Calculating food cost: the formula, the stock count and the variance from theoretical

A site’s food cost, also called the food cost ratio or food cost percentage, is calculated over a closed period: a week, a month. It does not measure what you bought, but what you used, and sets it against what you sold.

Everything is counted excluding VAT, purchases including delivery charges, with credit notes and discounts deducted. INSEE defines the cost of goods sold in the same way: purchases and incidental costs, plus opening stock, minus closing stock. Your accountant does this calculation at least once a year: the French chart of accounts (plan comptable général) adjusts purchases (accounts 601 and 607) for the change in stocks (account 603). Food cost repeats it over a short period, for food alone.

The denominator counts food only, with discounts deducted. Drinks are tracked separately, with their own purchases and stock: they have neither the same margin nor the same waste. Start from the item categories in your POS rather than from VAT rates: in France, a soft drink served by the glass is taxed at 10%, like a dish. For a set menu with a drink, split the price by product, as the VAT rules allow, rather than applying the highest rate to the whole.

The cost of a dish, its yield and its selling price are calculated on its recipe card, which is what provides the theoretical figure.

Without a stock count, there is no actual food cost, only purchases divided by sales. That ratio follows deliveries: a big order received on the 30th makes it jump, a delivery pushed back to the 2nd makes it drop, for the same consumption.

One period’s closing stock is the next period’s opening stock: one count per close is enough, always at the same moment, for example on the last evening of the month after service. A delivery received after the count belongs to the next period, and so does its invoice: this is the cut-off. Otherwise both months are wrong, one in each direction.

Count everything that has a value, including opened products and prepared items (sauces, kitchen stocks, mise en place, at their recipe card cost). Value them at purchase price excl. VAT, using the same method from one count to the next. How to organise the count is covered in the stock management guide.

A neighbourhood restaurant is closing October. It counted its food stock on the evenings of 30 September and 31 October, and reconciled the month’s invoices.

The restaurant also sold €11,000 excl. VAT of drinks. Added to the denominator, they would bring the ratio down to 13,400 ÷ 53,000 = 25.3%: more flattering, and wrong, since their cost is not in the numerator.

Sample figures, invented for the calculation.

Theoretical food cost tells you what the menu should have cost if every portion sold had followed its recipe card, with no waste and no complimentary meals. It is calculated on the same sales and over the same period as the actual figure.

It is only accurate if every POS item links to a recipe card, and if the recipe cards keep up with purchase prices. A dish without a recipe card counts as zero and inflates the variance; so does a recipe card at last year’s prices.

In the example, the variance is 2.9 points, or €1,220. Read both: the points to compare months and sites, the euros to know whether it is worth an hour of investigation. A steady variance suggests a habit or a wrong recipe card; a variance that jumps suddenly suggests an event: a delivery, a stock count, a refrigeration failure.

A variance is not proof of waste. Go through six causes, from the quickest to check to the slowest.

Data entry errors. An invoice counted twice, a forgotten credit note, a case entered as a single unit, a dish given away without going through the till: the figure is wrong before anything has been consumed.

Purchase prices. The actual figure pays invoice prices, the theoretical one recipe card prices: if beef goes up and the recipe cards do not follow, the variance grows without a gram more on the plate. A discount obtained after the fact reduces the cost of purchases (the chart of accounts records it in account 609): always assign it to a period in the same way.

Staff meals. They consume without selling. Record them at recipe card cost, then take them out of the cost of food consumed or show them separately, in the same way every month and at every site.

Transfers between sites. A case lent to another site is recorded on both sides: if it is forgotten, it penalises the site that gives and flatters the one that receives.

Waste. Out-of-date products, dishes sent back, frying oil thrown away: a waste sheet kept every day gives you the amount.

Portions. A heavy hand with the meat or the chips repeats itself on every plate. It is the last thing to look for, by comparing actual and theoretical consumption product by product.

With the credit note entered, the transfer recorded and staff meals taken out, the actual figure comes to €12,750, or 30.4%; with recipe cards updated, the theoretical figure rises to €12,390, or 29.5%. That leaves 0.9 points, or €360: €260 of known waste and €100 to look for in portions.

Sample figures, invented for the calculation.

Actual food cost is calculated at each full stock count, usually every month, in line with the accounting calendar. In between, count the products that weigh most (meat, fish, cheese) every week and compare their actual consumption with the theoretical: a variance found on the 8th can be corrected for three weeks.

Compare the actual figure with the theoretical for the same period, not with last year’s ratio. An actual figure that rises with the theoretical points to the menu or purchase prices; an actual figure that rises on its own points to the kitchen, the stock or data entry.

In a group, each site has its own stock count, sales, invoices and transfers; deliveries from a central kitchen count as transfers, at cost. Each site is compared with its own theoretical figure, and the group average is read last, because it offsets one site against another. The method for running several restaurants covers these comparisons in detail; a margin tracking tool keeps them up to date site by site.

Calculating on purchases alone. Without two stock counts, the ratio follows deliveries, not consumption.

Mixing food and drinks. The ratio drops without the kitchen having changed anything. Track two ratios, with two stocks.

Counting at different times. One count before service, the next one after: a day’s consumption moves to another month.

Changing valuation method. Last invoice price one month, average price the next: the variance moves without anything having been consumed.

Looking for a culprit before a cause. A team under suspicion rounds off its stock counts. Check data entry and prices first: they are the quickest checks.

Food cost, food cost ratio, food cost percentage: are they the same thing? Yes: the cost of ingredients consumed divided by sales excluding VAT. Just specify the scope (food only or drinks included, staff meals deducted or not) and the period. Two figures calculated on different scopes cannot be compared, even under the same name.

What food cost should you aim for? There is no right figure for everyone: the target depends on the concept, the prices and the sales mix. Set it from your recipe cards: your theoretical food cost is the level your menu allows. First close the gap between actual and theoretical; lowering the theoretical means changing recipes, the menu or prices.

Can you calculate food cost without a stock count? Not actual food cost. Purchases divided by sales give an approximation that follows deliveries: acceptable over a long period if stock changes little, misleading over a week. If you cannot do better, count the most expensive products every month, and the whole stock at each accounting close.

Why does my food cost change when the menu has not? Because of purchase prices, the sales mix or the stock count. A price rise weighs on the actual figure straight away, and on the theoretical as soon as the recipe cards are updated. A month in which high-ratio dishes sell better raises the theoretical figure itself: first check whether it moved with the actual.

Do you need software to track food cost? For a single site, a spreadsheet is enough, with regular stock counts and up-to-date recipe cards. The work grows with the theoretical figure: linking every sale to its recipe card, and every recipe card to the latest purchase prices. When the menu changes often or several sites share recipes, a tool that makes these links gives you that time back.