Variance isn't bad. Unexplained variance is bad.

Ask most hospitality operators what they want from a stock take and they will tell you: no variance.

That is the wrong goal.

A stock take that consistently shows zero variance is not a sign of a tight operation. It is often a sign that either the stock take is not being done properly, or the business is too vanilla to have any movement worth measuring. In a real operation, with real deliveries, real staff, real service, variance is inevitable. The question is not whether it exists. The question is whether you can explain it.

Explained variance is manageable. Unexplained variance is a leak you cannot fix because you cannot locate it.

What variance actually tells you

Variance is the gap between what you expected to have and what you actually counted. That gap is not a verdict. It is a prompt. It is asking you a question. The answer to that question is where the value is.

Here are the categories that variance tends to fall into, and what each one should make you do:

Supplier shortfall. You ordered twelve cases. Eleven were delivered. The driver was in a rush, no one checked, and now you are short a case of wine with no paper trail. Variance here is a signal to raise a credit note immediately. Not next week, not when you remember. If you do not catch it at the point of delivery, you will pay for stock you never received.

Team error. Portions are running large. The bar is over-pouring. The kitchen is not weighing. This variance is consistent and directional: the same product, trending short week on week. That is a training conversation, not a disciplinary one. The variance is doing you a favour by making it visible before it compounds.

Theft. This is the one everyone thinks about first and the one that is least commonly the cause. But it does happen. What makes theft-related variance distinctive is that it tends to be specific: one product, one outlet, one period. Broad variance across a site is rarely theft. Targeted, unexplained drops in a single category warrant a closer look.

Recipe drift. The kitchen has been improvising. A dish has changed without the costs changing with it. Garnish portions have crept up. A substitute ingredient costs 20% more than the original. This variance lives between your theoretical GP and your actual GP. That is why recipe costing exists: not as an admin exercise, but as an operational check.

Legitimate business variation. Your best table of the year ordered a bottle off the back of the menu that moves twice a year. A private event ran long and the bar did twice the normal volume of a specific spirit. These numbers will look like variance until you understand the night. Context is everything here.

The problem with variance as a number

Most hospitality businesses produce a variance number. Far fewer produce a variance story.

If you print out your stock take results on a Friday morning and they show you that spirits are 4% short, what do you do with that? You can note it. You can worry about it. You can argue about it in a management meeting. But you cannot act on it, because 4% short could mean any of the things listed above, and the response to each one is completely different.

A variance number without context is noise. Variance with context is a management tool.

The context you need comes from two places: what happened during service, and when the variance was first detected. Understanding what unexplained variance costs annually makes the effort of building that context feel proportionate.

Cash-up and the service picture

Cash-up is where the financial picture of a shift comes together. And the operators who get the most from it are the ones who are not entering data the morning after. They are reviewing data that was captured in real time.

When your POS syncs automatically into your end-of-day reconciliation, you are not guessing what happened last night. You know what was sold, by category, by payment type. The cash-up becomes a review, not a reconstruction.

That matters for variance because it means you can set last night's stock movement against what the till actually shows. If spirits are down and the bar was quiet, that is a different conversation than if spirits are down and you had a hen party at the back who were ordering doubles all night. Same variance number. Completely different meaning.

The shift from dreading cash-up to understanding service through it is the shift from reactive to operational. It takes minutes, not an hour, and the picture you get is reliable because you are reviewing, not guessing.

Stock takes and the moment of catch

The other half of the variance picture is when the count happens. This matters more than people realise.

A stock take done on paper, transferred to a spreadsheet, reviewed three weeks later is not a management tool. It is archaeology. By the time you find the variance, the shift that caused it is a memory, the staff involved have worked thirty more shifts, and there is nothing actionable left.

The same stock take done on a mobile app, with counts reviewed on desktop immediately after, is a completely different process. The variance is caught at the point of count. If a shelf looks short, you check it now. Not next Tuesday.

This is also where the staff conversation changes. Teams who understand why stock takes are done, not as a gotcha exercise but as a way to catch supplier errors, flag training needs, and protect the business, engage differently with the process. The question "why do we do this?" stops feeling rhetorical when the answer is "because last month we found a supplier had been short-delivering two cases of lager a week for six weeks, and we recovered that credit."

That is the moment stock takes stop being a chore and start being a tool the team trusts.

Variance is information. Treat it that way.

Hops has supported over 15,000 stock takes across its operator base. The pattern is consistent: operators who approach variance with curiosity rather than dread find problems faster, fix them at the root, and protect more margin over time.

The goal of a stock take is not zero variance. The goal is explainable variance, and a process tight enough that anything you cannot explain gets investigated rather than absorbed.

That is what operational control actually looks like: not a perfect number, but a clear picture.

If your current process is producing variance numbers without the context to act on them, Hops was built for exactly that.