Food cost variance: reading the number without panic

Food cost variance is the gap between the food cost you expected and the one you got. Sometimes it is last month's 29 per cent turning into this month's 31. Sometimes it is the difference between what the recipes say the food should have cost and what the stock count says it did. Either way, it arrives as a number on a report, and the temptation is to act on it at once: tighten portions, change a supplier, talk to the team about waste.

That is usually the wrong first move. A swing of two points can come from a real leak in the kitchen, but it can just as easily come from a stock count that missed a shelf, a delivery booked to the wrong month or a change in what guests ordered. Acting on a variance before you know which kind it is wastes effort at best, and at worst damages the trust of a team that did nothing wrong.

This guide explains the two kinds of food cost variance, how sensitive the number is to small errors, and the order of checks that separates a counting problem from a kitchen problem. Every figure in it is an example.

The phrase is used for two different comparisons, and it helps to know which one you are looking at.

The first is the change from one period to the next. Food cost was 29 per cent of food sales last month and 31 per cent this month: a variance of two points. This comparison tells you that something changed, but not what, because many things change between months, from prices to the menu mix to the weather.

The second is the gap between actual and theoretical food cost in the same period. The actual figure comes from the stock count and purchases. The theoretical figure comes from the recipes: each dish sold, multiplied by what its recipe should cost. If the recipes say food should have cost 28 per cent and the count says 31, the variance is three points of food that left the kitchen without being accounted for by the recipes. Our guide to actual vs theoretical food cost covers that comparison in depth.

Both are worth tracking. The period-to-period change is easier to produce, because it needs no recipe costs. The actual against theoretical gap is more useful, because it separates the effect of the menu mix from the effect of the kitchen.

Food cost percentage is a ratio of two figures, and the top figure is built from a stock count. Small errors in the count move the percentage further than most people expect.

A worked example. A kitchen with food sales of 40,000 in a month calculates its cost of goods sold as opening stock plus purchases minus closing stock. Suppose the true cost of goods is 11,600, which is 29 per cent. Now suppose the closing count misses a shelf worth 400. Closing stock is understated by 400, so the cost of goods appears to be 12,000, and the food cost becomes 30 per cent.

In other words, in this example, every 400 of stock missed or double-counted moves the percentage by a full point. Miss 800, and the kitchen looks two points worse without a single portion having been over-served. Count the same shelf twice, and the kitchen looks two points better.

Because of this, the first question about any food cost variance is not what happened in the kitchen. It is whether the numbers that produced it can be trusted. Stock count accuracy comes before everything else.

Work through the following checks in order. Each one is quicker and less disruptive than the next, and most variances are explained before you reach the end of the list.

Start with the closing stock, then the opening stock, because the opening figure is last period's closing figure and may have its own errors.

Our guide to running a restaurant stock count properly covers the counting discipline that prevents most of these.

Every delivery and every invoice must sit in the period in which the stock arrived. A delivery that landed on the last day of the month but whose invoice was booked in the next month makes this month look cheaper and next month dearer, and the reverse is just as common.

Check credit notes too. A credit for short-delivered or rejected goods that has not been entered leaves the kitchen paying, on paper, for food it never had.

The bottom of the ratio matters as much as the top. Make sure food sales are recorded net of tax and service charge, and that drinks are not mixed in. Then check the giveaways: comps, voids, staff meals and discounts. A month with more comps than usual has more food used and less sold, which raises food cost without anything going wrong in the kitchen. Recording them properly does not reduce them, but it explains them.

Compare the prices on this period's invoices with last period's for the main lines. A rise in the price of a few heavily used items, such as proteins, dairy or oil, can move the food cost by a point on its own. That is a real cost change, but it is a buying and pricing question, not a kitchen discipline one. Our guide to restaurant supplier management covers how to keep track of it.

If guests ordered more of the dishes with a high food cost and fewer of the ones with a low cost, the overall percentage rises even if every plate was perfect. A warm month that sells more fish and fewer pasta dishes can do it. So can a special that sold well.

This is where theoretical food cost earns its keep. If the theoretical figure rose by the same amount as the actual one, the mix explains the change, and the kitchen did nothing wrong.

Only now look at the kitchen itself. Is the waste log showing more than usual? Have portion sizes drifted, especially on proteins and garnishes? Have trim yields fallen because of a new supplier or a new cut? These are real leaks, and they are worth fixing, but they are rarely the whole story of a two point swing.

Theft happens, but it is the explanation you should reach after the others have been ruled out, not before. A team accused on the strength of a count that turns out to have been wrong will not forget it.

To see how the checks work together, follow one invented month through them. The kitchen's food sales were 40,000 and its food cost came out at 31 per cent, against 29 the month before. Two points on 40,000 is 800 of food cost that needs explaining.

The recount comes first. The chef recounts the twenty most valuable lines and finds that the freezer's second shelf was missed in the closing count: 300 of stock. Corrected, closing stock rises by 300, cost of goods falls by 300, and the food cost drops to 30.25 per cent.

