Where wrong numbers come from
When a store's inventory numbers stop matching reality, the investigation usually looks in the wrong place. People check for theft, for mis-picks, for system errors. Those exist, but the largest single source of drift in a small store is normally goods-in: stock that was counted quickly, or not at all, on the day it arrived.
The reason is structural. Receiving happens under time pressure, often by whoever is nearest, frequently without a document to check against. The paperwork says one thing, the pallet says another, and nobody has the information to notice.
Everything downstream inherits that error. Your reorder points are calculated from it, your alerts fire against it, and your team stops trusting the numbers, which is the expensive part because an untrusted number gets checked manually forever.
The check that has to happen before the boxes open
One thing separates a useful receiving process from a pointless one: the person receiving has to know what was supposed to arrive.
Counting into a void produces a number with nothing to compare it to. Forty cartons feels correct. Only an expected quantity turns it into either a match or a discrepancy, and only a discrepancy triggers action.
So the first step is not counting. It is having the open purchase order to hand, with quantities and expected items on it, reachable by whoever is doing the receiving rather than only by the person who placed the order. That is the connection the purchase order process exists to create.
The twenty-minute sequence
For a typical delivery at a small store, this is the whole procedure.
- Before signing, check carton count against the delivery note and check for visible damage. Note any discrepancy on the carrier's paperwork before signing, because after signing your position is much weaker.
- Match the delivery note to your purchase order. Same reference, same items. If the note references an order you do not recognise, stop and ask.
- Count what is actually there, by item, against the PO quantities. Not a spot check on the first delivery from a supplier, and not a spot check on anything high value.
- Inspect a sample properly. Open cartons from different parts of the pallet, not just the top one. Check the variant, not just the product.
- Record the result against the PO reference: received in full, short, over, or damaged, with quantities.
- Update stock, and only then put it away.
- Tell somebody if anything did not match. This is the step that gets skipped and the one that matters most.
Twenty minutes for a normal delivery. Longer for a big one, and much shorter once it is habitual.
The four discrepancies and what each one means
Short delivery. Fewer units than ordered. Needs a supplier claim and, more urgently, a decision about whether the shortfall changes anything: an advertised launch, a customer waiting, a reorder that now needs bringing forward.
Over delivery. More than ordered. Easy to ignore and a mistake to ignore, because you will usually be invoiced for it. Decide whether you are keeping and paying, or returning.
Wrong item or variant. The most common error with suppliers who carry similar codes, and the one most likely to survive receiving unnoticed, because the carton says the right thing.
Damage. Photograph it before moving anything. Damage claims are frequently refused on the basis of insufficient evidence, and the evidence is free to collect in the first five minutes and impossible to collect later.
Who needs to know, and how fast
A discrepancy recorded and not communicated is only marginally better than one not noticed. Three groups usually need to hear.
Whoever placed the order, so they can raise it with the supplier while the delivery is recent and the supplier's own records are fresh. Same day, ideally.
Whoever handles invoices, so a short delivery does not get paid at full quantity. This is where a short delivery actually costs money, and it costs it quietly.
Whoever is waiting for the stock, if anyone is. Marketing with a campaign scheduled, support with a customer promised a date, the person who set a launch date around this pallet.
The third is the one that most often gets missed, because the receiver has no way of knowing who was waiting. That is a visibility problem, not a diligence one: if the fact that a campaign depends on this delivery lives only in the marketing calendar, the person on the loading bay cannot be expected to connect them.
The person who finds the discrepancy is rarely the person who knows what it breaks. That gap is the whole reason to say something out loud rather than just record it.
When receiving is somebody else's job
If a 3PL receives on your behalf, the process does not disappear, it becomes a reporting expectation. You want their receiving discrepancy report within a day, not at month end, and you want it to reach a person rather than an inbox.
The failure mode is specific: warehouses do report discrepancies, and those reports arrive as routine emails among many others, and nobody treats them as requiring action. A month later a count is wrong and nobody can explain it, and the explanation was in an email in March. More on setting that expectation in working with a 3PL.
Putting it away is part of receiving
The step after counting gets treated as unskilled and is where a second class of error is created. Stock that is received accurately and then put away carelessly produces exactly the same symptom: a number that does not match what a picker can find.
Three things matter, and none of them is elaborate.
One SKU, one location, unless you have a system that can handle more. The moment the same item lives in two places without the system knowing, your count is right and your picker is wrong, which is the most confusing failure to diagnose because the numbers all look fine.
New stock behind old. For anything with a date, a batch or a version, this is the difference between rotation happening and rotation being something you intended.
Similar variants apart. Adjacent bins holding two colours of the same product generate mis-picks indefinitely. Separating them is a five-minute change that removes a recurring error.
None of this needs a warehouse management system. It needs somebody to have decided it once and written it where the person putting stock away can read it.
Counting, and how often
Good receiving reduces the need for full stock counts but does not remove it. The practical middle ground for a small store is rolling counts: a handful of lines every week rather than everything twice a year.
Prioritise by value and by movement. Your fastest-moving and highest-value lines are where drift costs most and where errors compound quickest. Slow-moving low-value lines can wait.
Then, when a count disagrees, treat the disagreement as information rather than as a correction to apply. A line that is repeatedly short is telling you something about receiving, picking or shrinkage, and adjusting the number without asking which one throws the information away.
Peak, when this gets skipped
Receiving discipline collapses under pressure, which is unfortunate because peak deliveries are the ones you can least afford to get wrong. Stock arrives, everyone is busy, it gets put away to be counted later, and later does not happen.
Two mitigations. Decide in advance which deliveries get the full process regardless, usually anything high value or anything a campaign depends on. And schedule receiving as a named slot with a named person during peak weeks rather than letting it be interrupt-driven, because interrupt-driven work is the first thing to lose in a busy week. The wider version of that planning is in the BFCM preparation checklist.
What this looks like in Store Huddle
The receiving check is a task attached to the purchase order it belongs to, so whoever is on the bay opens one thing and sees what should be in the delivery. Recording the outcome closes the task, which means the record of what arrived sits with the record of what was ordered rather than in a separate book.
A discrepancy becomes a message in the purchasing room with the product attached and the people who need to know already in the room, which handles the hardest part: telling the person who was waiting for the stock without having to work out who that was.
The Fulfillment role gives the people doing goods-in the rooms they need without giving them everything else, and photographs of damage sit in the conversation where the claim gets discussed rather than on somebody's phone.
Common questions
Why do inventory numbers drift?
In most small stores the largest source is goods-in: stock counted quickly, or not at all, without a document to check against. Everything downstream inherits the error, including reorder points and alert thresholds.
What should we check before signing for a delivery?
Carton count against the delivery note and any visible damage, noted on the carrier's paperwork before you sign. Your position on a damage claim is much weaker after a clean signature.
Should we count every delivery in full?
Count in full for a new supplier, for high-value stock, and for anything a campaign or customer is waiting on. Elsewhere a sample is reasonable, provided it is taken from different parts of the pallet rather than the top carton.
Who needs to be told about a discrepancy?
Whoever placed the order, so they can claim while it is recent; whoever handles invoices, so a short delivery is not paid in full; and anyone waiting on the stock. The last is missed most often because the receiver cannot know who that is.
How often should we count stock?
Rolling counts of a few lines weekly work better than everything twice a year. Prioritise fast-moving and high-value lines, and treat a disagreement as information about a process rather than just a number to correct.