The 11pm cash count doesn't match. Again. Not by a lot — Rs. 1,200, maybe Rs. 1,800 — small enough that you could talk yourself into "counting error" for the third week running. You don't say anything to anyone. You just start watching the counter more closely, and watching people is exhausting, and it still doesn't tell you what's actually happening.
This is the moment most restaurant owners in Pakistan hit before they do anything about staff theft — not a single dramatic incident, just a slow accumulation of small numbers that don't add up, and no system underneath the till to explain why.
Here's what actually works: not suspicion, not a stricter manager standing over the counter, but data the POS was already generating — you just weren't looking at it the right way.
Table of Contents
- The honest taxonomy of restaurant theft
- Void & discount logs
- Cash-vs-sales variance at shift close
- Per-staff sales tracking
- Recipe-level inventory variance
- Rider COD reconciliation
- What changes when staff know the system sees
- A fair-process note
- Staff chori — Roman Urdu Q&A
- FAQs
The honest taxonomy of restaurant theft
Before you can catch anything, it helps to know what you're actually looking for. Restaurant theft in Pakistan tends to fall into a handful of recognizable patterns, and each one leaves a different kind of trace:
- Unrung sales (kachi parchi) — an order taken, food served, payment collected, and no ticket ever entered into the system. The cash goes straight into a pocket instead of the drawer. No POS record exists, so this one leaves no direct trace in the system — it only shows up as a gap between what the kitchen actually cooked and what the register actually rang.
- Void abuse — a real sale is entered, then voided after the customer pays and leaves, with the cash pocketed and the sale erased from the books.
- Discount abuse — staff apply discounts, comps, or "manager specials" to paying customers and pocket the difference, or apply friends-and-family discounts that were never authorized.
- Over-portioning and ingredient walk-outs — kitchen staff serve oversized portions to friends, or ingredients leave through the back door without ever reaching a plate that gets billed.
- Rider COD leakage — a delivery rider collects cash on delivery and the amount handed back to the restaurant doesn't match what the customer was actually charged.
Each of these is invisible on its own, in a single instance. What makes them catchable is that every one of them leaves a pattern over time — and patterns are exactly what a POS system is good at surfacing, if you know where to look.
Void & discount logs
Every void and every discount should be tied to a name, a timestamp, and a reason — not just a number that disappears from the day's total. The question worth asking isn't "did staff void an order today" — voids happen for legitimate reasons constantly, wrong items rung in, kitchen mistakes, customer complaints. The question is who voids far more often than everyone else on the same shift, doing the same job.
Look at void logs by staff member, not by transaction. One or two voids a shift is normal. A specific cashier voiding six line items a week while colleagues on identical shifts void one or two is a pattern worth a direct, calm conversation — not an accusation, a conversation.
Discount logs work the same way. If your system lets any staff member apply a discount without a manager approval step, that's the first fix, before you even get to reading the logs — the log tells you where to look, but eliminating unauthorized discount authority in the first place removes a whole theft path.
Cash-vs-sales variance at shift close
This is the most basic check and the one most restaurants still do badly: what the POS says should be in the drawer versus what's physically counted at shift close, recorded shift by shift, not averaged across the week.
A single off-by-a-few-hundred night is noise — genuine counting mistakes happen, change gets miscounted, a customer gets shorted or overpaid by accident. What matters is the same cashier's shifts showing a consistent negative variance over two or three weeks, while other staff working equivalent shifts land close to zero. That consistency is the signal. One bad night tells you nothing. Ten shifts of small, one-directional variance from the same person tells you almost everything.
Per-staff sales tracking
Put two staff members on the same counter across similar shifts — same hours, same footfall, same average ticket size in the neighborhood — and look at their logged sales totals over a few weeks. If one consistently logs meaningfully lower sales than the other despite comparable shifts, that gap is either a scheduling and footfall difference worth checking, or it's unrung sales walking out the door in cash that never touched the system at all.
This only works if your POS actually attributes every sale to the staff member who rang it, per login, not per till. A shared login defeats this check completely — it's the single most common reason owners can't see this pattern even when it's happening in front of them.
Recipe-level inventory variance
This is where kitchen-side leakage shows up, and it's the check most restaurants skip because it takes real setup — recipe costing per dish, mapped to raw ingredient usage. Once it's set up, though, it's the hardest pattern to hide from.
The mechanics: for every dish sold, the system calculates theoretical ingredient usage based on the recipe — how much chicken, oil, rice should have been consumed to produce the plates actually sold. Compare that theoretical number against actual stock draw-down for the same period. A gap between theoretical and actual usage is variance, and consistent variance on the same ingredient, week after week, is either a recipe or portioning problem worth fixing operationally, or it's product leaving the kitchen without ever being billed. Both are worth knowing. Only one of them is theft.
