Cycle counting instead of the annual shutdown
Counting everything once a year is the most disruptive and least useful way to find out what you have.
The annual stocktake is a tradition rather than a method. It closes the business or consumes a weekend, it is performed by tired people counting unfamiliar items, and it produces one accurate figure a year that is out of date within a fortnight.
Frequent interruptions add overhead that is easy to miss in process estimates. For a discussion of that effect, see this guide.
Counting a small number of lines every week produces better records, costs less in total, and finds problems while they are still traceable.
Count by value and movement, not alphabetically
Not every line deserves the same attention. The standard approach is to split stock into three bands by annual value, and count each band at a different frequency.
- The small number of lines that account for most of the value — count monthly.
- The middle band — quarterly.
- The long tail of low-value lines — once or twice a year.
Fast-moving lines are worth counting more often than their value alone suggests, because they have the most opportunities to go wrong.
Movement is a better predictor of error than value. A cheap item picked forty times a week drifts faster than an expensive one that sits.
For a broader reference on product and location identification, GS1 standards document widely used supply-chain identification standards.
Small batches, fixed slot
Ten to twenty lines takes twenty minutes. Put it in a fixed slot — first thing on a quiet morning — and it becomes routine rather than a project.
Fixed slots also solve the practical problem: counting is displaced by everything else unless it is somebody's scheduled task with a name against it.
Count blind
A counter who can see the expected figure will find that figure. This is not dishonesty; it is how attention works when a number is already on the sheet.
Blind counts — the sheet shows the item and the location, not the quantity — produce materially different results, and the difference is the drift you were trying to measure.
Recount before you investigate
A meaningful proportion of variances are counting errors. Recounting the discrepancies before raising anything saves the investigation and avoids adjusting a record that was correct.
Where a second count agrees with the first, the variance is real and worth tracing. Where it does not, the count was the problem and there is nothing to trace.
The cost comparison people never run
The annual stocktake is chosen because cycle counting looks like ongoing effort while the shutdown looks like a fixed cost.
Twenty minutes a week is roughly seventeen hours a year. A full count in most small businesses consumes considerably more than that once preparation, the count itself, reconciliation and the disruption are added up — and it produces a worse result.