The system says 42, the shelf has 31. In most retail businesses the gap comes from one of three process failures, not from theft or bad counting.
Every retailer we onboard has the same first conversation. The system says one number, the shelf says another, and after a while nobody trusts either. The instinct is to blame staff, or the previous system, or theft.
In practice, across the deployments we have done, the gap almost always traces back to one of three process failures. None of them are counting errors. All of them are timing errors.
The carton is on the shelf. The delivery driver has left. The paperwork is in a folder by the door, and nobody has entered it into the system.
This is the single most common cause, and it is worst in businesses where receiving and selling are done by different people on different shifts. The goods are physically available, so they get sold — and the system, which never knew they arrived, records a negative or zero stock position that it then silently corrects with an adjustment nobody reads.
Two controls fix it:
A customer brings back a shirt. The refund is issued, the money leaves the till correctly, everyone is happy. The shirt goes back on the shelf.
But if the refund was processed as a negative sale rather than a return against the original invoice, stock never comes back into the system. The item exists twice in the physical world (on the shelf) and zero times in the system. At stocktake you find more than you expected, which feels like a pleasant surprise until you realise it means the opposite happened somewhere else.
The control is unglamorous: returns must reference an original document. That single rule keeps refund value, stock movement and margin adjustment attached to the same transaction. It also gives you a return rate by product, which is genuinely useful information you otherwise do not have.
Stock left store A on Tuesday. It was signed for at store B on Wednesday. Only store A recorded the transfer.
Now store A is short by a quantity nobody can explain, and store B has stock in the system it does not physically have — which is worse, because it means store B will keep selling something it cannot fulfil.
The control is a two-step transfer: dispatch, then receive. Between those two steps the stock sits in a clearly labelled in-transit state that belongs to neither store. It is visible in reporting, it is excluded from both stores' available quantity, and it has an owner and a timestamp.
Why in-transit matters more than it sounds: without it, every transfer is a coin flip on which store's numbers are wrong. With it, an unreceived transfer shows up as an open item on a report — and open items get chased.
Two SKUs for the same item. Created at different times, by different people, both selling, one holding the real quantity. This is common after a system migration or a catalogue expansion.
Unlike the first three, this one is fixed by cleaning rather than by process. The practical test is simple: pull your top 200 products by sales value and check for near-duplicate names. It takes an hour and usually finds five to fifteen duplicates.
Do that for a month and your count will not be perfect, but the gap will have a name — and a named problem can be managed. An unnamed one just gets written off every quarter.
Twice a year is typical for a single store, with cycle counting of one category per week in between. Full counts are expensive and disruptive; regular small counts find problems while they are still small.
Under 2% by value is a reasonable target for general retail; under 1% for high-value categories. Above 5% you have a process problem, not a counting problem.
It reduces entry errors but not timing errors. Most of the causes above survive barcoding entirely, because they are about documents not being raised, not about codes being mistyped.
Send a sample of your stock report. We will tell you which of these is happening, on your data.
Request a free trial WhatsApp us sales@gdyun.online · Reply within one business day · Support 09:00–21:00 GMT+8