Food cost 2026: the numbers the traditional method never sees

The wrong denominator: why your 29% is not food cost

Dividing the month's purchases by the month's sales does not give you food cost, it gives you a purchasing rate, and that confusion props up half the wrong diagnoses that reach my desk. The casual-dining spot in Bogotá I mentioned closed March with 218 million pesos in sales and 9 million in profit, a 4.1% that happens to resemble the 4.0% pretax profit reported as the limited-service median in the National Restaurant Association's Restaurant Operations Data Abstract 2025, using 2024 data. The owner celebrated his 29%. What he actually had was inventory noise: buy three weeks of beef in the last five days of the month and the indicator spikes without the kitchen changing a single gram of any recipe. Plate-level costing uses real consumption against a standard recipe, which is why it describes what left through the pass rather than what came in the back door.

What is the real per-plate food cost ceiling in 2026?

The ceiling is 32% of the plate's selling price, and that 32% is already the maximum tolerable figure, not the target.

At Masterestaurant we work against 28-30% on a mature menu, because the contribution margin left over has to pay payroll, rent and utilities before reaching break-even, and those three line items are NOT charged to the plate. The arithmetic is blunt: with the 2.8% pretax profit the National Restaurant Association measured as the full-service median in 2024, a dish that drifts from 30% to 35% eats the entire month's profit without a single odd invoice crossing your desk. And here I was wrong for years, recommending the menu average as a management indicator: the menu average is weighted by what the guest orders, not by what you wish you sold, and that mix is moved by the app's algorithm, not by your printed menu.

The channel bleeding you dry: 27.4% dine-in, 36.8% delivery

That same dish at 30% food cost in the dining room reaches an effective 39% once it passes through a marketplace charging 24% commission, 1,800 pesos of packaging and a free-delivery promotion you half-subsidize. When we split the Bogotá operator's calculation by channel, dine-in came in at 27.4% and delivery at 36.8% against the price the guest paid in the app: that 29% average was hiding nine points of bleeding behind the channel that was saving it. Platform commission across Latin America runs between 18% and 30%, an order of magnitude roughly ten times the 2.36% blended Visa and Mastercard interchange rate The Motley Fool reports for the United States in 2025. The comparison matters because plenty of owners treat marketplace commission as a minor financial cost, when it behaves like a silent partner with its own menu. Mini-conclusion: if you do not calculate food cost by channel, you are not managing, you are gambling.

The waste nobody invoices: 14% of sales in surplus food

Foodservice surplus food was worth 157 billion dollars in 2024, equivalent to 14% of the sector's sales, according to ReFED. Translate that 14% into your own operation: every point of waste you recover goes straight to contribution margin, with no customer acquisition cost and no discount to negotiate. And waste does not live where people look for it; it lives in the overportioning nobody weighs, in the product bought by volume for a discount your turnover cannot support, and in the menu item the cook assembles by eye because the standard recipe was never written down. Run the counterfactual: if the Bogotá location cuts waste from 14% to 9% on its 218 million in sales, it recovers roughly 10.9 million a month, more than the 9 million in profit it reported. So the biggest profitability project in that restaurant is not raising prices or fighting the app's commission, it is in the garbage bin.

The waste nobody invoices: 14% of sales in surplus food — in practice

That is where I would start on Monday. A 32% food cost weighs differently depending on check size, and that is the tension almost nobody resolves. One Haus data for 2025 in the United States puts quick-service average checks between 8 and 12 dollars per person, fast casual between 11 and 16, casual dining between 15 and 35, and fine dining above 60, with 50 to 150 as a common range. In a 10-dollar QSR, 32% leaves 6.80 dollars to absorb payroll, packaging and rent across high transaction volume; in a 90-dollar fine-dining room, the same percentage frees 61 dollars per guest, yet specialized payroll and slow table turns devour it just as easily. The conclusion comes before the premise here: a percentage without the absolute margin per guest decides nothing. Calculate contribution margin in currency per plate and per service hour, not only the percentage you enjoy showing people.

