Cycle counting works when it is small, frequent and unpredictable
Annual counts are a compliance ritual with management pretensions. They close the warehouse, exhaust the team, and produce one large adjustment that says nothing about when or where the drift began. Cycle counting replaces that with a rota: a short list of locations counted every working day, chosen by value and movement, with the fastest turning lines coming round most often. The count is small enough to be done properly and frequent enough that a variance is days old when it is found rather than months old.
Two design choices decide whether it works at all. The counter should not see the expected quantity, because a number on a screen is a powerful suggestion to a tired person at the end of a shift. Blind counts, with a recount on any variance outside tolerance, cost a little more time and produce data worth acting on. Second, a variance has to trigger an investigation rather than only an adjustment. An adjustment posted with no cause teaches a warehouse that the numbers are approximate.
Causes are usually few and repetitive. A location shared by two similar products. A case pack quantity entered wrongly on the supplier record. A pick shortage resolved on the floor without any system transaction. A customer return put away before it was ever received. Each is fixable once named, and naming requires a reason field somebody is expected to complete. Accuracy improves in steps rather than smoothly, and every step follows one of those causes finally being closed off.
- Daily rota weighted by value and movement, not by aisle order
- Blind counts, with a recount on any variance outside tolerance
- A cause recorded before the adjustment is posted
- Accuracy trended by location and product group over time
- Repeat offending locations escalated rather than simply recounted