How to cut theft and billing loss in a shop or restaurant

Loss comes from four directions and they need different answers: customer theft, internal theft, process error, and supplier or receiving shortfall. Owners tend to assume the first, and the money is very often in the third and fourth.

So measure before you install anything. Take a proper stock count, compare it against what the till says you sold, and look at where the gap sits — by category, by shift, by day of week. A gap that appears only on one shift is a very different problem from one spread evenly.

This is unglamorous and it is what makes everything afterwards worth doing. Without it you are buying cameras to watch a problem you have not located.

The pattern that costs small shops most is not a grab and a run. It is a customer engaging the owner in conversation — a question about a product, a note to be changed, a complaint — while goods leave the counter or a second person works the other side of the shop.

It works because it exploits politeness and single staffing, and because the owner's attention is genuinely occupied rather than merely distracted. It is also nearly invisible afterwards: nothing was forced, nobody ran, and the shortfall shows up in a stock count weeks later.

Controls that actually help: keep high-value, small items behind or beside the counter rather than at the front, never leave the till area with a customer at it, and put a mirror or camera view where the till operator can see the blind side. Where the shop is single-staffed, a rule that the counter is not left unattended matters more than any device.

This is also the case detection technology is genuinely good at, because the giveaway is a combination — an item moving while the attendant's attention is elsewhere — rather than a single suspicious act.

In food businesses the recurring loss is a table that clears and leaves without the bill passing through the till. Sometimes that is a customer walking out; more often it is an order taken verbally, served, and never entered.

Fix the process first. Every order goes into the system before it reaches the kitchen, the kitchen does not cook from a verbal instruction, and the table is not marked clear until the bill is settled. That one sequence removes most of the exposure without watching anyone.

Then look at the exceptions the system records: tables opened and closed with no payment, items removed after being sent to the kitchen, bills settled long after the table was cleared. These are cheap reports and almost nobody reads them.

Cameras close the remaining gap when they are watched in time — an alert while the party is still on the premises is actionable, and a recording found at closing is a note for your records.

Internal loss rarely looks like money leaving the drawer. It looks like a void on a completed sale, a refund with no returned item, a staff discount applied to a stranger's bill, or a no-sale opening of the till.

Every till system logs these. Review them weekly, grouped by staff member and by hour, and look for concentration rather than volume — the same person, or the last hour of a shift, is the signal.

Then make the control structural instead of investigative: require a manager code for voids and refunds above a threshold, print a receipt for every refund, and reconcile the drawer at shift change with the person handing over present.

Handle what you find carefully. A pattern in a report is a question, not a conclusion, and training gaps produce the same numbers as dishonesty far more often than owners expect.

Count often, and count the categories where the money is rather than everything. A full count once a year gives you an annual number you cannot act on; a weekly count of your top twenty lines tells you which week something changed.

Receiving is the most overlooked leak in the chain. Goods signed for without being counted, short deliveries accepted, and invoices paid against a delivery note nobody checked are pure loss that never involves theft at all. Have one person count in against the order, and make that a step rather than a formality.

Watch waste and breakage records too. A rising wastage figure is sometimes exactly what it says, and is sometimes where stock that left another way gets recorded.

Most small businesses already have cameras, and most of that footage is never watched. Reviewing it is a task nobody has time for, so it happens only after a loss is confirmed — by which point the recording is evidence for a claim rather than a chance to prevent anything.

What changes the value is timing. An alert while a situation is still happening lets someone walk over; the same clip found at closing does not. That is the difference between a camera that documents your losses and one that reduces them.

Placement matters as much as technology. Cover the till, the entrance, the stock room door and the receiving bay, at angles that show hands and faces rather than the tops of heads. A camera that cannot resolve what happened is not evidence, whatever its resolution claims.

And whatever you deploy, get the obligations right — visible notice, a stated purpose, a retention period, restricted access. Our guide on CCTV monitoring rules covers what that requires.

When something points at a member of staff, slow down. Confront early on thin evidence and you lose the ability to establish what happened, and you may create a dispute you cannot defend.

Preserve the records first — the till exceptions, the relevant clips, the stock counts — and check whether an innocent explanation fits, because frequently one does. Then follow a consistent process rather than an improvised one, and keep footage use inside the purpose you told staff about.

The goal is a business that loses less, not a case against a person. Most of the controls in this guide reduce loss whether or not anyone is stealing, which is exactly why they are worth putting in first.