Counting stock without closing the shop
Most small businesses count stock once, find the numbers are wrong everywhere, and quietly never do it again. The problem is almost always the method, not the shop.
Published
Do not count everything
A full count needs the shop shut, and a shop that shuts to count will not count often. Counting rarely is worse than counting partially, because the longer the gap the less you can tell about what caused a variance.
Count a section at a time on a rolling basis: the fast movers weekly, everything else over a month or a quarter. You will find problems while they are still small enough to explain.
Count what is there, not what should be there
Whoever counts should not be able to see the expected figure. Given a number to match, people match it — not dishonestly, but because the eye stops looking once it agrees.
For the same reason the person who counts should not be the only person who can adjust. That single separation catches most of what a stock count is for.
Freeze the section while you count it
If stock moves mid-count, the variance you find is arithmetic rather than information.
Either count a section when it is not selling, or record every sale from it during the count and adjust. Whichever you choose, write down the time the count started — a variance with no timestamp cannot be investigated.
Read the variance instead of just fixing it
The adjustment takes a second. The reason is the whole point, and a till that makes you pick a reason is doing you a favour even when it is irritating.
- Short in a fast mover, repeatedly: usually theft, or a receiving process nobody follows.
- Short across a whole supplier: check what was delivered against what was invoiced.
- Long on something: nearly always a receiving error or a product that exists twice in the system.
- Short on breakables and short-dated goods: expected, and it should be written off as waste rather than left as a mystery.
Fix the data before blaming the people
Duplicate products, a unit that is sometimes a bottle and sometimes a case, and a barcode shared by two items will each produce a permanent unexplainable variance.
Clean those up first. A count against bad data is a count against nothing.
General information, not legal, tax or financial advice. Where this page describes a legal requirement it reflects the legislation as commonly applied and legislation changes — confirm anything that matters with SARS or your own accountant.