Cash keeps going missing at the counter

The day's cash does not match the day's sales. It is not a large gap, but it is there most weeks, and you have no way of knowing whether it is a mistake, a missing bill, or something worse.

Start from the most useful assumption: most discrepancies are process failures, not theft. Wrong change given in a rush, a bill settled and never marked as paid, a discount given verbally, a mis-keyed amount. If you begin by assuming theft, you will damage trust with people who did nothing wrong, and you still will not find the cause. Build a system that tells you the difference.

Why you currently cannot tell

Everyone shares one login. If the till is operated under one account all day, no transaction is attributable to a person. Without attribution there is no way to narrow anything down, so every discrepancy is a whole-restaurant mystery.

Cash is counted once, at the end of the day. A single count across three shifts and four people cannot be traced back to when the gap appeared.

Voids and discounts are not recorded as events. A bill voided after payment, or a discount given without approval, looks exactly like a shortfall — and looks exactly like the other thing too.

No sequence control on bills. If bills can be torn out, skipped, or written in a second book, the sales total itself is not reliable, so the comparison is meaningless from the start.

The controls that cost nothing

Count cash at every shift change, not once a day. Two people count, both sign the sheet, the float is reset to a fixed amount. This is the single highest-value change available to you: it turns "the day was short ₹800" into "the 4–10pm shift was short ₹800", which is a problem you can actually investigate.

Write down the float amount and keep it fixed. Variable floats make every count an argument about the starting point.

Make discount authority explicit. Who may give one, up to what value, and where it gets written. Most "missing" money in restaurants is discounts nobody recorded.

Handle voids as paperwork. A voided bill gets a reason and a signature. No exceptions, including for the owner.

Post the day's total where staff can see it. Transparency reduces opportunistic behaviour more effectively than suspicion does, and it makes honest errors easier for people to own up to.

What per-person accountability adds

The structural fix is that every transaction carries a name, without adding any friction to service.

PIN-based staff login in a restaurant POS gives each person a 4-digit code and a role. Every order, every bill, every void, and every discount is attributed automatically. Nobody has to write anything down, and a shift change takes three seconds.

That changes the investigation completely. Instead of a daily gap with no context, reports give you a payment-mode breakdown — cash versus UPI versus card — per shift, plus a log of who voided or discounted what. A ₹800 gap becomes traceable to a shift, a person, and usually a specific transaction.

In practice, what most owners find when they first get this data is not theft. It is one staff member who consistently mis-keys cash amounts and needs ten minutes of training, or a discount policy that everyone has been interpreting differently for a year.

If it does turn out to be theft

Deal with it on evidence, privately, and once. The value of the system above is that it gives you facts instead of suspicion — which protects the people who did nothing as much as it identifies anyone who did.

The reason to build these controls is not to catch someone. It is so that the question stops being unanswerable, and so that a shortfall in March can be explained in March rather than argued about all year.