The period comes next. A delivery of dry goods worth 250 arrived on the last day of the month but had been entered as if it had arrived the next day; its invoice, however, was booked this month. The stock was used next month, so it belongs in this month's closing stock. Adding it takes the food cost to 29.6 per cent.

The sales figure is correct, but comps were higher than usual: a supplier's tasting evening and two complaint tables account for an extra 120 of food used and not sold. That is real cost, but it is explained, and it is not a kitchen leak.

The mix explains most of the rest. Fish sold strongly in a warm month, and the theoretical food cost rose by about half a point. What remains, a little over a tenth of a point, is inside the kitchen's normal range. The month that looked like a two point problem turns out to be a missed shelf, a misplaced delivery, a known event and a sunny few weeks.

Not every month resolves so neatly. Sometimes the checks leave a gap that only the kitchen can explain. But working in this order means that when you reach the kitchen, you are talking about a gap that is real.

Monthly food cost is what the accounts need. Weekly food cost is what a kitchen can manage. The two answer different questions, and it helps to keep both.

A weekly figure is noisier, because a week has fewer sales to spread the count's errors across and because deliveries land unevenly from week to week. Its value is speed: a change shows within days, while the people who were working can still say what happened. A monthly figure is steadier and ties to the accounts, but by the time it arrives, the month is history.

A practical pattern is to count the twenty most valuable lines weekly, produce a weekly food cost from them, and count everything monthly. The weekly figure warns; the monthly one confirms. If the weekly figures for a month add up to something very different from the monthly figure, the difference itself is worth investigating, because it usually points at the count.

If your stock system produces an inventory variance report, it lists, for each item, what the count says was used against what the sales and recipes say should have been used. It is the most useful view of food cost variance, because it shows where the gap is rather than only how big it is.

Read it in this order:

Even without a stock system, the same approach works by hand on your top twenty lines by value.

Not every movement needs action. Every count has some noise, and chasing it wastes time. Set a tolerance for your own kitchen, based on how much the figure usually moves when nothing is wrong.

A practical way to do it is to track the food cost every week for two or three months, note how far it moves from week to week in ordinary times, and treat anything inside that range as noise. In one example kitchen, the weekly figure might move between 29 and 30.5 per cent in normal weeks; a week at 32 is worth investigating, and a week at 30.2 is not.

Two other rules help:

A variance is easier to read when the tracking behind it is consistent. A few habits help a great deal:

Our guide to calculating food cost percentage step by step sets out the calculation itself, and our guide to building a food cost calculator in a spreadsheet shows how to keep the tracking in one place.

Once the checks have found the cause, the response follows from it.

Explain the result to the team, whatever it is. A variance that turned out to be a counting error is worth saying out loud, because it tells the kitchen that the number is checked before anyone is blamed for it.

Food cost variance needs three things: what was sold, what was used and what each dish should cost. servearound covers what was sold, and the records that explain some of the gap. The counts, the purchases and the recipe costs stay in your own system or sheet today.

The menu report in Analytics shows units sold, revenue and share of sales for every dish and category over any range of dates. That is the sales side of a theoretical food cost, and the place to see a change in the mix. The payments report shows discounts, voids and comps by member of staff, so you can see whether a month with a higher food cost was also a month with more giveaways. Bills for a date range can be downloaded as a spreadsheet for your own reconciliation.

servearound does not track stock, hold recipe costs or produce a variance report today. Inventory tracking, for the stock behind the menu, is coming. If you would like to see the sales side for yourself, get in touch.

It is the difference between the food cost you expected and the one you got. It can mean the change from one period to the next, or the gap between the actual food cost from your stock count and the theoretical food cost from your recipes and sales.

There is no general figure. Measure how much your own food cost moves in ordinary weeks over two or three months, and treat movement inside that range as noise. Investigate anything outside it.

In many kitchens, counting and timing errors explain a large share: a missed location, a wrong unit, a delivery or invoice booked to the wrong period. Check those before looking at portions, waste or theft.

Actual food cost is opening stock plus purchases minus closing stock, divided by food sales. Theoretical food cost is the recipe cost of every dish sold, added up and divided by the same sales. The gap between them is the food that left the kitchen without being explained by the recipes.

Weekly, if you can count your main lines weekly, and monthly at a minimum for the accounts. Weekly tracking shows a change while it can still be explained.

Yes. If guests order more of the dishes with a higher food cost, the overall percentage rises even when each dish is made perfectly. That is why the theoretical figure matters: it shows how much of a change the mix explains.

Not until you know why it rose. If the cause is a count error or a price rise, cutting portions punishes guests for a problem they did not cause. Find the cause first.

Food cost variance is either a change from last period or a gap between actual and theoretical cost, and the number is more sensitive than it looks: in this guide's example, every 400 of stock missed moves the percentage by a full point. So read it without panic, and in order. Check the count, then the period, then the sales figure, then prices, then the mix, then waste, portions and yields, and consider theft last.

Set a tolerance from your own ordinary weeks, act on trends rather than single periods, and track consistently: the same count day, the same valuation, deliveries in the right period, comps and waste recorded as they happen. Then tell the team what the investigation found, whatever it was.