Cycle counting: count a little stock often instead of everything once a year
Why we'd rather count a few shelves every week than the whole warehouse once a year, and how to put a dollar figure on what the counts turn up.
Say your year-end count comes up three cases short on 1g mylar bags. You can write them off, but you can't tell whether they walked out in May or got mis-picked in October, and by now nobody remembers much about either month.
The trouble with counting once a year
The annual wall-to-wall count goes about the same way everywhere. You stop the floor, print the sheets, count everything and book one big adjustment at the end. You get a number out of it, but not when any of the stock went missing. By then there's no shift to ask and no delivery left to recheck, and whatever habit caused the gap has been going the whole time, so about all you can do is book the write-off.
Smaller counts, more often
Cycle counting swaps the big event for a routine. You take a short list and count it properly: one location, maybe one category, twenty items on a Tuesday morning. Each count is small enough that the floor keeps working, and because you count often, a variance points at the last few weeks, which is recent enough to track down.
A few habits that help:
- Count your fast movers most often. They get touched the most, so they pile up errors fastest.
- Go back to wherever you last found a variance. The cause is often still there (a pick face that's easy to confuse, a step one shift skips), so count it again next month.
- Book the correction as soon as each line is counted rather than at the end, so the system agrees with a shelf from the moment someone has counted it.
Put a dollar figure on the gap
What you want from a count is the variance: the gap between what the system says and what's on the shelf, in dollars. That figure shouldn't just sit on a closed count sheet. It should be checked against a threshold, and somebody should hear about it when it's over.
That applies when stock comes up over, too. An overage means the book was wrong, which means every forecast and reorder that used it was off as well. If the same items keep coming up over, check whether receipts are being entered twice.
Counting in the app
A count here covers one location and, if you like, one category, so it can stay Tuesday-sized. It's a blind count: the counting screen hides the system quantity for a line until that line has been counted, so nobody anchors on the number they expect to see. Counting a line books the correction straight away as a stock movement labeled with the count's number, so the shelf and the system match from then on. The system quantity each line was counted against is kept, which means the report shows the variance you found and not a zero after the fix.
If you're not sure what to count next, the counts page suggests a list for each location from the items that are due. Items the book was recently wrong about come first (someone tried to take more off the shelf than the system said was there), then anything that's never been counted, then whatever has the most money sitting on the shelf.
When you close a count, the variance gets priced: a report and a CSV with each line's expected, counted and dollar variance, plus the shrinkage rate on what was counted. If the net variance is bigger than the alert threshold, you get an alert that says whether stock is missing or over and names the line with the biggest gap, so you know where to start looking.