How to calculate and cut your food cost %
The food cost formula, how to cost a plate with real yields, theoretical versus actual cost, and the levers that bring the number down.
Model food cost % for a dish across dine-in and delivery, and see how much margin it leaves. Below the calculator: how to work out food cost for a whole month, and what to do when the two numbers disagree.
What the customer pays for the dish
Raw materials per portion — protein, carbs, oil, spices
The small line most operators forget — garnishes, salads, sauces
Ingredients + garnish
Industry target: 28–32% QSR · 30–35% casual · 35–40% fine dining
68.8% of menu price — before labor, rent, utilities
Up to 32% is the target band for most formats. Past 38% the dish is underwater before labor.
Healthy
This dish sits in the target band for most service formats. Watch ingredient inflation — a 5% spike in raw material costs flips this to tight.
68.8% gross margin gives you headroom for ~30% labor + ~15% rent and still leave ~10–15% operating margin.
The industry rule of thumb says 30–32% food cost. That’s a starting line, not a target. A QSR running at 28% with a focused menu can out-earn a casual restaurant running at 35% with a sprawling one, even though the casual operator charges more per cover.
What actually matters is the spread between this dish and the others on your menu. Your top sellers should sit at the lower end of your target band, and the dishes guests order for the experience can run higher if the dollars per plate hold up.
The dial most operators sleep on: garnish. A dish modelled at 30% on its protein and carb alone can slide several points once sides, sauce and salad are added. The calculator above forces that accounting honestly.
Restronaut builds custom digital systems for F&B operators — including inventory-to-recipe costing dashboards that update in real time.
There are two food costs, and they answer different questions. The calculator above works out the first.
Plate food cost % = (cost of everything on the plate ÷ menu price) × 100
Plate cost comes from the recipe. It tells you whether one dish is priced right, and it is the number you set before the dish goes on the menu.
Period food cost % = ((opening inventory + purchases − closing inventory) ÷ food sales) × 100
Period cost comes from the stockroom. It tells you what the kitchen actually used over a week or a month, including everything that never reached a guest. The two should agree closely, and when they don’t, the gap is the most useful number you have — see actual versus theoretical below.
A costed recipe includes every item that leaves the pass with the dish:
Cost every ingredient at its usable price. Trim, bones, peel and cooking loss mean you pay for more than you serve. Divide the purchase price by the yield: chicken bought at $6.00 a pound with an 80% yield costs $7.50 a pound on the plate. Skipping this is the most common reason a costed recipe comes out several points lower than the kitchen really runs.
Period food cost also picks up everything that left the stockroom without being sold. None of it is in any recipe, so record each one separately or it becomes an unexplained gap:
Delivery platform commission, card fees and cleaning supplies belong elsewhere. Keep beverages on their own line too: bar cost runs on different margins, so a combined number moves when the sales mix shifts even if nothing in the kitchen changed.
It depends on the format, because the format decides how much of the price is food and how much is service, room and labor.
| Service format | Typical food cost |
|---|---|
| QSR / counter service | 28–32% |
| Casual dining | 30–35% |
| Fine dining | 35–40% |
The calculator checks each dish against one set of bands:
| Verdict | Food cost % |
|---|---|
| Healthy | 32% or under |
| Tight | 32–38% |
| Underwater | above 38% |
Two things move a restaurant within its band without anything being wrong. A lean kitchen can carry a higher food cost, because what has to work is prime cost — food and labor together. And one dish above band is not a problem if it sells in volume and earns more dollars per plate than the dishes around it.
The first is a plate costed with yield and without it. The second prices a dish from a target. The third is a month, worked from inventory.
Healthy. Cost the chicken at the $6.00 purchase price instead and the plate comes to $3.80, or 25.3%. The yield is 3.8 points the recipe card would never have shown.
The target gives a floor, not a price. If the dish sells nearby for $17, pricing it at $14.95 leaves money on the table; if nothing like it sells above $13, the recipe needs to change, not the price.
Recipe costs times dishes sold come to $19,600, a theoretical 30.6%. The kitchen used $2,200 more food than its sales account for, 3.4 points of revenue. That gap is what to chase, not the 34.1%.
Theoretical food cost is what the kitchen should have used: each dish’s recipe cost multiplied by how many were sold. Actual food cost is what the inventory says it did use. The first needs costed recipes and a POS item report; the second needs two counts and the purchase invoices in between.
Variance = actual food cost % − theoretical food cost %
A high actual food cost on its own tells you there is a problem and nothing about where. The variance splits it in two. If theoretical is already high, the problem is on the menu — recipes and prices. If theoretical is fine and actual is not, the problem is in the kitchen, and it is one of a short list:
Many operators aim to keep the gap within a point or two. More telling than its size is its direction: a variance that widens month on month is a control problem, and one that jumps in a single month is usually a miscount or a missing invoice.
Menu price = plate cost ÷ target food cost %
A $4.36 plate at a 30% target needs $14.53. That is where most pricing advice stops, and it is where the useful part begins, because the formula gives every dish the same percentage and a good menu doesn’t.
Guests pay for dishes, not percentages. A steak at 40% food cost can earn $12 a plate while a pasta at 22% earns $9, and if they sell in similar numbers the steak is the better dish. So price from the target first, then check two things before settling: what the dish sells for nearby, and how many dollars it leaves compared with what it would replace.
