You sell a product that your system says you have 14 of. A customer wants six. You go to the shelf and find two. Now you're apologising, refunding, or scrambling to reorder something you thought was covered. If that's happened to you more than once this month, your inventory isn't broken because of one big mistake. It's death by a hundred small ones.
Where the numbers actually go wrong
Stock records drift for boring, everyday reasons. None of them feel like a big deal in the moment, which is exactly why they add up.
- Receiving shortcuts. A delivery of 48 arrives, someone eyeballs it and marks the PO as received in full. It was actually 44. That gap is now baked into your system.
- Sales that skip the system. A staff member sells something quickly, promises to ring it up later, and forgets. The cash is in the drawer but the stock count never moved.
- Breakage and shrinkage. A bottle smashes, a box gets water damaged, something walks out the door. If nobody writes it off, the system still thinks it's on the shelf.
- Wrong SKU on a similar item. Two products look almost identical. The 500ml gets scanned as the 750ml. Both counts are now off, in opposite directions.
- Returns handled badly. A customer returns an item and it goes back on the shelf, but the return was never logged, or it was logged and the item went straight in the bin.
Any one of these on its own is a rounding error. Do a hundred transactions a day and the drift compounds fast. Within a few weeks, the report you make decisions from is fiction.
Why the once-a-year stocktake doesn't save you
Plenty of businesses count everything once a year, usually with the doors shut and everyone exhausted. It gives you one accurate snapshot on one day. The problem is what happens the other 364 days. You spend most of the year working off numbers you know are wrong, and the annual count just tells you how wrong after the fact. It also doesn't teach you anything. You correct the total, but you never find out which products keep going missing or which supplier keeps shorting you.
Cycle counting: count a little, all the time
The fix is to count small chunks of your inventory on a rolling schedule so every item gets checked several times a year, and you never shut down to do it. This is called cycle counting, and it's the single most useful inventory habit you can build.
The trick is to count more often the more the item matters. Not everything deserves equal attention.
- Split your items into three groups. Your top sellers and highest-value stock (the roughly 20% of products that drive most of your sales) go in group A. Steady mid-range items go in B. Slow movers and cheap bits go in C.
- Count A items monthly, B items quarterly, C items twice a year. A small shop might count 15 to 20 SKUs each morning before opening. It takes ten minutes and it fits into the day.
- Count against the system, not from memory. Pull the expected quantity for those SKUs, physically count them, and record the difference. Your ERP should let you enter a count and post the adjustment in one step.
- Log a reason for every discrepancy. Damage, theft, miscount, receiving error, wrong scan. This is the part people skip, and it's the part that actually pays off.
The reasons matter more than the corrections
After a month of logging reasons, patterns show up. Maybe one supplier accounts for most of your receiving gaps, so you start counting their deliveries properly. Maybe a specific product has constant shrinkage, so you move it behind the counter. Maybe two SKUs keep getting mixed up, so you relabel the shelf. You're not just fixing the count, you're removing the cause.
A good target for accuracy is 97% or better on your A items. If you're sitting at 80%, don't panic. Just start counting and watch the number climb over a few months.
Getting your system to carry the load
Cycle counting only works if the tools make it quick. In VISIONS you can build count sheets by category or location, scan items with a handheld or phone, and post adjustments straight to your stock and accounting records so your cost of goods stays right. You can also pull a variance report that shows exactly which items drifted and by how much, ranked worst first, so your morning count targets itself.
Start tomorrow. Pick your 20 best-selling products, count them before you open, and enter the results. Do it again with the next 20 the day after. Within a couple of weeks you'll trust your numbers again, and you'll stop promising customers stock you don't have.