Food Cost Percent: Purchases Are Not the Same as Consumption

Food cost percent answers how much of the matching sales was consumed by food. It does not tell you how much profit the restaurant made.

Use inventory to find food consumed

Start with beginning inventory + purchases − ending inventory. If you begin with $2,000, buy $3,000 and finish with $1,500, you used $3,500. On $10,000 of net food sales (what customers paid for food, after discounts and refunds, before sales tax and tips), food cost is 3,500 ÷ 10,000 × 100 = 35%.

Dividing purchases alone by sales would produce 30%. That misses the $500 reduction in inventory. Use the same dates, valuation basis and units for the counts, purchases and sales. Reconcile transfers, credits, staff meals, waste and count errors before interpreting the result.

Two weeks, one big delivery: why purchases mislead

A fictional counter restaurant sells $16,000 of food in each of two weeks. In week 1 a $2,500 case order lands on Saturday; in week 2 the kitchen works that stock down. Count the same way at the same moment each time: here, Sunday after close, before any Monday delivery. Week 2 opens with exactly what week 1 closed with. Food used is everything that left inventory according to the records: sold, wasted, staff meals and anything unrecorded.

The kitchen used the same $5,280 of food both weeks: 5,280 ÷ 16,000 × 100 = 33.0%. Dividing purchases by sales says 40.6% (6,500 ÷ 16,000), then 23.1% (3,700 ÷ 16,000): a swing of 17.5 percentage points with no change in the food used according to the records. Even over both weeks, purchases alone give 31.9%, because inventory fell $360 from the first count to the last and that food was used too.

The count mistake that reverses itself

Now suppose the Saturday delivery was still stacked in the receiving area at the week 1 count and nobody counted it. Closing inventory is recorded as $2,720 instead of $5,220. Food used becomes 4,000 + 6,500 − 2,720 = $7,780, and the week reads 48.6%. In week 2 that stock is counted at the end, but the opening carries the mistake: 2,720 + 3,700 − 3,640 = $2,780, or 17.4%. In this fictional example the cause is known: $2,500 of received stock was left out of one count, so the spike and the dip are the same size and the two-week total is still 33.0%. In a real restaurant that pattern is a prompt to reconcile the counts, receiving dates and purchases before drawing any conclusion about waste, portions or anything else. The pattern itself does not prove the cause, and a variance alone does not establish theft.

Three counting routines make this kind of error less likely and easier to find. Count at the moment the sales period closes, and count everything that has been received, including stock still on the dock. Keep purchases on the receiving date and net of supplier credits: if $300 of that delivery was rejected and credited, purchases are $6,200 and the rejected goods are not in the count either. Carry the reconciled closing count into the next period’s opening count. If you later find an error in a count, correct both periods consistently and record what changed and why, so the periods still reconcile. The ordering guide covers the counting routine itself.

Check it in the worksheet

The food cost calculator page has a second section, “Actual food cost for a period”, with exactly these four inputs: opening inventory, purchases, closing inventory and net food sales. It returns the percentage, the dollars of food used and the sales left after food, and it shows the formula with your numbers. It has no target or expected-cost input, and it refuses a closing count larger than opening plus purchases, so a typo such as $52,200 for $5,220 produces an error instead of a negative cost. To practice the formula first with checked answers, use the inventory-count lesson.

Keep the denominator comparable

If your numerator contains food only, compare corresponding net food sales. If you include food and beverage consumption, use matching sales. Including sales tax or employee tips makes the percentage look artificially smaller. A platform bank payout net of commissions is not the same as the full food sale.

Recipe cost answers a different question

A recipe costing $3 per portion at a $15 menu price has 20% theoretical food cost. This assumes its specified portion, yield and ingredients. It does not measure actual waste, giveaways, overportioning or an incorrect recipe.

Use the recipe calculator to check purchased quantity, usable yield and portion. A 20 lb case costing $60 with 75% usable yield gives 15 lb usable at $4/lb. There are 16 ounces in a pound, so six usable ounces cost (6 ÷ 16) × $4 = $1.50. Do not apply that same yield loss twice.

Explain a gap in dollars

Suppose actual food cost is 35% and the comparable expected mix suggests 30% on $20,000 sales. The difference is 5 percentage points: $20,000 × (5 ÷ 100) = $1,000. It is not proof of theft, and it is not a 5% relative increase: 35% is 16.67% higher than 30% (5 ÷ 30 × 100).

Check high-spend ingredients first: counted quantities and units, invoice prices, recipe portions, yield, sales mix and recorded waste. Serving 8 oz instead of 6 oz of a $4/lb ingredient adds $0.50 per dish. On 1,000 dishes, that is $500—half the example gap.

Turn the finding into an operating change

Weigh portions, provide the correct portioning tool, correct invoice/UOM errors, revise a par that causes spoilage, or reprice a dish using its complete cost and customer value. Do not cut quality blindly because the percentage is high.

Use the weekly scorecard and the inventory and waste worksheets. Review the next comparable period to see whether the change reduced dollars of waste or improved contribution. A lower percentage caused by a price increase is different from using less food.