Guide
Your stock count is wrong. Here is where it went.
Shop owners blame theft. In most shops the largest single cause is goods received in a different unit from the one they are sold in.
2 min read
Six causes, ranked by how much they usually cost
| Cause | Typically | Detectable by |
|---|---|---|
| Unit mismatch at receiving | Largest, and invisible | Comparing purchase unit to selling unit on any line |
| Wrong item billed | Steady, small | Two similar lines where one always overstates |
| Returns not booked back | Spikes after sale periods | Return count against stock adjustments |
| Damage and expiry | Predictable by category | Whether write-offs are ever recorded at all |
| Internal use and samples | Small, chronic | Nothing. This one needs a rule, not a report. |
| Theft | Real, usually not first | Pattern by shift once the others are eliminated |
The order matters because most shops investigate the last one first. A camera does not fix a case of twelve booked in as one unit, and that error repeats on every delivery.
Why the annual count is the wrong tool
An annual stock-take tells you the size of a problem roughly eleven months after it started. By then nobody can attribute it to a delivery, a period or a person, so nothing is learned and nothing changes. Next year produces a similar number and a similar shrug.
Counting a rotating slice weekly costs less total time and surfaces a discrepancy while its cause is still recent enough to find.
Start with value, not volume
In most shops a small fraction of lines carry the majority of stock value. Those are the ones worth counting weekly. A fast-moving two-rupee item can be counted monthly or quarterly and the error will not matter.
Chasing perfect accuracy on cheap fast movers is the most common way this discipline collapses. It generates work, finds small differences, and exhausts everyone before it reaches the lines where the money is.
Write down why
A write-off with no reason is a mystery. A write-off with a reason becomes a pattern after about six weeks, and patterns are actionable.
Four or five codes is enough. Damaged, expired, sample, internal use, unexplained. If unexplained is growing while the others are flat, that is when to start looking at people.
What to actually measure
Not the discrepancy. Discrepancy will never be zero and a shop chasing zero is wasting effort.
Measure how long it takes to notice one. If that is falling, the discipline is working, whatever the underlying number says.
Questions about anything here, or a situation this does not cover? contact@anantatechhub.com