Food Cost Percentage Calculation for Restaurants

Food cost percentage is one of the few numbers that tells you, every week or month, whether the kitchen is doing what the menu says it should. The formula is short. The trouble is that most operators calculate it from bad inputs, compare it to a target they found online, and then cannot explain why it moved.

This guide covers the calculation, how to cost a single plate, and the part that actually matters: the gap between what your food cost was and what it should have been.

The formula

Food cost % = (opening inventory + purchases − closing inventory) ÷ food sales × 100

The top half of that is cost of goods sold for food: what you had at the start of the period, plus everything you bought, minus what is still on the shelf at the end. Whatever is gone was used, wasted, given away or stolen. The bottom half is food sales for the same period, excluding sales tax and, normally, beverage sales.

Keep food and beverage separate. Bar costs run on very different margins, and blending them hides problems in both.

A worked example

Say a restaurant closes its books for a four-week period.

$8,400 + $31,250 − $7,900 = $31,750. Divide by $102,400 and you get 0.31, or 31.0 percent.

Some practical points on the inputs:

• The closing inventory of one period is the opening inventory of the next. Count once, at the same time of day, at period end. Counting on a Monday morning after a big weekend gives a very different number from a Friday afternoon after a delivery.
• Purchases should be what was received in the period, by invoice date, minus credits for returns and shorts. Do not use payments, since you may pay last month's invoices this month.
• Value the count at the latest invoice price or a consistent costing method. Switching methods between periods creates false swings.
• Transfers count. If you move product to another location or the bar takes lemons, record it or the kitchen pays for it.

If you skip the inventory count and just divide purchases by sales, you get a number that moves around with your delivery timing. A big order on the 29th makes the month look terrible and the next month look great. Purchases-only is fine as a weekly early warning, but not as your real food cost.

The stock count procedure covers how to count so the closing inventory is worth trusting.

Costing a plate

The period calculation tells you what happened across the whole kitchen. Plate costing tells you what each dish should cost, and you need it before you can say whether 31 percent is good.

To cost a plate, list every ingredient in the recipe at the portion actually served, and price each at its usable cost.

Usable cost is where people go wrong. A case of whole salmon at $9.50 per pound does not produce $9.50 fillets. After skin, pin bones, belly and trim, you might get a 70 percent yield of usable portions, so the real cost is $9.50 ÷ 0.70 = about $13.57 per usable pound. The same goes for produce (a head of romaine loses its core and outer leaves), and for anything that shrinks in cooking.

Here is a hypothetical salmon plate:

If the dish sells for $26, its plate cost percentage is 6.97 ÷ 26 = 26.8 percent.

Do this for every menu item, multiply each plate cost by the number sold in the period, add it up, and divide by food sales. That gives you theoretical food cost: what you should have spent if every plate was made exactly to recipe and nothing was wasted.

Plate costs go stale fast. Re-cost your top sellers whenever a major ingredient price moves, and the whole menu at least twice a year.

Actual vs theoretical

This is the comparison that earns its keep.

Say the theoretical food cost for the period was 28.5 percent and the actual was 31.0 percent. The gap is 2.5 points, or about $2,560 on $102,400 of sales. That is product you paid for that did not turn into revenue at recipe cost.

A small gap is normal. No kitchen hits theoretical exactly, because some trim loss, staff meals and the odd dropped plate are unavoidable. Many operators treat a variance around one to two points as acceptable and anything persistently larger as worth investigating, though the right threshold depends on your concept and how complete your recipe data is.

The usual causes, roughly in order of how often we see them:

Waste

Overproduction that gets tossed at the end of its shelf life, spoilage from poor rotation, and food made wrong and remade. A food waste log turns this from a guess into a list of items and amounts. Overproduction usually traces back to prep pars that are too high, which the prep list guide addresses.

Portioning

A cook who serves 7 ounces of salmon instead of 6 adds 85 cents to every plate. Across 400 plates a month, that is $340 from one dish. Heavy-handed portioning is rarely deliberate. It comes from eyeballing instead of weighing, from scoops and ladles that do not match the recipe, and from recipe cards that are not posted at the station.

Receiving errors

You pay for 40 pounds and get 36. A case of tomatoes arrives with a quarter of them soft. The invoice price is higher than the quoted price. If nobody weighs, counts and checks prices at the door, these go straight into food cost. A proper receiving checklist catches most of them.

Unrecorded comps and staff meals

The plate sent to table 12 because the steak was overcooked, the dessert for a regular, the family meal the line cooks make at 4 pm. All of these are legitimate. If they are not rung in the POS as comps or recorded as staff meals, they show up as unexplained variance.

Theft

It happens, and it is usually less than people fear. Product walking out the back door tends to show up as one or two high-value items (proteins, liquor, cheese) with variance that the other causes do not explain. Before assuming theft, rule out the more common causes above, because accusing staff based on a variance number that turns out to be a counting error does lasting damage.

Bad data

Recipe cards that do not match what the kitchen actually makes, items missing from the theoretical calculation, inventory count mistakes, and invoices entered in the wrong period. On a new costing system, bad data usually explains most of the early variance. Fix the data before chasing people.

What target should you aim for

You will see 28 to 35 percent quoted as a typical restaurant food cost. That range is reasonable for many full-service restaurants, but the right target depends on what you sell.

A steakhouse or seafood restaurant often runs higher, sometimes high 30s, because the core product is expensive and the price ceiling is limited. A pizza or pasta concept can run in the 20s, because flour, tomatoes and dried pasta are cheap. Fine dining may run a higher food cost but makes it up on beverage and check size. Quick service tends to be lower on food cost and higher on packaging.

Food cost also trades against labor. A concept that buys pre-cut, pre-portioned product pays more per pound but needs fewer prep hours. Many operators watch prime cost (food plus beverage plus labor as a percentage of sales) for exactly this reason. A kitchen that cuts food cost by two points by adding a prep cook has not gained anything.

So the useful target is your own theoretical food cost. If your menu is costed at 28.5 percent, then 28.5 percent is what perfect execution looks like, and your job is to keep actual close to it. If you think 28.5 percent itself is too high, that is a menu pricing and engineering question, not a kitchen execution question.

How often to calculate

Monthly is the minimum, and most accountants will produce it monthly whether you ask or not. Weekly is better for anyone who wants to act on the number while the cause is still fresh. A full inventory count every week is a lot of work, so many kitchens do a full count monthly and a count of their top 20 items by value weekly, which covers most of the money.

A note for multi-unit operators

Across several locations, food cost becomes a comparison tool. If four sites sell the same menu and one runs three points higher, the menu is not the problem. Comparing actual against theoretical by site, and by item within each site, points you straight at where to look. It also shows you the site whose number looks too good, which can mean an inventory count that is wrong in the other direction.

When you find a cause, give the fix an owner and a date and check it the following period. The variance only shrinks if someone follows up.