Every one of these methods has the same shape: the money and the record are changed together, so the count at close comes out right. That's why owners so often discover a problem months or years late — not because they weren't checking, but because the check they were doing can't see it.
One thing worth saying up front: card payments largely self-police — the money goes straight to your bank and reconciles itself. Cash is where all of this lives. The higher your cash share, the more of your takings run through the methods below.
Why the count can't catch it
The nightly count compares two numbers that come from the same place: the till. If a sale was never rung — or was rung wrong — both sides of the equation are wrong by the same amount, and the count "balances".
Method 1 — The sale that was never rung
Take the cash, ring nothing
The oldest and simplest — and in a busy shop, the hardest to see.
A customer orders, pays £8 in cash, and the order is called straight to the kitchen. Nothing is ever entered into the till. The £8 goes into a pocket, an apron, or sits under the note tray until a quiet moment. As far as your till is concerned, that customer never existed.
How you might spot it: receipts never offered to customers; kitchen orders that outnumber till transactions; a shift where the shop looked busy but the till says it wasn't. Comparing kitchen tickets (or screen orders) to till transactions for one evening is a revealing exercise.
Method 2 — The under-ring
Ring less than the customer paid
Also called skimming the difference. Subtler than method 1, because every sale has a record.
A £9.50 meal is rung in as £4.50 — a smaller item, or one item off a multi-item order. The customer pays the real price and rarely checks the screen or the receipt. The extra £5 comes out later, because the drawer is now £5 richer than the till believes.
How you might spot it: one person's average sale value sits oddly below everyone else's on the same shift; the same cheap item sells suspiciously often; customers occasionally mention their receipt showed the wrong thing. Your end-of-day report usually lets you compare average sale by staff member — it's worth a look.
Method 3 — The void after payment
Ring it properly, void it after they leave
Everything looks right while the customer is standing there. The edit happens after.
The sale is rung correctly and £12 cash goes into the drawer. Once the customer is gone, the transaction is voided — "customer changed their mind", "rung by mistake". The till now expects £12 less, so £12 can leave the drawer and the count still comes out right. The paperwork shows a void, which is a normal thing that happens in every shop — just not this often.
How you might spot it: your till keeps a void log — most owners never open it. Look at voids per shift and per person, and when they happen. Voids clustered on one person's late shifts, or voids of cash sales specifically, deserve a closer look. Requiring a manager code for voids helps, but only if the code actually stays private.
Method 4 — The refund that never happened
Refund a sale nobody returned
The till's own paperwork is used to legitimise cash leaving the drawer.
A £7.50 refund is processed for an order that was never brought back — often at a quiet moment with no customer anywhere near the counter. The till records money out, the drawer goes down by the same amount, and everything matches. The refund is the theft; the record is the cover.
How you might spot it: refunds with no matching complaint, no returned food, and no customer in sight. Compare your refund log against reality: in a takeaway, genuine cash refunds are rare events. More than a couple a week from the same person is a pattern, not bad luck.
Method 5 — The discount applied after they've paid
Charge full price, record a discount
Staff discounts, promo codes and "mates rates" — applied to orders that paid in full.
The customer pays the full £10. On the till, a 50% staff discount, promo, or loyalty deduction is applied, so only £5 is recorded. The other £5 is now invisible surplus in the drawer, lifted at the next opportunity. Discount abuse also runs the other way — genuinely free food for friends — but the cash version is the one your count will never show.
How you might spot it: run the discount report by staff member — most tills have one. A discount rate that's double everyone else's isn't generosity. Tie every staff discount to a named person and cap it, and make promo buttons manager-only if your till allows it.
Method 6 — The over-skim
Build a surplus, then take it out
Short-changing customers, rounding "in the drawer's favour" — then a no-sale opening to collect.
Small overcharges and short change accumulate through a shift: £4 or £5 the drawer holds that the till doesn't know about. A "no sale" opening — the button that opens the drawer with no transaction — is the collection moment. The victims here are your customers as much as you, which makes it worse: it's your shop's name on the bad reviews.
How you might spot it: the no-sale count per shift is on most end-of-day reports. A few a day is normal (change for notes, a stuck receipt). Fifteen on one person's shift is not. Complaints about wrong change that cluster on particular shifts are the other tell.
The pattern behind all six
Notice what every method has in common: each one is invisible in the till's own records, or hides inside records that look routine — a void, a refund, a discount, a no-sale. Reports can show you where to look (and the tips above are genuinely worth doing), but a report line can never tell you whether that void at 21:14 was an honest mis-ring or a £12 exit. Only one thing can: seeing what actually happened at the counter at that moment.
That's the gap between the record and reality — and it's exactly the gap TillGuard was built to close. It cross-checks what your till recorded against what actually happened at the counter, and sends you only the moments where the two don't line up, each with a short clip. Most flags turn out to be honest mistakes; those cost you money too. And to be clear about what it is: it flags, it never accuses — every judgement about a person is yours.
Try TillGuard on one site — first quarter free
We set it up in under an hour using the cameras and till you already have. It runs quietly, and each month we go through exactly what it found — every flagged moment, with its clip. A clean month is a good result too: then you know, instead of assuming.
Message us on WhatsAppFirst five sites: three months free, founding rate locked for life · No long-term contract