Shrinkage: finding out where the stock actually went

Shrinkage is the difference between the stock you should have and the stock you do have. It is normal for it to exist; it is not normal to not know why.

Five causes, in the order you should investigate them.

1. Administrative error — usually the largest

Not loss at all. Data problems that look like loss:

• A delivery entered in the wrong unit — a carton of 24 entered as one
• A product sold under the wrong code, so one item's stock is short and another's is long
• A price or cost entered wrongly, distorting valuation
• A transfer out recorded without the matching transfer in
• Duplicate product records splitting one item's stock across two rows

The check: look for pairs. An item short by 12 and a similar item long by 12 is almost always a scanning or coding error, not two separate thefts.

Fix this category first. In most shops it is the majority of the variance, and every hour spent investigating theft before ruling out data error is an hour wasted — and risks an unfair accusation.

2. Supplier shortfall

You were invoiced for 50 and delivered 46. Without receiving against a purchase order, this is invisible.

The check: compare ordered, received and invoiced quantities for the last month. A supplier who is consistently short is a commercial conversation, not a mystery.

The fix: receive against orders, and check the delivery at the door. Seepurchase orders.

3. Damage and expiry not recorded

Broken, expired and spoiled stock that left the shelf without a record.

The check: is there a write-off process at all, and is it used? A shop with zero recorded waste has a recording problem, not a zero-waste operation.

The fix: an easy write-off path with a reason list. If recording waste is harder than quietly binning it, it will be binned.

4. Customer theft

Real, and concentrated in small, high-value, easily-concealed items.

The check: rank your variance by value per unit. If the top of the list is cosmetics, razor blades, phone accessories and batteries, that is an external-theft profile.

The fix: placement, sight lines, security tags on the top ten items. Not a system change.

5. Internal theft

The last thing to conclude, not the first.

The check — patterns, not incidents:

• Variance concentrated in one branch, one shift or one person's hours
• Refunds and credit notes with no corresponding stock return
• Voids clustered at the end of a shift
• Discounts concentrated on one employee
• Stock adjustments approved by the same person who counted

The fix: controls first. Seepreventing till theft. Controls that make theft difficult are better for everyone than investigations after the fact — including for the honest staff, who are otherwise all under suspicion.

Measuring it

Shrinkage as a percentage: (expected stock value − actual stock value) ÷ sales for the period.

What matters is not the absolute number but thetrend and theconcentration. A steady 1.5% that is evenly spread is a cost of doing business. A 1.5% that is entirely in one category, one branch or one shift is a problem with a location.

The weekly report to look at

Five lines, five minutes:

1. Variance value this week, by category
2. Refunds and credit notes by staff member
3. Voids by staff member and time of day
4. Stock adjustments and who approved them
5. Waste recorded, by reason

None of these proves anything alone. Together, over a few weeks, they point at the cause — and usually the cause is number 1 on this page, not number 5.

Wameed's anomaly detection surfaces exactly these patterns; seeWameed AI.