Food cost
How to calculate restaurant food cost
Food cost percentage is the single number most operators check first, and the formula is genuinely simple. The part that goes wrong is rarely the arithmetic — it is what gets counted, and when.
The formula
Food cost percentage equals cost of goods sold divided by food sales for the same period, expressed as a percentage. Cost of goods sold is beginning inventory plus purchases, minus ending inventory. Spend $28,000 on food purchases in a period, start with $6,000 in inventory and end with $5,400, and cost of goods sold is $28,000 + $6,000 − $5,400 = $28,600. Divide by $95,000 in food sales and food cost runs 30.1%.
Where the number quietly breaks
Three things move this figure without anything actually changing in the kitchen:
- A count taken a day late or early. Ending inventory is a snapshot; move the snapshot by a day and cost of goods sold moves with it, especially around a delivery.
- An invoice entered against the wrong period. A vendor bill dated the 1st but entered on the 5th inflates the wrong month's purchases.
- Transfers between locations, uncounted. Stock moved from one store to cover another shows as a purchase at the receiving store and inventory loss at the sending one, unless the transfer itself is recorded.
What a good target looks like
Most full-service and fast-casual operators target food cost in the high-20s to low-30s percent, with the right number depending heavily on menu mix and format — a protein-heavy steakhouse and a pizza concept do not share a target. The more useful benchmark than an industry average is your own best-performing location on the same menu, because it already proves what is achievable with your suppliers.
Counting against what was actually invoiced
The reliable way to catch drift before the period closes is counting inventory against what the vendor invoiced, by category, rather than waiting for a period-end reconciliation to surface a number nobody can trace back to a cause.
See restaurant food cost management
A food cost figure is only as good as what fed it. Ask whether a tool counts against invoiced purchases by category, or produces a single blended number that cannot be traced back to a specific ingredient or vendor.