Plate costing: the traditional method against the Masterestaurant method

Food cost percentage hides the size of the margin

Two dishes with opposite percentages produce opposite cash results, and the percentage will never tell you that. A $9 dish at 27% food cost leaves $6.57 of contribution margin; a $26 dish at 34% leaves $17.16, two and a half times more money per unit sold even though it breaks the 32% ceiling the classic manual teaches. If your fixed monthly structure demands $42,000 of margin to hit break-even, the first dish requires 6,393 covers and the second one needs only 2,448, and that gap of nearly 4,000 services decides whether the month closes in black. Costing dishes by percentage tells you to pull from the menu the very item paying the rent. Count money first, calculate the percentage afterwards, and treat it only as a warning light. In-house butchering yields run between 78% and 88% depending on the cut, and that spread turns your invoice price into a useless number for costing.

Yield loss you never measure gets paid in full

At 82% yield, a tenderloin bought at $14 per kilo actually costs you $17.07 per usable kilo, 22 points above what the invoice you type into the spreadsheet says. On a 220-gram portion that hides $0.67 of cost, and if the item sells 45 portions a day, the year swallows $11,005 that never showed up in any report. I got this wrong for years, loading a standard yield by family instead of measuring it by cut and by supplier. Weigh five real breakdowns of each protein, average them and lock the figure into the recipe card; that is half a morning of work against a five-figure leak. Spreading the cook's salary across portions produces a phantom cost that moves with volume and wrecks every menu decision you make afterwards. Full-service labor closed 2024 with a median of 36.5% of sales, while limited service came in at 31.7% (National Restaurant Association, Restaurant Operations Data Abstract 2025); those are STRUCTURAL charges, not recipe charges.

Payroll does not belong in the dish: it lives at break-even

Load 36.5% of payroll onto a dish during a slow month of 2,100 covers and the same dish looks 40% more expensive than it did in a 3,000-cover month, so you end up raising prices when what you are missing is customers. The Masterestaurant method leaves the dish carrying raw material plus direct consumables, and sends payroll, rent and utilities (2% to 5% of revenue per Toast, Average Restaurant Electricity Bill 2025) into the calculation of how many covers the venue needs to breathe. Selling through a marketplace is not selling cheaper, it is selling a different product with a different cost structure. Third-party delivery costs effectively 30% to 40% of the order total once commission, promotions and refunds are added up (OPA!, True Cost of Third-Party Delivery 2026). Apply 35% to the $26 dish from the first passage: net revenue falls to $16.90, and after subtracting $8.84 of raw material the margin drops from $17.16 to $8.06, a 53% collapse caused by the channel alone.

Delivery: the same dish needs two costings and almost nobody runs the second

That dish is still profitable in the dining room and roughly neutral on the app. The call is neither abandoning the platforms nor joining them blind: cost every item twice and publish on delivery only what survives the commission with positive margin, using channel-specific pricing whenever the contract allows it. With sector net margins running between 3% and 5% in full service, missing your costing by two percentage points stops being an accounting detail. Take a venue doing $1.2 million in annual sales: two points equal $24,000, and against a 4% profit ($48,000) exactly half the working year disappears. That two-point error walks in through three familiar doors: yield loaded at 0%, consumables left uncosted, and purchase prices frozen six months ago. Diego F. Parra and the Masterestaurant team rebuilt dish costing around contribution margin in dollars precisely because of that arithmetic, after watching managerial P&Ls close in red with 27% food cost.

Why a two-point costing error eats half the year's profit?

When the cushion is four points wide, the recipe card stops being kitchen paperwork and becomes the most sensitive financial instrument in the business.

Benchmarks do not apply the same way across three different operation sizes, and that is where most owners misuse them. In a small single-shift venue, with kitchen equipment running from $50,000 to $150,000 (Rezku, 2025) and a low average ticket, dollar margin per dish outranks everything else: review your 10 best sellers and raise price on the three with the lowest absolute margin. In a mid-size venue with two shifts and active delivery, the job is dual costing by channel, because that 30% to 40% commission decides what you publish. In a group of three or more locations, unify recipe cards and centralize purchasing; there the lever is no longer the dish but the buying price, and one point won at the negotiating table beats redesigning the whole menu.

How to read these numbers in YOUR operation?

The sequence never runs backwards. Every reference figure in this text comes from verifiable public sources, and it is worth stating how far each one reaches.

The labor numbers of 36.5% and 31.7% come from the National Restaurant Association's Restaurant Operations Data Abstract 2025, medians across U.S. operators; the 25% to 35% labor cost range is published by the Bureau of Labor Statistics for sector 722. Delivery economics come from OPA! (2026), utilities from Toast (2025) at $2.90 per square foot in electricity and $0.85 in natural gas per year, and startup investment ranges from Rezku (2025), with a $375,000 median to open. These are U.S. market references: elsewhere the levels shift, the mechanics do not. None of these figures replaces your own accounting, and treating them as if they were your operation is the second most common mistake after ignoring yield loss.

What to do on Monday with the menu you already have?

Start with the items making up 70% of your units sold, which on a 40 to 50 dish menu usually means 12 or 15 references.

For each one calculate contribution margin in dollars, not in percentage: selling price minus raw material at measured real yield minus consumables. Sort that list from highest to lowest absolute margin and cross it against units sold over the last quarter. High-margin, high-rotation dishes get protected and placed in the best spot on the menu; low-margin, high-rotation dishes get reformulated or repriced; low-margin, low-rotation dishes leave without debate. Add up the projected aggregate margin and compare it against your fixed monthly structure, including the business owner's policy that averages around $3,000 a year (MoneyGeek, 2025). If the sum does not cover the structure, your problem is not one dish: it is the entire menu. Percentage hides size.

The four differences that change the outcome

A $9 dish at 27% food cost leaves $6.57 of margin; a $26 dish at 34% leaves $17.16. The second one breaks the 32% ceiling and still pays three times more structure per unit sold. The traditional method tells you to pull the dish that keeps the register alive, and that is the mistake I meet most often in the menus that cross my desk. Real yield loss gets measured by family, never assumed. On whole protein butchered in house, yield lands between 78% and 88% depending on the cut, which means your portion cost runs 12 to 22 points above invoice price. Loading 0% yield loss is the quietest way to mis-cost every portion for a full year. Payroll does not belong on the plate. Spreading the cook's wage across portions produces a cost that rises when sales fall, an indicator that degrades exactly when you need it sharp.

The four differences that change the outcome — in practice

In the Masterestaurant method payroll, rent, utilities and recurring OpEx form the monthly fixed cost, and that block gets covered by the sum of margins rather than dish by dish. Recosting runs on a calendar. Twelve dishes concentrate around 70% of sales in a mature menu, and those twelve get recosted every 45 days against live supplier invoices. The rest can wait for the semiannual cycle. Reviewing all 48 at once, once a year, amounts to not reviewing them.