The weekly count that catches what the annual count misses
A once-a-year stocktake tells you how much you lost and nothing about when, how or who. Counting a little every week costs less in total, interrupts nothing, and is the only version that catches anything.
Every figure below is hypothetical. The pattern is not.
Most Nigerian SMEs count stock once a year, usually in December, usually badly, usually with everybody in the building and the business closed. At the end of it somebody announces a shrinkage figure, everyone is briefly upset, and the number is written off.
That exercise costs two days of trading and tells you almost nothing. It gives you one data point, twelve months wide. You learn that ₦2.4 million of stock is unaccounted for. You do not learn when it went, which items, or under whose shift — so you cannot act, and next December it happens again.
Cycle counting
Count a small slice of your stock every week, on a rota, without closing anything.
Roughly forty items a week gets a 2,000-line warehouse fully counted twice a year, and every count is a fresh comparison against a recent baseline. Nothing shuts. Nobody works a weekend.
The change is not really about effort. It is that a variance found within a week can still be explained, and a variance found in December cannot. The delivery note is still in the folder. The person who signed it is still on shift. The CCTV has not overwritten. You are investigating an event, not an accounting result.
Count the valuable things more often
Not every line deserves equal attention. Sort by annual value — quantity moved times unit cost — and you will find the usual shape:
| Class | Share of lines | Share of value | Count |
|---|---|---|---|
| A | ~20% | ~80% | Monthly |
| B | ~30% | ~15% | Quarterly |
| C | ~50% | ~5% | Twice a year |
Your A items are where the money and the risk are. Counting bolts as carefully as compressors is how counting programmes die of boredom.
Count blind
This is the rule that decides whether the whole exercise is worth anything, and almost everyone gets it wrong.
The count sheet must not show the expected quantity.
Give somebody a sheet that says "expected: 240" and, at the end of a long afternoon, a stack that looks roughly like 240 becomes 240. Not from dishonesty — from the entirely human tendency to stop counting once you reach the number you were told to expect. The variance you were hunting for is erased by the sheet itself.
Blind sheets list the item and a blank. The system does the comparison afterwards.
Two more rules that cost nothing:
- Rotate who counts what. A person who counts the same aisle every week is the only person who knows what is in it. That is a control weakness regardless of anyone's honesty.
- Recount variances before you investigate. Most first-count discrepancies are counting errors. Confirm it is real before it becomes an accusation — the same discipline as when your dip doesn't match your books.
What variances are telling you
Shrinkage is a category, not a diagnosis. The pattern tells you which kind you have.
Spread evenly across everything, small. Measurement and recording noise. Tighten receiving discipline, not security.
Concentrated in a few high-value lines. Now you are interested. Note that theft rarely looks like a missing pallet — it looks like a steady 2% on the one item that is easy to carry and easy to resell.
Always in the same period or the same shift. The pattern is the finding. Do not announce it; just widen the count frequency on those lines and see whether it persists.
Big positive variances. More stock than the books say is not good news. It usually means a delivery went unrecorded, which means an invoice may be unraised — that is money you never billed, and it is more common than theft.
The number to track
Not naira. A percentage:
shrinkage rate = value of net variance / value of stock issuedAround 1% is typical for a general SME warehouse; well-run operations get under half of that. Establish your own baseline over three months, then watch for movement. The baseline is the point. Without one you cannot tell a bad month from a bad year, and every conversation about stock loss stays an argument about feelings.
What this actually buys you
Three things, in order of how much they are worth:
- Losses get caught while they are still explainable — days old rather than months.
- Your stock figure becomes usable for decisions. Reordering, pricing, and knowing whether you can promise a delivery all depend on a number you currently do not quite believe.
- Nobody gets accused on a hunch. A team that knows counting is routine, rotating and blind is a team where the honest majority is protected — and where the alternative, an atmosphere of vague suspicion every December, quietly costs you good staff.