Weekly stock counts and the fast route to real GP margin
General information for UK hospitality operators, not legal advice. If something here is wrong or out of date, the editorial policy explains how to report it.

Weekly stock counts turn a GP problem into a specific week, delivery or line you can check, rather than a monthly mystery.
Weekly stock counts are short, targeted counts of the lines that carry the most money, run every week alongside the full count the accounts still need at month end. They turn a gross profit problem into a specific week, a specific delivery or a specific line that can still be checked, rather than a monthly figure that only confirms something went wrong somewhere. The trade-off is more counting sessions, each of them shorter, in exchange for a variance report that covers seven days instead of thirty.
What gross profit measures, and why VAT comes out first
Cost of sales is opening stock plus purchases minus closing stock. Gross profit is net sales less cost of sales, expressed as a percentage of net sales, and the word net carries most of the weight. The price on the menu includes VAT at the standard rate of 20 per cent that applies to most goods and services, and that money belongs to HMRC rather than to the business. Working GP out on the gross till price flatters the figure by several points and makes every comparison worthless, whether against the target the operator has set, against last year, or against another site in the group.
Both ends of that sum depend on a physical count of what is actually on the premises, which is where the theory and the practice pull apart. A count taken once a month leaves a single data point to explain four weeks of trade, produced days after the last of those weeks has closed.
Weekly stock counts against a monthly cycle
Take a site whose food GP has landed three points below the target its operator sets for that menu. On a monthly cycle the drift starts, runs unnoticed for up to four weeks, and only surfaces once the period has closed and the figures have been produced. By then the rota has moved on, the delivery in question is one of several since, and nobody can say with any confidence which shift was short-staffed or who was on the section.
Counting weekly gives four times the resolution across the same period and a window small enough to interrogate. A manager can take one week, cross it against one supplier delivery and one part of the rota, and rule causes in or out while the people involved still remember the shift. None of that removes the need for a full count at month end, which is what reconciles opening and closing stock for the accounts. It changes what happens in the twenty-nine days in between.
Variance is the output, not the count
A count on its own is a valuation. The number worth acting on is the variance between theoretical and actual usage. Theoretical usage is what the sales mix and the recipes say should have left the storeroom or the cellar, based on what actually sold through the till. Actual usage is what the count says has gone. The gap between the two is where the story sits, and a count with nothing to compare it against says very little.
Both sides have to be kept current for the comparison to hold. A recipe costing that has not been revisited since the last supplier price rise will throw off the theoretical side as surely as a rushed count throws off the actual side. Recording waste as a movement in its own right, rather than letting it vanish into the variance figure, also matters, because it is the difference between knowing there is a waste problem and merely suspecting one.
The likely causes of GP drift, roughly in order
Once a variance appears, the instinct in many kitchens and bars is to assume theft. In practice it usually sits further down a shorter list. Over-portioning and free pours are training problems rather than dishonesty, and they cluster around particular shifts or particular staff who have never been shown the standard measure or the portion weight. Waste and spillage that never gets recorded is common and easy to miss, especially on prep-heavy sections where trimming and spoilage happen out of sight of the pass.
Staff meals, comps and manager discounts handed out without being booked through the till appear as usage with no matching sale. Short deliveries signed for without being checked against the invoice erode stock before anything is sold, which is a goods-in problem rather than a counting one. Supplier price creep on unchanged volumes looks identical to a portioning problem on the GP line until purchase prices are checked against history. Theft is real, and weekly counting is one of the ways it gets caught, but putting it at the top of the list usually costs a fortnight of chasing the wrong line while the actual cause carries on.
