Food cost percentage: the one number that reads your health
How to calculate food cost percentage properly from stocktakes rather than invoices, what separates theoretical from actual, and what to do when the gap widens.
If you could watch only one number in your restaurant, watch food cost percentage. It is the fastest signal that something has changed: a supplier price, a portion size, or a back door left open.
The definition
food cost % = cost of food consumed ÷ net food sales × 100
The most common mistake is putting purchase invoices in the numerator. Purchases are not consumption: the month you fill a freezer with beef looks catastrophic, the next month looks brilliant, and both are wrong.
The correct version goes through your stocktake:
cost of goods consumed = opening stock + purchases − closing stock
A monthly example:
| Line | Amount |
|---|---|
| Opening stock | 180,000 |
| Purchases | 420,000 |
| Closing stock | 210,000 |
| Cost consumed | 390,000 |
| Net food sales | 1,200,000 |
| Food cost % | 32.5% |
Without the stocktake you would have read 420,000 ÷ 1,200,000 = 35% and gone hunting for a problem that does not exist.
Theoretical versus actual
This is the part that genuinely changes how a restaurant is run.
- Theoretical: what your sales should have cost, from recipes multiplied by quantities sold.
- Actual: what they did cost, from the stocktake.
variance = actual − theoretical
| Variance | Reading |
|---|---|
| Under 1% | Tightly run |
| 1% – 2% | Normal for a busy site |
| 2% – 5% | Check portions, waste and receiving |
| Over 5% | Real leakage: theft, unapplied recipes, or a bad count |
On a restaurant selling a million a month, every percentage point of variance is 10,000 walking out of the door.
Where the gap usually comes from
- Loose portioning. The chef adding “a bit extra” cheese to every pizza costs you more than any supplier.
- Receiving without weighing. Sign for 20 kg, receive 18.5, and you paid for the difference.
- Unlogged waste. A burned plate thrown away without a record shows up later as stocktake variance nobody can explain.
- Staff meals. They have a cost. Record them as staff meals rather than as nothing at all.
- Rushed counts. A stocktake done after midnight by tired hands produces numbers you cannot act on.
Making the number appear by itself
Calculating by hand means you learn the number two weeks after month end — long after the moment when you could have acted.
When recipes, inventory and sales sit in one system, as they do in GoSufra, both halves are produced automatically: theoretical from recipes attached to sales, actual from stocktakes, goods-received notes and waste logs. The variance shows up weekly instead of surprising you after it has already cost you a full month.
Rough benchmarks
Approximate, and not a rule:
| Type | Common range |
|---|---|
| Coffee and drinks | 18% – 25% |
| Quick service | 25% – 32% |
| Full service | 28% – 35% |
| Seafood / steak | 35% – 45% |
A restaurant at 40% with good profit is in better shape than one at 28% with empty tables. The percentage is a tool for comparing you with yourself, not a competition.
A four-week plan
- Week 1: Fix the stocktake. Count high-value items weekly and everything else monthly, same day, same hour.
- Week 2: Write recipes for your top twenty sellers and calculate theoretical cost.
- Week 3: Start logging waste and staff meals properly.
- Week 4: Compare theoretical to actual and start with the three biggest gaps.
Two months in, the variance becomes a number you track rather than a surprise you wait for.
Run all of this from one system
POS, kitchen, inventory, recipe costing, staff and accounting — connected, and free to start.
Create your free account