The small, constant cash variance: the signal nobody chases

In your hotel’s restaurant nobody lets a large shortage slide. The controller calls, the manager reviews tickets and the cashier explains. What does happen, every single day, is a small shortage: forty, sixty, eighty. It fits inside the tolerance, gets noted as “change error”, gets signed and forgotten. That is the one to chase, because a large shortage is an accident and a small, constant shortage is a method.

Large and isolated versus small and constant

A large, isolated shortage is an event. It has a date, a shift, a plausible explanation: a bill that went out with the change, a drop that was not recorded, a check closed wrong. It gets investigated in an hour and is almost always found. Nobody will repeat it because everybody heard about it.

A small, constant shortage is the opposite. It has no date because it has all of them. It triggers no call because it fits inside the tolerance you yourself defined. And it is not investigated because each one, seen alone, does not deserve it. The per-close tolerance was designed to absorb the normal noise of making change. The pattern lives inside that noise on purpose.

That is why we say nobody chases it. The hotel controller reviews closes one at a time, and one at a time they are all fine. The pattern only appears when you put thirty closes from the same cashier on a single sheet and read the variance column top to bottom.

How to read a pattern

To read a pattern you need every close stored by cashier, by revenue center and by shift, with its expected cash, its counted cash and its variance. If your hotel only keeps the daily total, there is nothing to read. With the history in hand, there are five things to look at, and all five fit in a simple table.

The noise of making change is symmetrical: sometimes you give too much, sometimes too little, and over a month it nearly cancels out. A pattern is not symmetrical. It always goes in the same direction, always below the line that triggers a review, and always follows someone or some place.

An illustrative example with numbers

The figures below are made up to show the calculation. They do not belong to any hotel. Imagine a hotel restaurant with a tolerance of 100 per close and two cashiers who worked a similar month.

Neither cashier triggered a single review all month. Both signed every one of their closes. And yet cashier B is noise: twelve times short and twelve times over, similar amounts, a net of 60 in a month. Cashier A is a pattern: twenty shortages, zero overages, always around 70, that is, always under the tolerance of 100.

Fourteen hundred a month does not look like much until you multiply it: 1,400 times twelve months is 16,800 a year, from a single drawer, without a single alarm ever going off. That is the cost of the pattern. And it is still not the most expensive part: the most expensive part is that the rest of the team already knows what the tolerance is and knows it is not reviewed.

Where a small, constant shortage usually comes from

Before thinking about bad faith it is worth going through the process causes, because they are the majority. A pattern of small shortages in a hotel restaurant almost always comes from one of these sources, and several have nothing to do with the cashier’s honesty.

• Small purchases paid from the drawer without a receipt: ice, limes, a charger. Cash goes out, nothing is recorded, and the close is short by exactly that.
• Card tips paid out to the server in cash from the drawer at the end of the shift, without recording the outflow.
• A float that was never topped up: the shift opens with 1,900 instead of 2,000 and nobody counted it with a witness.
• Approximate cash drops: the supervisor takes “about eight thousand” and the system records exactly 8,000.
• A discount applied after collecting in cash, without returning the difference to the guest or recording the refund.
• Rounding of change in favor of the drawer or the customer, which is only noise if it goes both ways.
• And yes, cash someone takes in amounts that fit inside the tolerance, precisely because they know the tolerance.

All of these sources have something in common: they are cash outflows that did not go through the system. The small, constant shortage is, almost always, the sum of legitimate movements nobody recorded plus some movement nobody should have made. The article on the places where consumption leaks in a hotel (The eight places where a hotel loses food and beverage revenue) walks through the leaks on the sales side; this one walks through the leaks on the drawer side.

The controls that cut the pattern

No single control cuts a pattern of small shortages, because the pattern is designed to pass the control you already have. What works is a short combination of rules, all simple, that together remove the space the pattern lives in.

1. A blind count at every close: the cashier enters what they counted before seeing the expected amount. Without this, the variance is always zero and there is no pattern to read.
2. One cashier per drawer and per login. If two people collect in the same drawer, the pattern has no name and cannot be followed.
3. Petty cash separate from the sales drawer. Every small purchase comes out of a separate fund with its receipt, never from the drawer.
4. Card tips paid out at close with a record per server, not as an informal outflow from the drawer.
5. Cash drops counted and entered with the exact amount, signed by the cashier and the supervisor.
6. A cumulative tolerance per cashier, in addition to the per-close tolerance. It is the rule that makes cashier A in the example visible.

Cumulative tolerance: the control almost nobody uses

A per-close tolerance of 100 tells the cashier exactly how much can go missing before anyone asks. Cumulative tolerance changes the question: it does not matter how much was short today, it matters how much has been short over the last thirty closes. If the sum of a cashier’s variances in a month exceeds, say, 300, or if more than half of their closes are negative, they get reviewed even if none went over 100.

With that rule, cashier A from the example shows up in the first week: five closes short by 70 add up to 350 and the line is already crossed. Cashier B never shows up, because their shortages and overages cancel out. The rule does not punish noise; it only punishes constant direction.

The best thing about cumulative tolerance is that it requires suspecting no one. It is a number the system calculates on its own and that the hotel controller reviews every Monday in a list of cashiers sorted by cumulative variance. Whoever is at the top gets reviewed first. What else the controller reviews each week is on the corresponding page (Controller).

How to talk to the cashier

A pattern is a fact, not an accusation. The conversation starts with the process causes: did you pay for ice from the drawer, did you take card tips in cash, did you count the float with someone? Most of the time the pattern is explained right there, and what gets corrected is the process, not the person.

If the process causes are ruled out, the next step is to rotate: that cashier to another center, another cashier to that drawer, for two weeks. If the pattern follows the person, you have your answer. If it stays with the drawer, the problem is in the place: a drop that is not recorded, a purchase paid from there, a float someone touches. In both cases the data led you to the answer without pointing at anyone too early.

What the system has to store

All of the above depends on the point of sale in your hotel’s restaurant keeping the history and showing it by cashier. This is the minimum list.

• Every close with cashier, revenue center, shift, float, drops, expected, counted, variance and the explanation written at the moment.
• History searchable by cashier for any date range, not only last night’s close.
• Cumulative variance per cashier over the last thirty closes, with the number of negative and positive closes.
• An automatic alert when a cashier crosses the cumulative tolerance, even if no single close crossed the individual tolerance.
• A record of cash outflows from the drawer by type: drop, tip paid out, refund, with the signature of whoever authorized it.

How that history and the cumulative tolerance are organized in the cash module is on the cash page (Cash and shift close).

What to do this week

1. Pull the last thirty days of closes for every cashier in the restaurant and the bar and put them in a single table with the variance column.
2. Count, per cashier, how many closes were negative, how many positive and what the sum is. Sort the list from largest to smallest cumulative shortage.
3. Set up petty cash today with its own fund for small purchases and forbid paying for anything from the sales drawer.
4. Define a monthly cumulative tolerance per cashier and announce it to the team. The rule works better when everyone knows it.
5. With the cashier at the top of the list, review float, tips and drops first. Only then, rotate drawers for two weeks.

Inn Restaurant stores every close by cashier and by revenue center in the hotel, calculates the cumulative variance per person and warns when someone crosses the line even if no individual close did. If you want to see that cashier list with your own shifts, book a fifteen-minute demo (contact).