Plate costing: the myth that costs you twelve points of margin

Before the first formula: three sources on the table and a calibrated scale

Costing without the three sources is not costing, it is drafting a hypothesis: you need eight weeks of purchase invoices priced at what you actually paid, the per-dish sales report for that same window expressed in UNITS sold rather than revenue, and a one-gram digital scale sitting in the kitchen. The deliverable of this step zero is physical, one folder holding those three files plus a calibrated scale, and the checkpoint is equally concrete: the report must cover at least 500 tickets and 100 % of the active menu. First money gets lost here for a reason that sounds minor and is not, taking the supplier list price instead of the price paid net of discounts and credit notes, a gap that inflates cost by 4 % to 7 % before you have written a single recipe. Yield gets weighed, never assumed, and that single discipline explains much of the twelve margin points we argue about at Masterestaurant with owners who swear their costing is current.

Step 1. Close the spec sheet on MEASURED yield, and keep the worst of three

Write every recipe in grams and milliliters, no pinches and no 'to taste', and for any input that gets trimmed, thawed or cooked, weigh the raw product and weigh the portion-ready product, three times, with three different cooks. Keep the WORST figure rather than the average, because the average describes a cook who does not exist. A one-kilo tenderloin yields between 680 and 760 clean grams; costing the kilo purchased instead of the usable kilo understates real cost by 24 % to 32 % on premium protein. Deliverable: a spec sheet per dish carrying unit cost with the yield factor applied. Checkpoint: 90 % of the menu closed, zero estimated factors on meat and seafood. Divide the dish's input cost by its pre-tax selling price and you hold theoretical food cost, though the number only turns useful once you group it by menu family, because the healthy ceiling shifts across the card: pasta and rice near 24 %, premium protein up to 32 % and not one point beyond, desserts usually below 20 %.

Step 2. Food cost per dish and per family, with 32 % as a ceiling and never a target

An external reference exists and deserves a straight look, the National Restaurant Association places US sector food and beverage cost at 33.2 % of sales, which makes a 32 % dish the outer edge of tolerable. Deliverable: a food cost table sorted high to low. Checkpoint: no individual dish above 32 % and a units-WEIGHTED average between 27 % and 30 %. Averaging without weighting hides exactly what hurts, that your most expensive dish to produce is usually your bestseller. Anything that does not physically enter the plate leaves the calculation, and this border is operational before it is accounting: kitchen and floor payroll, rent, utilities, licenses, software, maintenance and equipment depreciation all get covered at the monthly break-even, not inside each dish. CapEx — the walk-in, the remodel, that convection oven — goes to cash flow and depreciation; recurring OpEx goes to the monthly management P&L. And watch the rent, which rarely travels alone: common area maintenance fees add another 2 % to 3 % on top of base rent according to 7shifts.

Step 3. The $14,000 oven does not live in the plate: CapEx and OpEx belong to break-even

Add up every fixed cost for the month, divide by average contribution margin per guest, and you have break-even in covers. Checkpoint: that break-even stays under 72 % of your historical covers for the same month last year. Count physical inventory on the same weekday, at the same hour, always, and with the kitchen closed, because counting during service injects up to 2 points of noise that ruin the entire measurement. Actual consumption comes from an old, reliable subtraction, opening inventory plus purchases minus closing inventory, set against the theoretical consumption your spec sheets produce once multiplied by units sold. That difference in percentage points is your capital leakage map, and it carries concrete names: theft, over-portioning, badly weighed receiving, walk-in spoilage. Scale helps here, ReFED estimates US restaurants generate roughly 11.4 million tons of food waste a year and that 78.4 % of foodservice waste went to landfill in 2024.

Step 4. Weekly variance is the most honest number in the whole exercise

Checkpoint: total variance under 1.5 points; above 3, audit receiving and portioning before touching the menu. A 31 % dish leaving $11 beats a 24 % dish leaving $3, and that arithmetic gets ignored daily on menus across the continent. Plot two axes per dish, contribution in dollars — price minus input cost — and popularity measured as units over total, then place each one in its quadrant. High-contribution, high-popularity dishes are your workhorses: top right on the card, with a photo where the design allows, recommended by the floor team by NAME rather than by category. Low-contribution, low-popularity dishes come off without sentimentality. Deliverable: a four-quadrant matrix carrying one written decision per dish. Checkpoint: your four highest-contribution dishes account for at least 22 % of units sold by the close of month two. The costliest error is not arithmetic but boundary, allocating fixed costs inside the dish, because it produces food costs of 45 % to 62 % that compare to nothing and to nobody.

Five errors that bring the system down, in the order you will meet them

Four more follow, in this order of frequency: costing with last year's prices while food-away-from-home inflation accumulates 23 % over four years per the USDA Economic Research Service; averaging without weighting by sales; leaving equipment leases tangled inside OpEx, which distorts EBITDA and any serious valuation of the business; and cutting a high-food-cost dish that turns out to carry the largest absolute contribution. The fifth is more human and kills more systems than the other four combined, producing the report and never booking the thirty-minute meeting to read it. You will know the system stands when you can recite these seven figures without opening a file: 90 % of the menu with a closed spec sheet, no dish above 32 %, weighted average between 27 % and 30 %, weekly variance under 1.5 points, prime cost — inputs plus total payroll — below 65 % of sales, break-even under 72 % of your historical covers, and operating profit above 8 % at quarter close.

Closing: seven figures that tell you the costing was built right

With a typical full-service operating margin of 3 % to 5 %, according to the National Restaurant Association, that 8 % is not ambition, it is the cushion separating a business from a badly paid job. This week weigh one dish, your bestseller, and set the result against what your system claims it costs. The first is a boundary question. A dish costs what goes into the dish, full stop, and that border exists for operational reasons before accounting ones: inputs move with every sale, rent does not. Folding a fixed cost into a variable calculation destroys the only useful property food cost has, which is comparing the same recipe against itself over time. According to Peter Backman, an independent foodservice sector analyst, the discipline of separating variable from fixed costs is what lets an operator react in weeks rather than quarters. The second is yield.

Four differences that change the outcome

A one-kilo beef tenderloin does not yield a kilo of portions: it yields between 680 and 760 clean grams depending on the cut and the hands working it, and that 24 % to 32 % gap is where half of the profitability your sheet claims quietly escapes. Measuring yield means weighing the same operation three times, with three different cooks, and keeping the worst number rather than the average. Third comes variance. Costing once is an exercise; costing and then reconciling against actual inventory consumption is a restaurant cost control system. The gap between what the system SAYS was consumed and what the physical count proves disappeared is your capital leakage map: theft, over-portioning, sloppy receiving, spoilage. Skip that weekly subtraction and you do not have costing, you have an old photograph. Fourth, and the one that has cost me the most arguments with owners in board meetings, is that percentage is NOT the goal.

Four differences that change the outcome — in practice

The goal is the dollars left after paying for inputs, multiplied by how often that dish sells. Choosing a 24 % that leaves $3 over a 31 % that leaves $11 is an arithmetically losing decision, and it gets made every single day on menus across the continent.