Staff Theft in Hospitality - Part 3

How You Actually Catch It

I've lost count of how many times an operator has told me "my team wouldn't do that" or "I'd trust them with my life." It's meant as loyalty. It's actually the most dangerous sentence in hospitality.

Temptation isn't just reserved for people you already have concerns about.

Cash and stock sit in front of everyone, every shift, regardless of how long they've worked for you or how much you trust them. And in my experience, it's rarely the person you'd have suspected. It's the one you'd never have looked at twice, right up until they get caught. Not because they started dishonest, but because it worked, and then it worked again, and at some point greed pushes it further than it's ever gone before, until it's no longer subtle enough to miss.

I've told this story to more than one operator who's said those words to me: a daughter, taking money from her own father's safe while he slept upstairs. If you can't rule that out, you can't rule anyone out. I'm not suggesting that everybody's friends and family are fleecing them, but the above demonstrates that blind faith is the most expensive thing in your stockroom.

Controls aren't an insult to your team. They're what protects the ones who'd never take a penny from being the only ones anyone can prove didn't.

Trust isn't a control. Here's what actually is.

Start with the pattern, not the person

The mistake most operators make is looking for a single moment of proof, a smoking gun on camera or a till roll that doesn't add up on one specific night. That's not usually how it shows up. What shows up first is a pattern, sitting quietly in numbers nobody's compared side by side.

Void and refund rate, per person

Not an aggregate figure for the venue. Per team member, set against their own transaction volume. Someone with a disproportionate share of voids or refunds relative to how much they're actually serving is the first flag - not proof of anything on its own, but enough to start asking questions.

Average transaction value, same comparison

The vanilla syrup trick, and anything like it, depresses this number specifically. A real order, a pot of tea for two and cakes, gets replaced on the record by something cheap. Anyone whose average check value sits noticeably below colleagues working the same section, same shift pattern, over time, is worth investigating.

Cash taken versus peers on comparable shifts

This controls for the obvious pushback: "my shift's just quieter."

Compare like-for-like. Same day type, same time slot, same footfall pattern.

If one person's cash figures consistently sit apart from everyone else working the same conditions, that's not the shift. That's the person.

Dipping sales, rising tips

A specific pattern worth naming on its own: recorded sales for a team member drop while their personal cash tips climb. Customers are still paying, still tipping normally, the sale just isn't landing on the till roll the way it should.

Watch for till surpluses, as well as shortages

A skimmed cash sale rarely creates a till shortage because the sale was never rung through in the first place. Once the cash is removed, the till balances perfectly. The only evidence is a temporary surplus while the money is still in the drawer.

That's why end-of-shift reconciliations often miss this type of theft. A mid-shift till check is far more effective because it can catch the overage before it disappears.

Reluctance to take time off

Employees who never take holiday, resist shift changes or avoid covering different locations deserve a closer look. Many theft schemes rely on one person being present to keep them running. The moment someone else takes over, the pattern breaks and the issue becomes visible. Sometimes the strongest warning sign isn't what's happening when they're at work - it's their determination not to be away from it.

The shift-move test

This is the sharpest tool of the lot, because it isolates the person as a variable rather than just watching a number. If a pattern of shortages or high voids is clustering around a particular time slot, move the person in question to a different day or shift and watch what happens.

If the pattern follows them to the new shift, that's not the shift. That's the person, and you now have evidence that isolates them specifically, rather than a correlation that could be explained a dozen other ways.

If the pattern doesn't move, and stays with the original shift regardless of who's working it, you've learned something different and just as useful: you've probably been looking at noise, or at a structural issue with that time slot rather than an individual. Either result is progress. That's what makes this a genuine investigative method rather than a checklist item - It's designed to produce an answer, not just a suspicion.

Isolate the till once suspicion emerges.

If transaction analysis and shift comparisons consistently point towards one individual, the next step is till isolation. Assign them their own cash drawer for several shifts. Shared tills muddy the data and allow losses to be spread across multiple employees. A dedicated till creates accountability and provides much clearer evidence of whether a problem exists.

On the stock side, the same principle applies but in a different way

Everything above works from till data. For stock walking out the door, or anything that never touches a transaction at all, the equivalent tool is a bookend count.

Book-end stock counts

What's there last thing at night gets counted. Then counted again first thing in the morning, before anyone new comes on shift. If a gap only ever opens up overnight, you've isolated the window it's happening in, and from there, who had access during it.

An end-of-service count also gives you same-day data — losses get flagged and acted on the day they happen, not weeks later. It's a deterrent too: your team knows the checks are constant, and that changes behaviour on its own.

For anything ongoing rather than overnight-specific, reconcile consumption against till sales daily, not just at a periodic stocktake. A daily gap gets caught the same day it opens. A gap that waits for a monthly or quarterly stocktake to surface has had weeks to compound, and by the time it's found, there's no way to isolate which shifts it happened on.

Par-level counting for packaged and bottled stock

This only works if the principle from part two is already in place: backup stock in the cellar, spirits store, or bottle store locked, with no access to team during the shift other than management. Once that's true, front-of-house fridges and shelves get stocked to a fixed, round number rather than whatever's convenient. Six of everything. Eight. Ten. Whatever suits the volume, as long as it's consistent.

The reason the number matters is what it does to the counting. Instead of counting all your stock at the end of a shift, you're counting the gap on the shelf or fridge.

Ten bottles stocked, seven left in the fridge, that's three sold, and three is what should be sitting on the sales report for that line.

The same principle works for anything packaged and countable: crisps and snacks stocked to ten per flavour, four flavours, forty packets at the start of service. Thirty left at the end of the night means ten should be on the sales report.

What makes this worth running alongside the bookend count rather than instead of it: it's fast enough to run every single shift, not just once a day.

The detection of stock losses at the time sends a strong signal to your team that you're all over your stock and you don't miss a thing.

A full stocktake on packaged goods takes real time and usually only happens periodically. Counting a gap against a fixed par level takes minutes, which means variance gets caught the same night it happens, rather than surfacing weeks later in a stock count nobody can trace back to a specific shift.

What this actually requires

None of this depends on catching someone in the act, and none of it depends on trusting your own instinct about who "seems" honest.

It depends on comparing numbers that are already sitting in your system — person against person, shift against shift — and letting the pattern do the work your gut can't be relied on to do.

That's not distrust. It's the only method that doesn't rely on trust at all, which means it's the only one that protects the team members who've never taken a penny just as effectively as it exposes the ones who have.

This is part of the three part series. Visit the other blogs here...

Part 1 - The Official Numbers Say It Isn't Happening... It Is.

Part 2 - The Mechanisms.