The menu engineering worksheet does that second check for the whole menu at once, sorting every dish by margin and popularity.
Sorted by how fast each one moves the number. The first three need no supplier and no price change.
Recipe costs go stale the day a supplier raises a price. The ten dishes that sell most are most of your food cost; re-cost those first, with yield, and you will know whether the problem is the menu or the kitchen.
Scales at the stations, the right ladle and scoop sizes, and a photo of each plate at the pass. Portion drift is the biggest line in most variances, and it costs almost nothing to stop.
Every discarded item, with a reason, for two weeks. Over-prep shows up within days, and the fix is a prep list driven by forecast covers rather than by habit.
Proteins, dairy and anything costly and perishable. A monthly count tells you there was a problem four weeks after it started; a weekly count of ten items catches it inside a week.
Move the high-margin, popular dishes to where the eye lands first and let servers recommend them. A mix shift lowers the whole restaurant’s food cost without touching a recipe.
Ask for a different cut, pack size or grade before asking for a lower price. A specification change often saves more than a discount, and suppliers agree to it more easily.
It works, and it is last because guests see it. If a dish is still over band after everything above, the problem is the price or the recipe, and one of them has to change.
The calculator checks one dish. The worksheet holds the menu: one row per dish, with packaging counted only for delivery, and a verdict against the same bands the calculator uses. It keeps the two food costs apart, the plate cost from the recipe and the period cost from the stockroom, so you can see the gap between them.
The first rows are the worked examples above, so you can see the chicken plate reach 29.1% and the month reach 34.1% before you enter your own numbers. Excel or Google Sheets, no macros.
The food cost formula, how to cost a plate with real yields, theoretical versus actual cost, and the levers that bring the number down.
Food inflation rose again in August. What is pushing kitchen costs up, and how to reprice a menu without driving regulars away.
Sort every dish into Stars, Plowhorses, Puzzles and Dogs using your POS export and our free worksheet, then decide what to reprice, move or cut.
For a single dish: the cost of everything on the plate divided by its menu price, times 100. For a restaurant over a period: opening inventory plus purchases minus closing inventory, divided by food sales for the same period, times 100. The first tells you whether a dish is priced right; the second tells you what the kitchen actually used.
As operator rules of thumb: 28–32% for QSR, 30–35% for casual dining and 35–40% for fine dining. The calculator treats 32% or under as healthy, 32–38% as tight and above 38% as underwater. Read direction from these rather than a grade — a restaurant with low labor can carry a higher food cost, and prime cost is the number that has to work.
Cost of food used = opening inventory + purchases − closing inventory. Divide that by food sales for the same dates and multiply by 100. It only works if both inventory counts are taken on the dates the sales period starts and ends, and if purchases are booked when delivered rather than when invoiced or paid.
Theoretical food cost is what the kitchen should have used: each dish’s recipe cost multiplied by how many were sold. Actual food cost is what the inventory says was used. The gap is waste, over-portioning, theft, unrecorded comps and staff meals, and invoices that changed price since the recipe was costed. It is the most useful number in food cost, because it points at something you can fix.
Many operators aim to keep actual within a point or two of theoretical. That is a rule of thumb, not a standard. More useful than the size of the gap is its direction over time: a gap that widens month on month is a control problem, and one that jumps in a single month is usually a counting or invoicing error.
Divide the plate cost by your target food cost percentage. A dish costing $4.36 with a 30% target needs a price of about $14.53, which you would round to a price point such as $14.95. That gives a floor, not a price — check it against what the dish sells for nearby and against its popularity before settling.
Yes. When you set service format to delivery in the calculator, packaging cost (containers, lids, cutlery, bag) is added to ingredient and garnish cost before food cost % is calculated. For dine-in, packaging is left out. Platform commission is not food cost — it belongs in the delivery margin.
Yes, and they are the most common reason a costed recipe comes out low. Cost every ingredient at its usable price: the purchase price divided by the yield after trimming, cooking or portioning. Chicken bought at $6.00 a pound with an 80% yield really costs $7.50 a pound on the plate.
They count in actual food cost whether you record them or not, because the food left the inventory. Record them separately so you can see them — staff meals and comps as their own lines, waste on a waste sheet — otherwise they turn up as an unexplained gap between actual and theoretical.
Track them separately. Beverage and bar cost runs on different margins from food, so a combined figure moves when the sales mix shifts between them even if nothing in the kitchen changed. Report food cost against food sales and beverage cost against beverage sales.
Gross margin is menu price minus total food cost — the dollar amount left over. Gross margin % is that amount divided by menu price. It is what remains before labor, rent and utilities are paid. A dish with a higher food cost % can still earn more dollars per plate than a cheaper one, which is why menu engineering looks at both.
Count the high-value items weekly — proteins, dairy, anything expensive and perishable — and the full inventory monthly. A monthly-only count tells you there was a problem four weeks after it started. A weekly count of the ten items that make up most of your spend catches it inside a week.
What you net per delivery order after commission, packaging and discounts — against dine-in margin in real time. The gap is the cost of being on the platform.
Rent, wages, ticket size and variable cost worked back to the daily covers number every operator should know cold.
Wages as a share of revenue, benchmarked against the band for your service format — QSR, casual or fine dining.