A counting cadence that survives week three
Trying to count everything every week is the most common reason weekly counting is abandoned inside a month. A cadence that holds splits the work by risk rather than treating every line the same, so the stock that moves the GP figure is counted most often and the rest is not counted for the sake of it.
| Line type | Suggested frequency | Reason |
|---|---|---|
| Spirits and draught | Weekly | High value, high pour variance, easily over-served |
| High-value proteins | Weekly | Expensive, portion-sensitive and exposed to price moves |
| Fast-moving low-value stock | Rotating daily spot check | Catches early drift in minutes rather than a full count |
| Everything else | Monthly full count | Reconciles opening and closing stock for the accounts |
Consistency matters more than the exact split. A count taken at a different point in the trading week, or after the delivery one week and before it the next, produces a variance that reflects the timing rather than the trading. Fixing the day, the time and the order of events, and counting before the first delivery of the week lands, removes a whole category of argument about whether the number is real.
Turning a variance into next week's decisions
A weekly variance earns its time only if something changes the following week: a re-trained pour, a corrected recipe cost, a conversation with a supplier about a price that moved without notice, or a dish repriced from what its ingredients cost now. Getting the theoretical side right starts with costing, and the free GP calculator checks a single dish or a keg against current purchase prices without an account. Building menu prices from a costed recipe in the first place, the approach set out in costing a menu with a GP calculator, keeps the theoretical figure anchored to what ingredients cost today rather than to a price agreed a year ago.
Waste is the other half of the same job. Sites that treat food waste and stock control as one task, rather than two reports produced by different people, tend to close the gap between theoretical and actual usage faster, because the same person is looking at both sides of the sum.
The paperwork behind a count also outlives the week it covers. A VAT-registered business must keep its VAT records for at least six years, and a limited company must keep its accounting records, including the stock it owns at the end of the financial year, for six years from the end of the financial year they relate to. Purchase invoices, credit notes and closing stock figures serve the weekly variance and the tax record at the same time, which is a strong argument for capturing them once, in one place, rather than twice in two. Stock management software that records a count where the stock is held and works out variance per line is mostly a way of removing the re-keying that makes weekly counting feel like extra work.
Keeping this on the record
A weekly count is worth the effort only if what sits behind it, the purchase invoices, the waste log and the variance by line, is kept somewhere it can be checked later rather than discussed once and forgotten. Zynthio is built to hold a running theoretical stock level alongside the counted one, with deliveries adding to it, wastage and transfers taking from it and sales depleting it through the recipes, so a variance report can put the two side by side line by line. Stocktakes are created per department and entered on whatever device is to hand, with barcode lookup, and a photographed delivery note can be uploaded for its lines to be pulled out rather than typed in. Recipe costs inherit the preferred supplier price, and GP is worked out on the net selling price with VAT stripped out. That stock control sits in the EPOS and Stock add-on, which is flagged as coming soon rather than on sale today, so the stock management page describes what it is built to do, while the free GP calculator is available to use this week.
Frequently asked questions
How often should a full stock count happen if we also count weekly?
Keep a full count monthly for the accounts, since that is what reconciles opening and closing stock for the period. Weekly counts sit on top of it, focused on the lines that carry the most money or the most risk, so a problem is narrowed down long before the monthly figure appears.
What is a normal GP variance to worry about?
There is no single number that applies to every menu or bar, because mix, pricing and wastage differ by site. What matters is the trend against your own target and your own history: a line that has held steady for months and then drifts by more than a point or two in a week is worth a look.
Does weekly counting take significantly more staff time than monthly?
Counting fewer lines more often generally takes less total time than one large monthly count, because each session is short and focused. The time cost comes from doing it inconsistently, where staff re-learn the process each time rather than building a routine.
Can a weekly count replace supplier invoice checks?
No. A stock count tells you what happened to goods after delivery, while an invoice check confirms what was actually delivered against what was ordered and charged. Short deliveries and price rises are best caught at goods-in, with the count confirming usage afterwards.
Who should carry out the weekly count?
Ideally a manager or supervisor who was not the only person handling that stock during the week, so the count acts as a check rather than a self-report. Rotating who counts also reduces the chance of a blind spot becoming a habit.