Jewelry stock control: UHF for the count, NFC for the one-on-one

01 / FIELD NOTE

Keep the decision tied to the operating context.

The closing count is where jewelry retail shows its real operating problem. The stock is small, valuable, handled constantly, and spread across showcases, trays, safes, repair benches, photography corners and transfer boxes, and the record drifts with every move nobody recorded. A missing ring is not the only failure mode — an item in the wrong showcase, a watch at repair that still reads as available, a transfer sent but never confirmed at the receiving store. Each is a small gap between the physical world and the file, and the question the system answers is which of these gaps the count can make visible in time to act.

The items themselves are why the line-of-sight method struggles. A ring has no flat face for a label the way a carton does; its certificate, stone details and tray position belong to it alone; and scanning one by one through a showcase interrupts exactly the moments of the day when pressure is highest — handover, closing, the end of a busy session. The radio method changes the unit of work: the tray becomes the unit, and one scan reads every tagged piece in it. The inventory stops being a line of individual scans and becomes a group identification followed by an exception list, which is a different kind of work for a different level of attention.

The tag has to disappear into the item’s presentation, and that constraint dominates the form factor. A ring is usually tagged on its display medium or its packaging rather than on the ring itself; a watch carries a small label on the clasp or buckle where it stays out of the piece’s profile; necklaces and bracelets take hang tags or labels that hang clear of the metal; boxed pieces carry a label on the inner packaging. The rule is that the tag has to accompany the item through its display and handling routine, because a tag that is removed for every showing is a tag that will not be there for the count.

The two radio technologies answer different questions, and they are not interchangeable for this application. UHF reads a field at range and in bulk, so it answers the tray count, the showcase sweep and the transfer box in a single pass — which is the daily control question. NFC is a close-range, one-item protocol: the reader must be within a few centimetres and it addresses a single tag at a time, which makes it right for authentication, certificate access, or a deliberate interaction with one piece. Both can live in the same program of tags because the workflow is matched to the method rather than the other way round.

The rollout does not need fixed infrastructure to start. A handheld reader plus tagged trays answers the daily questions — opening count, closing count, transfer verification, repair intake — from the moment the first batch is bound, and it does so without touching the store’s layout. Fixed installations belong at specific control points once they are warranted: a vault or secure-room door that should record movement, a smart cabinet that can report its own contents, a counter where every transfer is reconciled. The sequence matters: prove the count and the exception list on a handheld first, then add a fixed point where a controllable boundary actually exists.

The system’s value lives in what the count is allowed to distinguish, because not every mismatch is the same event. An item found in the wrong showcase is a location change to record, not an alarm. An item absent from every expected area becomes a search with a last known location. An item in repair that still appears available is a status error in the file, and a transfer sent but not confirmed is a movement gap at the receiving end. Reading the tags is the easy part; deciding which of these each discrepancy is, before the software is configured, is what keeps the closing count from turning into a list of unsolved mysteries.

The identity on the tag stays minimal, and the detail lives in the inventory system — the piece’s code, certificate number, stone and metal particulars, tray home, store location, stock status. The reason is practical: a certificate number or a status can change, and changing data on a chip means touching the tag, while changing a record touches nothing physical. The binding discipline is where the record is made or broken: the tag identity and the item row have to be linked once, correctly, because every later read and every later exception inherits that first link. Two items sharing one row is a failure the reader cannot see, and the audit eventually pays for it.

The failures in this application are rarely technical. They are decisions made before the tags go on: a label chosen for the desk rather than for the tray, a workflow that adds steps to a closing routine nobody has time for, exception categories left ambiguous so every discrepancy lands in one undifferentiated list. The system survives where it fits the store’s existing rhythm — opening, closing, transfer, repair intake — and weakens where it does not. The honest limit follows: a count confirms presence and position in the zones the reader reaches; it does not replace security measures, and between counts the record stands on staff discipline, which is a property of the store, not of the hardware.