Plate costing: knowing what the dish costs is worthless if you know it late

1. Why does a flawless recipe cost sheet still lose money?

A recipe cost sheet loses money the moment its date is older than the last move in your inputs, and those moves now run faster than any quarterly calendar.

The producer price index for all foods in the United States closed May 2026 some 35% above the February 2020 level, according to USDA ERS / BLS (2026), while in Colombia menu and product prices climbed 9,8% since February 2025 according to ACODRES (2025). An operator in the 500 thousand to 1 million dollar band showed me a spreadsheet in January with 62 dishes costed to the cent; the file dated from September, the main protein had moved twice and cooking oil once, and the dish he believed sat at 29% food cost was running at 38%. He was not losing money from ignorance. He was losing it from the CALENDAR. Arguing about food cost percentages without crossing them against table turns is the most expensive trap in this trade, and I recommended the opposite for years before correcting it.

2. The decision unit is not the percentage: it is dollars per dish

A dish at 34% selling one hundred twenty units a week drops far more cash than one at 26% selling nine, and the percentage alone cannot separate them. At Masterestaurant we cost in contribution margin dollars per dish multiplied by units sold, then decide what gets defended and what gets pulled. The ceiling remains 32% food cost per dish as a MAXIMUM, never as a target. Alcohol confirms this from another angle: 46% of respondents surveyed by Technomic / Nation's Restaurant News (2024) name it among the highest-margin menu categories, and no serious operator judges that category on an isolated percentage. Below 500 thousand dollars in annual revenue, the owner needs no continuous recosting software; discipline over four inputs will do. Pin down the four that carry 60% or more of your purchasing spend — usually the main protein, oil, dairy and whatever commodity is spiking — and check them every fifteen days against invoices, not memory.

3. Under 500 thousand a year: four inputs, reviewed every two weeks

With food producer prices 35% above February 2020 according to USDA ERS / BLS (2026), a fortnight captures roughly 80% of the risk at zero technology cost. The action threshold stays simple: any input that moves 7% or more triggers a same-day recost, and price, portion or removal gets decided before the next weekend. This band does not drop out of the analysis, and whoever treats it as a minor case is usually the one who later buys it cheap. Between 500 thousand and 1 million dollars a year sits the crossover where the quarterly sheet costs more than the tool replacing it. At that volume, three points of stale food cost across one quarter equal somewhere between 3.750 and 7.500 dollars of margin that never comes back, and the January operator was running roughly nine points of drift on his best seller. The threshold here is monthly with a variance trigger: a full recost every thirty days, plus an immediate recost of any dish whose dominant input moves 5%.

4. 500 thousand to 1 million: quarterly latency starts eating EBITDA

Arabica coffee makes the warning concrete, having risen 70% during 2024 according to Bellwether Coffee, with a combined 50% United States tariff on Brazilian coffee in 2025 from that same source: anyone who costed in January and looked again in April sold every cup below replacement. Past 1 million dollars a year, costing stops being an office matter and moves into the preshift, because margin gets defended in the server's mouth and not on a spreadsheet. Your floor team should walk in each morning knowing which two dishes to push and which one to leave alone, with the day's number already loaded. That training pays hard: every avoided departure saves 150% of the salary in replacement costs according to StaffedUp (2025), and AI-assisted scheduling cuts labor cost between 8% and 12% with forecast accuracy above 90% according to TimeForge (2025). Add the delivery channel, where 37% of adults order at least once a week according to UpMenu (2024), carrying commissions that sink the margin of any dish costed as though it were dine-in.

5. Above 5 million and the high-end case

Above 5 million dollars a year, the decision is no longer whether to recost: it is who governs the recosting, with a named owner and a weekly board that reaches the committee. This range holds the large-format themed venue and the restaurant signed by a media chef, where the menu sells on experience and input cost gets neglected for exactly that reason, with real food cost drifting ten or twelve points off budget while nobody notices for half a year. Diego F. Parra runs the Masterestaurant framework against the usual habit: contribution margin in dollars per seat and hour first, menu second. The scale reference exists — Chipotle projected 315 to 345 openings for 2025, more than 80% with a drive-thru, according to Chain Store Age (2025) — and no chain grows that way on ninety-day sheets. In groups and chains above 10 million dollars a year, continuous recosting stops being an operational upgrade and becomes a valuation condition, since the multiple pays for margin predictability rather than the annual average.

6. Above 10 million: the group buys recosting, not dishes

My threshold there runs daily by exception: the system flags any SKU that moves 3% and corporate decides within 48 hours across the whole network. Starbucks opened 589 net stores in 2024, reaching 16.935 units according to QSR Magazine (2024); at that scale, one stale food cost point over a quarter is an eight-figure number. What happens if such a group keeps its quarterly sheet two more years? With food producer prices 35% above February 2020 according to USDA ERS / BLS (2026), it stacks six cycles of dead pricing and arrives at the negotiating table explaining why its EBITDA will not hold. Open your cost sheet today and look at the file date before you look at any number, because that date tells you how much money you are deciding blind. If it is older than thirty days, pull the five dishes that sell the most units, request the last three invoices for your dominant input, and recost those five this week; the rest can wait.

7. This week's move

The trade's tension resolves this way and no other: precision and frequency appear to compete for the same manager hours, yet a rough cost from yesterday governs better than an exact one from September. With food producer prices 35% above February 2020 according to USDA ERS / BLS (2026) and 9,8% increases in Colombia since February 2025 according to ACODRES (2025), the only variable you control is how often you look. Look every thirty days. The first difference is FREQUENCY, and it moves the most money: the quarterly card assumes price stability in a market where the food producer price index sits 35% above February 2020 per USDA ERS / BLS (2026), while continuous re-costing fires on input variance rather than on a date. Second comes the UNIT of measurement. Traditional costing argues about percentages; the Masterestaurant method argues about dollars per dish crossed with table turnover, which is the only way to judge whether to defend a 34% food cost dish selling a hundred and twenty covers a week or protect a 26% dish selling nine.

8. Four differences a CEO should underline

Third, costing stops living in the office and moves onto the floor: automated preshift and the Interactive Training Kit simulators turn the recipe card into a sales script, and the server shifts from taking orders to steering average check mix toward high-contribution items. And fourth, the one that decides whether the group scales: a master card with per-site local costs lets you compare the same dish across locations, expose capital leakage from yield differences, and run operational due diligence before committing CapEx to the next unit.