How to calculate restaurant food cost: the 30% myth and what really decides your margin

Step 1 · Build the recipe card before you touch a calculator

The recipe card is the only document that turns a dish into a number, and without it any food cost you calculate is an opinion with decimals. Write down every ingredient with its NET weight —what lands on the plate, not what comes through the door—, the cost per purchase unit and the yield after peeling or trimming: a tenderloin bought at 18 USD per kilo with 22% cleaning loss actually costs you 23.08 USD per usable kilo, and that 5 USD gap per kilo eats your margin without ever showing up on an invoice. The deliverable is simple: one card per menu item, signed by the chef, with total ingredient cost on the last line. You verify it by weighing the finished plate on a scale three times and confirming the grams match the card within a 5% margin.

Step 2 · Theoretical food cost: ingredient cost divided by selling price

Theoretical food cost for a dish comes from dividing the ingredient cost on the card by the pre-tax selling price, times one hundred. A risotto that costs 4.60 USD to produce and sells for 16 USD yields a theoretical 28.75%, and that figure tells you what you SHOULD spend if your kitchen executed the card with laboratory precision. It never will, and that is exactly the point: theoretical is the yardstick reality gets measured against. Calculate one for every active dish and sort them high to low on a single sheet, because your problems show up in that order. The 32% reference we use at Masterestaurant is the CEILING, not the target, and it applies to the individual plate; starters usually land at 18-24% and protein-heavy mains push toward 35% when nobody is watching them.

Step 3 · Actual food cost: opening inventory plus purchases minus closing inventory

Actual food cost for the period is calculated this way: opening inventory plus purchases minus closing inventory, divided by food sales for that same period, times one hundred. If you opened the month with 12,400 USD across walk-in and dry storage, purchased 38,900 USD, closed at 11,100 USD and sold 118,000 USD in food, your consumption was 40,200 USD and your actual food cost is 34.07%. The denominator shifts from the theoretical one —here it is TOTAL food sales, not the price of one dish— and mixing the two denominators produces that odd number matching nothing, which is why so many owners abandon the metric after two months. Count inventory on the last day of the month after service closes, always at the same hour and with the same two people counting.

Step 4 · Subtract theoretical from actual and name the gap

The gap between actual and theoretical is the one number in this whole guide that will hand you money back this week. Take your sales-mix weighted theoretical —the average of your cards weighted by units sold per dish— and subtract it from actual: if weighted theoretical is 29.8% and actual is 34.07%, those 4.27 points over 118,000 USD in sales are 5,038 USD that evaporated without ever reaching the register. A gap under 1.5 points is a healthy operation; anything above 3 points is a structural problem, not a bad month. The deliverable is one line in your monthly report with three columns: weighted theoretical, actual, and gap in points and in dollars. Verify it by rerunning the math against POS sales exported from the system, not against what your manager remembers.

Step 5 · Break waste down into its four owners

Waste is not one thing, which is why one measure never fixes it. There is PREP waste —peeling potatoes, trimming the loin—, measured by weighing scrap per shift and usually worth 1 to 2 points; PORTIONING waste, the cook plating 220 grams where the card says 180, a 22% overrun on that ingredient, corrected with a calibrated ladle and a scale on the line; SPOILAGE, whatever expired in the walk-in, attacked with FIFO rotation and a weekly count of your ten costliest items; and SERVICE waste, the plate that comes back from the floor, logged with a reason. Assign each one to a person by name and a different verification, because waste without an owner never gets corrected: it only gets reported.

The mistakes owners repeat most when calculating food cost

The most common mistake is calculating against tax-INCLUDED selling prices, which inflates the denominator and gifts you two or three points of food cost that do not exist. Second: forgetting that the price on delivery platforms is not the one on the printed menu —if Rappi withholds 28% commission and you loaded the same price, that 16 USD dish nets you 11.52 USD and your actual food cost on that channel jumps from 28.75% to 39.9%—, so calculate food cost by channel or you go blind exactly where you sell most. Third is loading payroll, rent or utilities onto the plate: those costs belong to break-even, never to the recipe card. And fourth, benchmarking your 34% against the 30% from a 1990s manual written for table service with no delivery.

Cross your food cost against what you promote on social

Last month's highest-viewed Reel is probably the dish with your worst margin, and that is not bad luck: photogenic plates carry expensive inputs and generous portions because they were designed for the camera. When content pushes demand toward the wrong dish, consolidated food cost climbs two or three points without anyone changing a recipe, and per Marketing LTB bookings rise 30% in the week following a creator's post, so the effect is far from marginal. Diego F. Parra has insisted on the same order for twenty years: measure first, price second, and only then decide what to promote. Cross your card list sorted by food cost against your publishing calendar and you will see, in one afternoon, which content is funding the decline of your EBITDA.

Closing checklist · how to know everything landed right

You know the calculation landed when you can answer five things without opening another file. One: you have a signed recipe card for 100% of active dishes and the real weight matches the paper within 5%. Two: your sales-mix weighted theoretical was built with POS units from the same period, not a different one. Three: opening and closing inventory were counted at the same hour of day and under the same valuation criteria. Four: the actual-minus-theoretical gap sits under 1.5 points, or you have written down which type of waste each excess point belongs to. Five: a food cost by channel exists, because the National Restaurant Association 2024 documents a 35% rise in food costs since 2019 and that headwind is not offset by an average number. Repeat the full cycle the first Monday of every month.

Where the calculation actually breaks?

The real difference is not the formula, it is the denominator. Theoretical food cost divides by the menu price of ONE dish; actual food cost divides by total food sales for the period.

Mixing the two denominators produces that odd figure matching nothing, which is why so many owners abandon the metric after two months. Waste is not one thing, so it never yields to one fix. There is prep waste (peeling the potato, trimming the loin), portioning waste (the cook plating 220 grams where the card says 180), spoilage waste (what expired in the walk-in) and service waste (the plate sent back).

Where the calculation actually breaks — in practice?

Each has a different owner and a different verification. The price on your physical menu and the price inside the delivery QR menu should NOT be the same number, because platform commission eats between 18 and 30 points.

Calculate food cost with dining-room pricing while selling through apps and you are measuring a margin that does not exist. Sales mix outranks the percentage. A consolidated 29% with 60% of sales concentrated in one low-margin dish is more fragile than a 33% spread across four healthy plates. The metric is never read alone: it is read against rotation and against contribution margin in currency.