What operators say: 41% of orders no longer pass through your till

According to Rodrigo Chaves, operations director of the Colombian Association of the Gastronomic Industry (ACODRES), most independents in the region set menu prices once a year and never readjust them by channel, so the marketplace ends up selling at dining-room prices with a delivery cost structure. With 41% of Latin American independent restaurant orders arriving through a digital marketplace in 2026, that stopped being an accounting detail and became the majority of the sector's growth operating under the wrong price. Scale context helps you locate yourself: Mexico's restaurant industry was worth 300 billion pesos in 2024 and accounts for 12.2% of the country's businesses per CANIRAC and INEGI, while Canadá billed CAD 96.5 billion in food and beverage services in 2024, up 4.0% versus 2023, per Statistics Canadá. Mini-conclusion for this group: the menu price and the app price are two prices, and treating them as one destroys more profit than anything else today.

Credit risk reads your food cost too

Regional variation in SBA loan default rates for restaurants reaches 8.7 percentage points, per Crestmont Capital's analysis of default rates by industry for 2026, and that dispersion is not geography, it is management. A bank looking at per-plate food cost documented against standard recipes, with channel-level costing and measured waste, is looking at a different risk than an owner who presents a purchases-over-sales ratio and calls it an indicator. I will say it without hedging: food cost stopped being a kitchen number the day it became the variable that explains whether you refinance at 14% or at 22%. And here is the paradox of the trade: the cheapest indicator in the restaurant to calculate, since a standard recipe costs one afternoon of work per dish, turns out to be the most expensive one when it does not exist, because without it there is no supplier negotiation, no price defense, and no credible income statement in front of a credit committee.

The 3 numbers you should tattoo on yourself

Three numbers, and with each one the action it triggers. First: 32% per-plate food cost over selling price as an absolute ceiling, calculated with standard recipes and real consumption. Action: this week, cost the ten dishes that make up 70% of your sales and pull or reformulate any that crosses 32%. Second: 18-30% marketplace commission, against the 2.36% Visa and Mastercard interchange rate for 2025 reported by The Motley Fool. Action: build a delivery menu with its own pricing that absorbs commission, packaging and promotion, and remove from the app every dish whose effective food cost exceeds 39%. Third: 14% of sales tied up in surplus food, per ReFED 2024. Action: weigh the hot line's waste for fourteen consecutive days and compare it against recipe; whatever number appears will tell you how much profit you are throwing away. Start with the first one Monday, with a scale and one sheet per plate.

Three differences that move cash

FIRST, the denominator. Dividing purchases by sales gives you a purchasing rate, not a food cost. Buy three weeks of beef in the last week of the month and your indicator spikes while nothing at all happened in the kitchen. Per-dish costing uses the actual consumption of the standard recipe, so the number describes what left the pass, not what came through the back door. SECOND, the channel. A dish running 30% in the dining room becomes an effective 39% once it goes through a marketplace with 24% commission, $1.20 of packaging and a free-delivery promo you half subsidize. According to Rodrigo Chaves, operations director at ACODRES, the Colombian gastronomic industry association, most independents moved their full menu to delivery without recosting it and found the hole in cash flow rather than in the P&L. Masterestaurant builds a shorter channel menu with the dishes that survive commission and photograph well in the app.

Three differences that move cash — in practice

THIRD, what belongs on the plate. I got this wrong for years, and I will say it plainly: I used to teach owners to spread labor and rent across dishes to know 'the real cost'. It is an accounting error with operational consequences, because a dish that appears to cost 58% cannot be managed and the owner ends up raising everything 8%. Only what is consumed preparing the dish belongs to the dish; labor, rent and utilities are structural costs paid out of break-even. The house rule: 32% maximum per dish, and 32% is the CEILING, never the target.