How much does employee theft cost a small shop?

The ACFE's Occupational Fraud 2026 report puts the median loss per case at 126,000 US dollars for organisations with fewer than 100 employees, higher than for any other size band. In sub-Saharan Africa the median across all organisation sizes is 97,000 dollars per case. These are losses per incident, not per year, and most cases run for months before anyone notices.

How long does till fraud usually run before it is detected?

Twelve months is the global median between a fraud starting and being found, according to the ACFE 2026 report. The gap widens sharply without a reporting channel: organisations with a formal mechanism found fraud in 11 months and lost a median of 100,000 dollars, while those without took 17 months and lost 150,000 dollars, half again as much.

Which POS settings catch cashier fraud fastest?

Four settings do most of the work. One login per person with no shared PIN, so every transaction has a name on it. Void and refund permissions restricted to a supervisor. A cap on the discount a cashier can apply without approval. And a blind cash declaration at shift end, where the cashier counts the drawer before seeing what the system expects. None of them costs money.

Does mobile money reduce till theft?

It reduces cash skimming and increases a different risk. Money paid into a merchant till is traceable and hard to pocket, which is the gain. The new exposure is staff steering customers to a personal number or a Pochi wallet instead of the business till, which looks like a completed sale to the customer and leaves no record in your books at all.

What is paper shrinkage and why does it matter?

Paper shrinkage is stock that your records say you hold but that never existed, usually from receiving errors, wrong unit counts or untracked wastage. Retrak figures reported by the Daily Nation attribute 16.2 per cent of Kenyan retail shrinkage to it. It matters because it hides real theft: if your stock records are wrong, a variance tells you nothing.

Employee theft at the till: the controls that actually stop it

Theft causes 72 per cent of stock shrinkage in Kenyan retail, and process errors another 16.2 per cent, according to Retrak figures reported by the Daily Nation. Look at what is left over for damage, expiry and everything else: under 12 per cent.

Most shop owners read that and picture a stranger putting a tin in a bag. The more expensive version of the problem is standing behind the counter, holding a login.

Where the money actually leaves

Five schemes account for most till and stock losses in an independent shop. Each one leaves a signature in your point of sale data, and each one has a control that catches it within a week rather than within a year.

Scheme	How it shows in your POS data	The control that catches it
Cash skimming (sale rung, cash pocketed)	Drawer short at close, small and frequent	Blind cash declaration at shift end, per cashier
Unrecorded sale (goods out, nothing rung)	Drawer balances, stock variance grows	Cycle counts on fast movers, weekly
Void and refund abuse	Voids and refunds cluster on one operator or one shift	Supervisor permission required for both
Discount and price override abuse	Same cashier applies discounts far above average	Hard ceiling per operator, override logged
Receiving fraud (short deliveries signed for)	Purchase value rises faster than sales	Three way match: order, delivery note, invoice

The pattern worth noticing: only the first scheme shows up in the cash drawer. The other four are invisible to a manager who reconciles cash every evening and calls that stock control.

The numbers small shops assume do not apply to them

Small businesses carry the highest median loss of any size band, not the lowest. The ACFE’s Occupational Fraud 2026 report puts the median at 126,000 US dollars per case for organisations with fewer than 100 employees, slightly above the figure for organisations with more than 10,000.

The regional picture is just as uncomfortable. South Africa, Nigeria and Kenya together accounted for 220 of the 397 sub-Saharan African fraud cases the ACFE analysed, or 55.4 per cent, with a regional median loss of 97,000 dollars per case (Ecofin Agency).

At sector scale, South African retail puts annual losses from crime and shrinkage at roughly 23 billion rand, with employee theft named alongside shoplifting and organised syndicates (Zawya). In the United Kingdom, the British Retail Consortium’s 2026 crime report puts the total cost of retail crime at 4.2 billion pounds for 2023/24, up from 3.3 billion, with 1.8 billion spent on prevention.

Kenyan chains have published the shape of it too. Nakumatt once measured 1.5 per cent of stock lost to shoplifting while estimating total shrinkage as high as 6 per cent of goods. On a shop turning over 40,000 dollars a month, six per cent is 2,400 dollars gone monthly, which is usually more than the rent.

The controls that cost nothing but a setting

You do not need cameras to start. You need four things switched on in software you probably already have.

One login per person, no shared PIN. This is the control everything else depends on. A till where three cashiers share a code has no audit trail, only a rumour. If your software does not support per user accounts on the same device, that is a reason to change software.

Voids and refunds behind a supervisor. A refund is a cash withdrawal with a receipt. Treating it as an ordinary cashier function is the single most common gap we see in shops that have just been defrauded.

A discount ceiling per operator. Ten per cent without asking, more with approval, every override written to a log that somebody actually opens on Monday morning.

Blind cash declarations. The cashier counts the drawer and enters the total before the system shows the expected figure. A cashier who can see the expected number will match it. This one change turns your shortage report from decoration into evidence.

Mobile money closes one hole and opens another

Money that lands in a merchant till is traceable, which removes most of the skimming opportunity. That is a real gain, and it is why shops that move volume onto a till number see drawer variances fall.

The new exposure is simpler than fraud: a staff member gives the customer a personal number or a Pochi la Biashara wallet instead of the business till. The customer pays, gets goods, walks out satisfied, and the sale never existed in your books. Nothing in your cash reconciliation will ever show it.

Two habits contain it. Print the till number or merchant code on the receipt and on a sign at the counter, so a customer paying elsewhere is visibly wrong. And reconcile the payment provider statement against POS payment lines daily, not monthly, which is the routine we set out in our M-Pesa merchant guide.

Receiving is where the bigger loss usually lives

Shops police the till and sign for deliveries without counting. That is backwards. A short delivery signed for is a loss you paid for, twice: once in cash, once in the stock record that now overstates what you hold.

Paper shrinkage is what makes the rest of your controls useless. When records are wrong for innocent reasons, a stock variance stops being a signal, and real theft hides inside the noise. Get receiving right first, then variances start meaning something.

Two routines are enough for most shops. A three way match on every delivery, with the person who counts different from the person who orders. And cycle counts on your top 20 lines every week, rather than a full stocktake once a year that nobody trusts by day three. Our guide to running more than one branch covers how this changes when stock moves between shops.

Detection: the cheapest control is a way to be told

Fraud runs for a median of 12 months before discovery, and the single biggest lever on that number is whether staff have a safe way to report what they see. The ACFE 2026 data is stark: organisations with a formal reporting mechanism lost a median of 100,000 dollars and found the fraud in 11 months, while those without lost 150,000 dollars and took 17 months.

Only 24 per cent of organisations with fewer than 100 employees have one. For a shop with eight staff, it does not need to be a hotline. A number that reaches the owner directly, and one visible instance of a report being handled without the reporter being punished, does the job.

Thirty days, in order

Create individual logins, delete shared codes, and tell staff why.
Move voids and refunds behind supervisor approval.
Set a discount ceiling and open the override log weekly.
Switch cash declarations to blind counts.
Split ordering from receiving, and count every delivery.
Start weekly cycle counts on your 20 fastest lines.
Reconcile mobile money statements to POS payment lines daily.

Where the software fits

Most of the list above is configuration, not purchase. digabloPos is free and runs offline, with per user accounts, permission controls on voids and refunds, real time stock, and mobile money recorded as its own payment line so that reconciliation takes minutes.

If you are weighing options, our POS comparison sets out which systems support per user permissions and which only pretend to. And if you have just found a shortage and need to know whether it is theft or process, tell us what you are seeing: the reports usually answer it within a day.