Count a little, all the time
The annual wall-to-wall count tells you in December what went missing in March, when the trail is nine months cold. Small, frequent counts catch the same shrink while the cause can still be found - and they turn the variance from a number you report into one you can act on.
Every warehouse counts. The question is whether you count once a year, and learn in December what happened in March - or a little at a time, and learn while the trail is still warm.
Why the big count fails quietly
The annual wall-to-wall count is a ritual: shut the floor, print the sheets, count everything, book one giant adjustment. It produces a number, and the number is nearly useless, for one reason - it has no when in it.
A discrepancy found nine months after it happened has no shift to ask, no delivery to re-check, no habit still running that anyone can watch. You book the write-off and move on, none the wiser. Do it again next year.
The alternative is smaller, not harder
Cycle counting replaces the ritual with a habit: a short list, counted properly, often. One location, one category, twenty items on a Tuesday. Because each count is small, the floor never stops; because counts are frequent, a variance points at a window of weeks rather than a year - and at a cause you can still find.
- Count what moves most, most often - fast movers accumulate errors fastest.
- Count where the last variance was found - errors cluster around habits.
- Book the correction the moment a line is counted, not at the end. A shelf somebody has counted should be right from the second they trusted it.
The variance is the real product
The point of counting was never the counting. It is the variance: the gap between the book and the shelf, priced in money. That number deserves better than a cell in a spreadsheet - it should be compared against a threshold somebody chose, and it should interrupt somebody when it crosses it. In either direction, because a count that comes up over is not good news: it means the book was wrong, and every forecast and reorder built on the book was wrong with it. Stock that is consistently over usually means receipts are being entered twice.
What this looks like here
A count here is scoped to a location, and optionally to a category - a Tuesday-sized list, not a ceremony. Counting a line books the correction immediately, as a stock movement with the count's name on it, so the shelf and the book agree from that second - and the book value each line was counted against is kept, so the report shows the real variance rather than the post-correction zero.
When the count closes, the variance is priced: a report and a CSV with every line's expected, counted and dollar variance, plus the shrinkage rate on what was counted. And if the net crosses the dollar threshold you set, it raises an alert that names the worst line - because "where do I look first" is always the next question.