RFID inventory system

A client counted their stock once a year the way most companies do: they closed, put five people on the floor with barcode scanners and printed sheets, and spent twenty hours reconciling what was found against what the system believed.

It was expensive, it was disruptive, and — because it was exhausting — it was not especially accurate.

Barcodes are the root of the problem. A barcode needs line of sight and one scan per item, which makes the count linear in the number of things being counted and puts a person in front of every one of them.

Anything that replaced it had to survive a warehouse rather than a demo: items stacked behind other items, tags that get damaged, and staff who would use the system a handful of times a year and could not be expected to remember a procedure.

RFID removes the line-of-sight requirement — a reader picks up many tags at once, from a distance, through packaging. The count stops being a sequence of scans and becomes a sweep of the space.

The software around it does the part that actually matters: taking the raw reads, resolving duplicates from tags seen more than once, reconciling the result against expected stock, and producing an exception report of what is missing, what is unexpected, and what is in the wrong place. Nobody reads a list of everything that was found. They read the list of what disagrees.

The count went from five people for twenty hours to one person for roughly three — about a 97% reduction in effort — and from an annual ordeal to something that can be run whenever it is useful.

It remains the clearest example I have of a result that came from changing the shape of the problem rather than optimising the existing solution.