are your verdict on the scan itself. Accept means yes, this really is going. Reject means the scan has it wrong, and the item dims so it stops competing for your attention.
• Need to buy and Ordered are what you are doing about it. Both open a quantity box, and the app works out what that commits at retail and what it makes at cost.

No button lit means nobody has looked at it yet, and the filter above has a Not reviewed option so you can work through exactly those. Clicking a button that is already on clears it again. Notes are for anything the scan cannot know: a call with the rep, an allocation, a promise.

What the money figures mean

Unit value is your own retail price for the largest bottle of that item, and the size it refers to is printed beside it. Committed value is that price multiplied by the quantity you entered. These are retail figures. Connect a token in settings and your wholesale cost comes across too, which turns the same figures into margin.

The figures do not know how many bottles you already hold, because this pilot reads an uploaded export rather than live stock. Connecting Shopify is what closes that gap.

Where your decisions live

Two stores, kept deliberately apart. GitHub holds what the scan produces, the flag list and the catalog, and it rewrites that file twice a week. Supabase holds what you decide: Accept, Reject, Need to buy, Ordered, quantities and notes.

Splitting them is the point. When both lived in the same file, Monday morning scan could overwrite Friday afternoon decisions. Now the scan cannot touch your side of it. Decisions save the moment you click, on every device, with no token and no setup, and the masthead says syncing when that is working. If the connection drops, the app keeps working, saves to the device, and pushes everything up the next time it can reach the cloud.

Notifications

You get one email per new flag, with confirmed and rumored listed separately. Items already flagged do not come back to your inbox, they simply stay on this page.

Checking supplier stock

Tick the items you care about, then Check supplier stock and upload one or more current supplier price lists. Matching is by barcode where the file has one, by house and name where it does not, and every match tells you which of the two it used. The report exports to PDF or Excel.

Coverage, stated plainly

Searching all 4,825 products individually twice a week is not practical. Each run covers a rotating slice of roughly 150 products, which brings the whole catalog round in about a month, and it rechecks everything already flagged as rumored on every single run. So a confirmed item is never more than a few days stale, while a quiet corner of the catalog is on a monthly cycle. Connecting live Shopify sales data is what would let the scan look at your fastest moving products first instead of working alphabetically through the shelf.

Your data

Nothing here has touched the live store. The catalog comes from an export you uploaded. Wholesale cost is never published with the site, it only appears once a token is connected in this browser.