How to do a stock take: a practical guide for hospitality operators

A stock take done properly takes less time than most people think and produces more useful information than most people realise. A stock take done badly — rushed, estimated, disconnected from the financial picture — takes roughly the same amount of time and tells you almost nothing you can act on.

This guide covers how to do it properly.

Before you start: what you need

A consistent counting structure. Your stock should be organised in a way that matches your count sheet or app, so the person counting moves through the space in a logical order without backtracking. If your storage areas are not organised, spend ten minutes before the count making them so. Reorganising mid-count multiplies errors.

A clear baseline. Your opening stock for the period (the closing stock from your last count) should be in the system before you start. Without a baseline, you have a list of quantities. With it, you have variance: the difference between what you were expected to have and what you actually have.

A confirmed period of purchases. Everything purchased since the last count needs to be recorded before you run the variance. If an invoice has not been entered, the system will show you short-delivered when you are not. This is the most common cause of unexplained variance that turns out not to be variance at all.

The right people. Ideally, two people: one counting, one recording. This reduces transcription errors and keeps the pace steady. For larger operations, divide the space into sections with one person responsible for each. If the count is being done solo, work methodically and do not rush.

The count itself

Start at the same point every time. Whether that is the dry store, the cellar, or the walk-in, consistency matters. Starting at the same point means sections do not get missed and the count can be picked up and resumed at any point without confusion.

Count what is there, not what you expect to be there. This sounds obvious but it is where most counts go wrong. If you know you usually have four cases of a product and you can see approximately four cases, the temptation is to write four and move on. Do not. Count the cases. It takes five seconds and is the only way the number is meaningful.

Record in the unit the system uses. If the system records wine in bottles, count in bottles. If it records spirits in centilitres, measure and record in centilitres. Unit inconsistency is one of the most common sources of systematic error in stock takes: a cellar that counts in bottles when the system expects cases will produce a variance every single time, and it is invisible until someone checks the unit settings.

Note anything unusual. A partial case that is being used across two locations. Product that has been opened and partially consumed. Deliveries that arrived after the last count and have not been entered yet. These notes are what turn an unexplained variance into an explained one.

Do not skip sections because "we always have plenty." The sections you are confident about are the ones that will surprise you eventually.

After the count: review before you submit

This step is where the value is, and where most operations shortcut.

Before submitting the count, a manager should review the variance report. Not to check the arithmetic, but to sense-check the story.

Look at what is unexplained. A -15% variance on a product might mean a delivery was short and not credited. It might mean that product has been used in a way the recipe does not account for. It might mean the count was done incorrectly. The variance does not tell you which. Your operational knowledge does.

Look at what is perfectly on expected. A count where every single line item comes in exactly as predicted is statistically unusual in a real operation. If every category is spot on, it may mean the operation is exceptionally well-run. It may also mean sections were estimated rather than counted. Both are worth knowing.

Make notes. A variance of twelve bottles of house red is a number. A variance of twelve bottles of house red on the week of a private hire event that ran two hours over is a context. The number without the context is not actionable. The number with the context explains itself.

What to count and how often

Weekly: wet stock in a high-volume bar or restaurant. Spirits, wine, draught lines. The value is high and the movement is fast. Weekly counts catch problems before they compound.

Fortnightly or monthly: dry goods, ambient, and food provisions, depending on volume. Lower value, slower movement, but still worth tracking regularly.

At every delivery: spot-check against the delivery note. You do not need to count the full stock room every time a delivery arrives, but checking that what was delivered matches what was ordered takes minutes and is where credits get raised before the opportunity disappears.

On menu change: whenever the dish mix shifts significantly, a full count resets the baseline. Otherwise you are comparing current stock against a theoretical usage that no longer reflects what the kitchen is producing.

The common mistakes

Counting without a baseline. You have quantities. You do not have variance. The count produces a list, not an analysis.

Not entering purchases before counting. Every delivery that has not been recorded looks like unexplained consumption.

Submitting without review. The count produces numbers. The review produces insight. Skipping the review means the data is filed, not used.

Doing the count in a hurry because service starts in an hour. A count done under time pressure is estimated, not measured. Schedule the count for a time when it can be done properly.

Counting the same section twice or not at all. Happens more often than operators realise, particularly in complex storage layouts. A consistent starting point and a ticked-off section list prevent it.

How the count connects to GP

Every count you do feeds into your GP calculation for the period. Opening stock plus purchases in minus closing stock equals cost of goods sold. Cost of goods sold divided by revenue gives you your gross profit percentage.

That is why the count matters. It is not a compliance exercise. It is not a management check on the team. It is the data point that makes your GP figure a real number rather than an approximation.

A carefully done count, reviewed against the context of the period, produces a GP figure you can act on. A rushed count produces a GP figure that looks plausible and tells you nothing.

Using a mobile app versus paper

Paper counts are slower, more prone to transcription error, and require manual transfer before they produce any output. If your current process involves paper sheets that someone types into a spreadsheet later, two errors can occur for every one item counted: the count itself and the transcription.

A count done on a mobile device, in the app, updates the system directly. Variance is visible on review without any manual calculation. The gap between completing the count and having useful output is minutes rather than hours.

The practical difference in count quality is significant. When the feedback loop is short and the output is immediate, the person counting understands what they produced. When feedback arrives two days later as a spreadsheet, the connection between the count and the outcome has been lost.

For new year and seasonal stock takes, the gap between a digital and a paper process is even more pronounced — the volume is higher and the time pressure is greater.

If you are looking for a stock take process that works the way real operations work, Hops was built around exactly that: count on mobile, review on desktop, variance visible immediately.