Food cost percentage: how to calculate it so the number means something

The formula is not the hard part.

Food cost % = cost of food used ÷ food sales

And cost of food used is opening inventory + purchases − closing inventory.

What makes the number unreliable is that both halves of that fraction are routinely built from data that has other things mixed into it. An operator comparing this month's 31% to last month's 28% is usually comparing two differently contaminated numbers and drawing an operational conclusion from an accounting artefact.

Theoretical versus actual is the finding

A single food cost percentage tells you very little on its own. The useful measurement is the gap between two numbers:

• Theoretical food cost — what the food should have cost, given what was sold and what each item's recipe costs.
• Actual food cost — what it did cost, from the inventory calculation above.

The difference is the variance, and the variance is where the money is. A restaurant running an actual food cost of 32% against a theoretical of 29% has a three-point problem with a physical cause: waste, over-portioning, spoilage, theft, unrecorded comps, or prices that moved without the menu moving. A restaurant running 32% against a theoretical of 32% has no such problem — it has a menu priced at 32%, which is a completely different conversation.

Without the theoretical, you cannot tell those two restaurants apart. Most operators are tracking only the actual, which is why food cost discussions so often end in a shrug.

Six things that corrupt the calculation

1. Beverage mixed into food. Liquor, beer, wine and non-alcoholic beverages carry entirely different cost profiles. Blended into one line they mask each other — a good beverage margin can hide a deteriorating food margin for a long time. Separate the categories on both sides of the fraction. The beverage side has its own instrument: the pour cost calculator.
2. Purchases used instead of usage. If you built stock ahead of a holiday, purchases overstate cost. If you ran it down afterward, they understate it. The percentage swings and nothing in the kitchen changed.
3. Employee meals left in. Real cost against no sale. It inflates food cost by a broadly constant amount every period, which makes it invisible in trend and wrong in level.
4. Comps not isolated. A comped dish consumed food. Netting comps against sales and leaving the cost in COGS moves the percentage for a reason that has nothing to do with kitchen performance. Comps deserve their own line, because their size is a management decision worth seeing.
5. Delivery commission in COGS. Commission is a cost of the sales channel, not a cost of goods. Booking gross order value as sales and commission as cost distorts both the numerator and the denominator.
6. Inconsistent count discipline. A count taken carefully at period end and casually at period start moves cost of goods used by the difference between the two, and attributes it to the kitchen.

The period problem

Calendar months are an accounting convention, not an operating one. February has 28 days; March has 31. A month containing five Fridays is not comparable to one containing four, and a percentage is not immune to that — sales mix shifts with day of week, and so does food cost.

This is why operators serious about the number move to a 4-4-5 or 13-period calendar, where every period contains the same number of each weekday. Comparisons then mean something without mental adjustment. The cost of the change is one conversation with whoever prepares your accounts; the benefit is that period-over-period movement becomes a signal instead of noise.

Frequency beats precision

A food cost percentage that arrives on day 20 of the following month is a historical statistic. One that arrives weekly, even built on a count of only the fast-moving categories, is a management tool.

Perfect monthly beats nothing. Approximate weekly beats perfect monthly, because it arrives while the order can still be changed. Counting proteins, produce and liquor weekly and the long tail monthly gives most operations a usable weekly number for a fraction of the effort of a full count.

What to do with the answer

A food cost percentage on its own justifies no decision. Paired with a theoretical, tracked weekly, split by category and read alongside labor as part of prime cost, it points at a specific cause: a supplier price that moved, a portion that drifted, a menu item priced when beef was cheaper, a station that wastes.

Those are all fixable. "Food cost is up" is not.

---

Related: Prime cost · Restaurant bookkeeping · KPI glossary · Sample flash report

Common questions

How do you calculate food cost percentage?

Cost of food used divided by food sales for the same period. Cost of food used is opening inventory plus purchases minus closing inventory — not purchases alone, which swings with stock building and drawdown.

What is the difference between theoretical and actual food cost?

Theoretical food cost is what the food should have cost given what was sold and what each recipe costs. Actual is what it did cost. The gap between them is the variance, and the variance is what identifies waste, over-portioning, spoilage or unrecorded comps.

Should employee meals be included in food cost?

They are a real food cost but they generate no sale, so left inside the calculation they inflate the percentage every period. They should be identified separately so the percentage reflects food sold to customers.

Why use a 4-4-5 calendar for food cost?

Because calendar months contain different numbers of each weekday, and sales mix and food cost both move with day of week. A 4-4-5 or 13-period calendar makes period-over-period comparison meaningful without mental adjustment.