Retail Shrinkage: Split the Number to Fix It

Your annual count lands $6,000 light and three people have a stockroom key. Before you have that conversation, pull the receiving log. Two boxes of twelve earrings were receipted; one box came off the truck. That leaves twelve units the system still carries as sellable stock, $480 at cost, and nobody flagged the short delivery. That's retail shrinkage, and it's rarely just one thing.

You're the one who explains the number when it lands in the stockroom, and "someone's stealing" is tempting because it's the hardest explanation to disprove. A weekly count of your highest-value SKUs finds the cause, instead of one count a year. A year-old variance has no witnesses.

What is retail shrinkage, and how do you find where it's going?

Retail shrinkage is the gap between what your inventory system says you should have and what you can physically count, valued at cost. It comes from external theft, internal theft, receiving and vendor errors, and paperwork failures such as damages, returns never restocked, and staff purchases never written off. Find where it's going with a weekly count of your highest-value SKUs, so a variance dates to a narrow window while whoever touched that stock still remembers it.

Compute the rate honestly, and pick one basis

Shrink rate is dollars of missing inventory at cost, divided by net sales or cost of goods sold, over the same period. Pick one basis and hold it — mixing retail and cost dollars produces numbers that were never comparable.

Use cost of goods sold as the denominator, since a markup change shouldn't move your shrink rate — but if you divide by sales, it will.

Shrink rate = (units missing × unit cost) ÷ cost of goods sold for the period, × 100

The retail industry's average sits around 1.6% of sales, per the National Retail Federation's 2023 National Retail Security Survey (fiscal 2022) — a benchmark rather than a target, and measured against sales rather than COGS. Well under or over that average says more about receiving and register discipline than about your neighborhood.

Build the count list: value first, pocketability second

Rank every SKU by extended cost: unit cost times units on hand. The ABC inventory convention — which sorts by annual usage value — puts roughly the top fifth of items at around 70% of the total, with no fixed threshold. On-hand value is a close-enough proxy for a shop your size. Rank yours; that's your core count list.

Add a second, overlapping list: small, expensive, easy-to-pocket items — fine jewelry, premium skincare, a $60 accessory the size of a coin. Value alone misses what walks out in a pocket; pocketability alone misses receiving errors like the earrings above.

Your five highest-value SKUs sit outside the rotation and get counted every week, no exceptions. The rest of your core list rotates separately at ten a week: a 40-item pool comes around in a month, a 60-item pool in six weeks. Work out which of those you're running and tell the team the interval, so nobody assumes an item was checked more recently than it was.

Past a few hundred SKUs the rotation stops working as a diagnostic: the top fifth of a 2,000-SKU shop is 400 items, a forty-week loop at ten a week, by which point the trail is as cold as the annual count. Keep the core list tight and let the long tail wait.

Four rules keep the number trustworthy once you're counting:

• Count before doors open or after close. A sale mid-count changes the number you're freezing.
• Count blind. Give the counter the SKU and description only, never the system's expected quantity.
• Recount on any variance. A second person recounts anything off by more than one unit before it's logged.
• No receiving or selling touches the counted stock during the window. A mid-count delivery or a shift lead grabbing a unit produces a phantom variance that's just timing.

Run it the same day and hour every week, so a variance dates to last Tuesday, not sometime in the last year.

Read the shape of the variance

A sharp negative on one high-value SKU, especially one on your pocketable list, points at theft (customer or staff) or a single bad receiving entry. Check the receiving log first; it's the cheaper explanation to check.

Small negatives spread evenly across a category rarely mean a dozen separate thefts. They usually mean damages, spoilage, staff purchases never written off, or a shelf-tag mismatch ringing sales under the wrong SKU.

A positive variance — more units physically on the shelf than the system claims — is a signal too, and it traces the same way: something received but never logged, or a sale rung against the wrong item repeatedly.

Watch for matched pairs: a negative on one SKU next to an equal positive on a similar-looking one is a barcode mix-up at the register. Someone's been scanning the $18 candle under the $22 candle's lookalike code for weeks, and the two numbers won't match until the label gets fixed.

Check the cheap causes before the expensive ones

Before anyone reaches for a camera, check these three causes in order, starting with the one that takes the least time:

• Receipted without counting what arrived. Check the packing slip against what landed on the shelf, not just the quantity on the purchase order.
• Cases counted as single units. The system logged one case as one unit, or the reverse — open the case and count the pieces; it's invisible otherwise.
• Damages and staff purchases never written off. A cracked bottle goes in the trash without an adjustment, a staff purchase gets paid off-register in cash — both look identical to theft on a count sheet.

Short deliveries and case-pack errors often turn up on rush orders, where a part-shipment gets signed for under time pressure — see setting a reorder point around your supplier's lead time.

Only once you've checked the packing slip, the case count and the write-off log does theft become the leading explanation. Even then, a frequent count tells you which week and which SKU — more useful than an accusation after a count you can't trace back six months.

If it's cash rather than stock going missing, that's a companion problem: see why a short till is almost never theft.

Write it off the same week, or you'll chase it twice

Once you've traced a variance, record it. A damaged unit thrown out without a write-off still sits on your books as inventory you don't have, so next month's count repeats the loss. Log every damage, staff purchase, theft and receiving correction with a reason, the same week you find it. Any system with a stock ledger will hold those adjustments; VoVi keeps one and sends low-stock alerts, on every plan, at $0 a month for the software.

Whatever you run, look for a system that logs adjustments against a reason code rather than a raw quantity change, so you can pull every write-off from the last 90 days and catch damages climbing before they become next quarter's shrink number.

If a count turns up stock that's accurate but not moving, that's a markdown problem rather than a shrink problem: see clearing dead stock in one cut.

That $6,000 stops being frightening once you can trace it to a case, a slip, or a specific week on the calendar.

Common questions

What size variance is worth investigating?

Investigate when the loss is worth more than the time it takes to trace it. Weigh it by unit cost rather than unit count — a single missing $200 watch matters more than six missing $3 keychains — and work down the week's variances until the next one is worth less than the twenty minutes it takes to pull a receiving log. Always investigate anything on your pocketable list, whatever the size; that's where a mislabeled tag shows up first.

Should staff know which SKUs are on the weekly count list?

Yes — knowing which SKUs are on this week's list doesn't reveal the expected quantity, so the count-blind rule holds. The real worry is that a posted schedule also tells anyone inclined to take something which items aren't being watched, but your five highest-value SKUs sit on the list every week, so there's no off-week to wait for on the items worth taking. Keep the expected number private and the schedule can go on the wall.

Do I still need the annual count if I'm counting weekly?

Yes. The annual count is what your accountant uses for tax basis and what an insurer wants for a valued inventory on a claim — it has to cover everything you own, not just your top SKUs. The weekly count is a diagnostic tool that catches problems while you can still trace them; it doesn't replace the once-a-year count of everything on the floor.