Key points
- Freeze stock movement for the duration of the count, or the result is a snapshot of two different moments.
- Count blind: the counter must not see the book quantity, or the book quantity becomes the count.
- Work in teams of two — one counts, one records — and never let the storekeeper count their own store alone.
- Investigate variances before posting any adjustment. An unexplained adjustment hides the problem it was created by.
- A count that nobody signed is not evidence, and cannot support a claim against anyone.
A stock count is only evidence if four conditions hold: movement was frozen while it happened, the people counting could not see the book quantity, every difference was traced to a document before it was adjusted, and someone with authority signed the result. A count that fails any one of those is a number, not a control.
What has to happen before the count?
Before counting starts, movement is frozen in writing, the store is physically arranged, every item has one code, and blind count sheets are printed and numbered. Most failed counts fail here rather than during the counting.
- Announce the freeze in writing, with a start and end time, and record the last receipt and issue number used before it.
- Separate what is not yours or not saleable: damaged goods, customer returns, consignment stock and items awaiting inspection.
- Resolve duplicate codes first. Two codes for one item guarantee a variance that no investigation can explain.
- Print numbered sheets by location, not by item list, so a counter walks the shelf rather than searching for it.
- Assign teams of two and make sure nobody counts the store they are responsible for.
How should the counting itself be run?
Counting is done blind, in teams of two, against numbered sheets, with a pre-agreed recount threshold. Blind counting is the single most important rule: the moment a counter can see the expected quantity, the expected quantity is what gets written down.
- One counts, one records. The recorder never counts and the counter never writes.
- Set the recount threshold before the count, not after seeing the results, and recount with a different team.
- Handle goods in transit explicitly: stock dispatched but not yet delivered, and stock received but not yet entered, are listed separately with their documents.
- Count partially opened packs by opening them. An assumed quantity inside a sealed carton is an assumption, not a count.
- Collect every sheet, including the unused ones, and account for all sheet numbers before anyone leaves.
What happens after the count?
Every difference is traced to a document before any adjustment is posted, and both the count and the adjustments are signed. Posting an adjustment first and investigating later removes the only evidence that would have explained it.
| Likely cause | What to look for | Fix |
|---|---|---|
| An issue that was never recorded | Goods gone with no issue note, or a note with no entry | Post the entry and make the note mandatory before dispatch |
| A transfer between stores counted once | Two stores whose differences are equal and opposite | Match the transfer document and correct both stores |
| Duplicate or wrong item code | One item appearing under two codes, one short and one over | Merge the codes and recount the affected items |
| Unrecorded damage or return | Goods physically set aside with no document | Document it, then adjust with the reason recorded |
How often should a count happen?
A full count belongs at every period you intend to close and report on, and a rolling count belongs in between. The purpose of the full count is a defensible closing figure; the purpose of the rolling count is to find the cause of a difference while the documents that explain it still exist.
Rolling counts — often called cycle counting — cover a small part of the store frequently instead of the whole store rarely. They are cheaper, they do not stop the business, and they surface a control problem in weeks rather than at the end of a year. Prioritise by value and by movement: the fast-moving, high-value items justify being counted most often.
What if the count reveals a large shortage?
Preserve the evidence before doing anything else. Freeze the affected store, keep the original count sheets, and do not post an adjustment while the difference is still being examined — an adjustment closes the accounting question and destroys the trail at the same time.
- Recount the affected items with a different team, in the presence of someone independent of the store.
- Reconstruct the movement history for those items from the documents: receipts, issues, transfers and returns.
- Separate what is explained by paperwork from what is not, and quantify each part.
- Report the unexplained portion in writing, with the evidence attached, before any adjustment is prepared.
- Only then post the adjustment, with the reason recorded, and record the control change that prevents the repeat.
The five mistakes that void a count
- Counting while the store is still receiving and issuing. The result then describes two different moments and cannot be reconciled to either.
- Giving the counter the book quantity. The count stops being independent evidence the moment it is anchored to the expectation.
- Letting the storekeeper count their own store alone. Even with complete integrity, it removes the separation the count exists to provide.
- Posting adjustments before investigating. The difference disappears from the books and the cause stays in the warehouse.
- Finishing without signatures. An unsigned count cannot support a claim, a insurance notification, or a conversation with a bank.
A count tells you the quantity. Only the documentary cycle tells you who is responsible for the difference.
Counting is half the job — valuing it is the other half
A counted quantity becomes a financial figure only once it is valued on a stated basis, applied consistently. Decide the basis before the count, write it down, and use the same basis in every period: the value of closing stock is simultaneously an asset on the balance sheet and the cost of sales for the period, so an inconsistent basis moves the profit figure without anyone touching the profit.
Two things are decided at the same time as the basis: how damaged and slow-moving items are treated, and which costs form part of the item’s value. Both are judgements, and both belong in a written policy rather than in the head of whoever prepared the schedule this time.
If the count keeps producing differences
Repeated unexplained differences are a documentary problem, not a counting problem. The Foundation Package builds the cycle that makes each movement traceable, and Past Years Cleanup closes periods whose stock figures were never established. If you are not sure which of the two you need, the seven-question self-check will point at one of them.