A butcher: lot traceability scanned at point of sale, not reconstructed
A family butcher in North Yorkshire was maintaining traceability compliance on paper and goodwill. The owner's son, in post two years, could see the problem but not the solution.
The reading
Three generations in the business, a shop front and a wholesale arm, five butchers behind the counter. The firm bought from six regional farms and supplied around forty restaurants. Traceability compliance was handled by the owner's father, in his seventies, who kept a cloth-bound ledger in the office and transcribed deliveries into it in pencil.
The ledger worked, in the sense that the firm had never failed an inspection. It worked less well in the sense that a specific query from a restaurant customer about which farm a particular joint had come from took, on average, forty minutes to answer.
The son, in post two years, had suggested a system. His father had asked him what was wrong with the ledger. The son could not answer that question in a way his father accepted.
The ledger that worked
The ledger worked because the father was in it every day. The traceability chain existed in his head, backed by the pencil entries. When he was out of the shop, the chain was effectively unreachable. When he eventually retired, which was being discussed, the chain would need to be reconstructed from the ledger alone, which was possible but slow.
We thought the son was right but had been arguing the wrong case. The ledger was not inadequate for compliance; it was inadequate as a succession plan.
- 6 supplier farms, ~40 wholesale customers, 1 shop front
- Paper ledger as single source of traceability
- 40-min average response to a traceability query
- Full dependence on the father's active presence
- No link between the ledger and the till
What we built
We put a label printer at the goods-in bench and a scanner at each till. Every incoming carcass or box received a lot label on arrival, generated from a short entry the father (or anyone) made on an iPad: farm, date, cut category. At the till, the butcher scanned the lot before weighing. The sale was linked to the lot automatically.
The technical choice worth noting: we kept the father's ledger. The system printed, each evening, a page in the same format he had always used, which he signed and filed. He continued to feel that he was running the traceability system, because in a meaningful sense he still was. The system did the recording; he did the oversight.
Stack: an on-premise mini-server in the back office, a Postgres database, a thin web UI, Zebra label printers, Bluetooth scanners.
The numbers at ninety days
| What | Before | After |
|---|---|---|
| Traceability query response time | 40 min | 15 sec |
| Ledger entries made by one person | 100% | 100% signed, 20% written by hand |
| Lot-to-sale links | Reconstructable | Automatic |
| Inspection preparation time | 1 day | 20 min |
| Succession risk | High | Low |
The system respects the person
The son's original mistake had been to propose a system that replaced his father. The system we built included his father, deliberately, because he was in fact doing a useful thing and the firm valued it.
The one-week training test applies: a new butcher scans the lot at the till on day one and the compliance happens as a by-product. But the deeper test, which matters in family businesses, is whether the system respects the authority of the people already there. We think, in this case, it does.