Rider COD reconciliation
For delivery orders, the amount a rider hands back after a cash-on-delivery run should match the order total exactly, every time, reconciled per rider and per shift — not bundled into a single end-of-day lump sum that makes individual discrepancies invisible.
The mechanics here are simple in principle: every COD order has a system-recorded total. Every rider's end-of-shift handback gets checked against the sum of their COD deliveries for that shift. A rider running consistent shortfalls against their own delivery total — not a one-off missing Rs. 200, but a repeated pattern — is the same signal as a cashier with a recurring drawer variance. The fix is the same too: reconcile per person, per shift, not in aggregate.
What changes when staff know the system sees
Here's the part worth being honest about: most staff in most restaurants are not stealing. The value of these systems isn't primarily catching the small number who are — it's that visible, consistent tracking removes the opportunity and the temptation for the rest.
When voids need a reason and a manager sign-off, when every sale is tied to a named login, when shift variance gets reviewed as a matter of routine rather than only when something feels off — theft stops being a low-risk, low-effort option for anyone who might otherwise drift into it. That's deterrence, and it's a quieter, more permanent effect than catching any single incident after the fact.
It also protects honest staff. A cashier who's never had a bad variance night has a clean record the system can show, not just their own word — which matters when trust is ever questioned, in either direction.
A fair-process note
Patterns are not proof. A consistent variance, a high void rate, a sales gap — these are reasons to look closer, not reasons to accuse. Real explanations exist for almost every one of these signals: a cashier covering unusually chaotic shifts, a kitchen genuinely over-portioning by habit rather than by theft, a rider working a route with more customer disputes than most.
Before any conversation moves from "let's look at this together" to something with consequences attached, look at the data across enough shifts to rule out coincidence, talk to the staff member directly and give them the chance to explain what the numbers show, and check whether the same pattern shows up for anyone else in a similar role. Treating a pattern as a verdict before that process is how honest staff get pushed out and real problems stay hidden, because everyone learns to stop trusting the system that flagged them unfairly.
Staff chori — Roman Urdu Q&A
Staff cash chori kar raha hai — kaise pata karen? Variance report aur void log dekhein — kisi ek staff member ka shift-close variance baar baar negative aa raha ho, ya void rate baqi staff se zyada ho, to yeh ek pattern hai. Pattern 2-3 hafton mein saaf nazar aata hai — ek raat ka farq kuch nahi batata, lekin lagataar farq zaroor kuch batata hai.
Kya har cash mismatch chori hoti hai? Nahi. Chhoti, ek-baar ki galti aksar sirf counting mistake hoti hai. Jo cheez dekhni hai woh hai consistency — same staff member ka baar baar, same direction mein variance aana.
Kitchen se ingredients gayab ho rahe hain, kaise track karein? Recipe-level inventory variance dekhein — jo ingredient theoretically use hona chahiye tha (recipe ke hisab se) aur jo actually stock se nikla, dono ko compare karein. Consistent gap ek signal hai, chahe woh portioning issue ho ya leakage.
FAQs
What's the difference between unrung sales and void abuse? Unrung sales never enter the POS at all — an order is served and paid for in cash with no ticket created, so there's no direct system record, only a gap between kitchen output and rung sales. Void abuse is the opposite: the sale is entered correctly, then deliberately erased after the customer has already paid, leaving a void log entry that points straight at who did it.
How long does it take to see a real theft pattern in the data? Most of these checks — void logs, shift variance, per-staff sales — need two to three weeks of consistent data before a real pattern separates from normal day-to-day noise. Acting on a single bad shift or one high-void night usually means acting on coincidence, not evidence.
Can a restaurant catch theft without expensive systems? The core mechanics — daily variance counts, a written void reason log, per-shift rider reconciliation — can be done manually with a notebook and discipline. What a proper POS adds isn't the concept, it's consistency: automatic per-staff attribution, recipe-level variance calculation, and reports that don't rely on someone remembering to write things down every single night. See how this works inside a full system on the POS feature page and the inventory costing page.
What should an owner do before confronting a staff member about a suspected pattern? Pull the data across enough shifts to confirm it's a real pattern and not a coincidence, check whether the same signal shows up for other staff in comparable roles, and go into the conversation asking what happened rather than stating a conclusion. A pattern is a reason to investigate, not a verdict — treating it as one risks losing honest staff over an explainable gap.
Staff cash chori kar raha hai — kaise pata karen? Variance report aur void log dekhein — kisi ek staff member ka pattern baaqiyon se alag ho, to woh dekhne layak hai. Pattern 2-3 hafton mein saaf nazar aata hai, ek raat ka farq nahi.
See how void logs, per-staff attribution, and role permissions work inside a real system on the POS feature page, how recipe-level variance is tracked on the inventory costing page, and how COD gets reconciled per rider on the delivery management page. For the full platform, start at Kliovo Dine.