Plate profitability: the errors bleeding your menu and the method that actually fixes it

Eleven dishes selling below variable cost on a 68-item menu

A 42-seat restaurant in Guadalajara was billing 118,000 USD a month and closing at 1.9% profit, and the diagnosis sat neither in payroll nor in rent but in the menu: costing all 68 items with standard recipe and real waste showed eleven dishes selling below their direct variable cost, and the house best-seller was among those eleven. The owner defended it as a LOSS LEADER, an argument that sounds reasonable until you put the arithmetic on the table: that dish had spent three years financing itself with beverage margin, meaning beer was paying for the kitchen while the monthly report claimed everything was fine. When menu engineering is executed properly, Oracle NetSuite documents profit gains of 10% to 15% sustained over time, and here the first decision was not raising prices: it was measuring how much money each item actually left behind after its cost.

A percentage carries no units, contribution margin does

Dish profitability is measured in contribution-margin dollars multiplied by monthly units sold, never in the isolated food-cost percentage, and that distinction decides what stays on the menu and what goes. Take two items you would find on any Latin American menu: a ceviche at 24% cost turning 40 plates a month contributes 380 USD, versus a pasta at 31% cost turning 620 plates and contributing 4,960 USD. Pull the pasta because its ratio looks ugly on the spreadsheet and you have just erased thirteen times more cash than you thought you were protecting. The ratio is a number without scale, it measures purchasing efficiency but not paying capacity; absolute margin does cover payroll, which per the 7shifts workforce report climbed 4% to 14.20 USD an hour in 2024 and is not billed in percentages.

Where fixed costs land decides whether your costing starts crooked?

Payroll, rent and utilities do NOT belong inside dish costing, and allocating them there is the mistake that turns every later decision into a badly informed one.

Those line items behave per period, not per unit sold: spread 14,000 USD of monthly fixed costs across the month's units and a slow month artificially inflates the apparent cost of every dish, so you end up raising prices exactly when demand was already soft. In the Masterestaurant method a dish absorbs only what its preparation consumes —ingredients with real waste, mother sauces prorated by recipe yield, packaging if it travels— and fixed costs get resolved at break-even, which is where they belong. Diego F. Parra repeats it with a stubbornness that sometimes annoys chefs: you do not eat percentages, you eat money, and 32% food cost per dish is a ceiling, never a target.

Sales mix outranks the recipe: the levers the sector has measured

Shifting the mix toward higher absolute-margin dishes pays off faster than renegotiating with suppliers, and public sector figures back it with improvement ranges no purchasing discount reaches. A professional photograph beside a dish lifts its sales by up to 30% according to Cornell University research on menu design, with roughly 6.5% of incremental sales per photographed item; menu psychology techniques raise average check by 15% or more without touching prices, per NeatMenu's 2026 analysis. If your star dish leaves 6.80 USD of margin and sells 620 units, an extra 15% of turnover means 93 more plates a month, 632 USD straight to the profit line. That money shows up in no negotiation with your protein distributor, it shows up in how the menu is arranged.

Digital channels: the same dish leaves different money depending where it enters

One dish with an identical recipe produces different margins depending on the channel, and that gap now weighs as much as the costing itself. QSR Magazine reported in 2024 that self-service kiosk checks run 8% to 15% above the counter, with Yum measuring close to 10%; McDonald's documented average-check gains of up to 30% with kiosks, and Future Ordering logged 35% in its own case. For QR-code digital ordering, Sunday places the check increase between 20% and 30% when menu, order and payment live on the same screen. Translated into cash: if your in-person average check is 18 USD and the digital channel lifts it to 22 USD across 1,400 monthly orders, that is 5,600 USD of extra revenue with the same crew on the hot line. Suggested upsell inside the digital flow adds another 20% to 30% of order value.

How to read these numbers in YOUR operation: three scenarios?

Benchmarks only help once you land them on your own size, so here are three different readings of the same data. Small operation, under 40 seats and a 30 to 45 item menu:

cost the ten references that concentrate 60% of your units and calculate each one's monthly absolute margin; with that exercise alone, the 10% to 15% of extra profit Oracle NetSuite documents sits within reach in a quarter. Mid-size operation, 60 to 120 seats across two services: add the channel axis, because the 8% to 15% kiosk lift QSR Magazine measured in 2024, applied to 2,000 monthly orders, is between 2,800 and 5,400 USD a year you are not collecting today. Group of three or more locations: standardize the spec sheet before comparing anything, because without a common recipe you are measuring cooks, not dishes.

Where these benchmarks come from and how far they reach?

The figures above come from three kinds of public source, and it is worth saying plainly what each can and cannot do.

Operator data —McDonald's, Yum, the Future Ordering case— are reported results from chains with volume and proprietary technology, so their magnitude is real but transferring it to a 42-seat independent demands a discount; Cornell's academic research on menu design and the 7shifts workforce report carry open methodology and broad samples, though their base is largely United States and the 14.20 USD hourly wage of 2024 does not travel to Latin America without adjustment. Numbers from technology vendors like Sunday or NeatMenu describe their own customers, that is, businesses that already chose to invest. None replaces your P&L. Use them as a reasonable expectation range, never as a promise.

The fix starts Monday, with one sheet and four columns

Sort your menu this week using four columns and the exercise wraps up in two hours: item, direct variable cost per unit, selling price, units sold last month. You calculate the fifth column yourself —price minus cost, times units— and that dollar figure is the only one that decides. Whatever comes up negative leaves the menu or gets its spec sheet redesigned; whatever brings thin margin and thin turnover is occupying storage space and hot-line minutes your champion dish needs. What would happen if tomorrow you pulled the eleven items that fail to cover their own cost? Your menu drops from 68 to 57, purchasing concentrates, waste falls because fewer dead ingredients sit in the walk-in, the cook stops chasing fifty preparations and mix margin rises without a single price having moved. Start there.

The three differences that change the outcome

The first difference is arithmetic and it is brutal: food cost percentage is a ratio without scale, while contribution margin is hard cash. A ceviche at 24% cost selling 40 plates a month contributes 380 USD; a pasta at 31% selling 620 contributes 4,960 USD. Pull the pasta because its cost looked worse and you just deleted thirteen times more cash than you protected. The second difference sits in where fixed costs land. Prorate payroll, rent and utilities inside the dish and the apparent cost explodes, so every downstream decision is born crooked, because those lines behave per period rather than per unit sold.

The three differences that change the outcome — in practice

Under the Masterestaurant method, only what is consumed preparing that dish enters the dish, and fixed costs get settled at the monthly break-even, which is where they actually live. And the third one, which hardly anybody works on, is the physical position of the dish inside restaurant menu design. Marginal profitability per dish does not hang on the spec sheet alone: it hangs on how often the guest picks it, and that is governed by visual hierarchy, by the boxed feature at the top right, by reading order, and by the absence of aligned price columns that invite guests to scan downward and pick the cheapest line.