Menu design: the 7 mistakes that drain your margin and the method that fixes them

Step 1: cost every portion at this week's prices, before you open a design file

Cost all 46 items on your menu at this week's invoice prices, dish by dish, and do not touch a typeface until that sheet is closed. The deliverable is a table with cost per portion, current selling price and percentage food cost for EACH dish; you verify it by adding up the ingredient cost of three random dishes against the invoices from your last order, and if the gap exceeds 5% the sheet is worthless. Costing expires and nobody sends a warning: restaurant menu prices rose 4,1% in 2024 and 3,8% in 2025 according to the USDA Economic Research Service, while in Colombia ACODRES reported a 9,8% increase in dish prices since February 2025. A costing sheet from two years ago carries, by simple arithmetic, somewhere between eight and fourteen points of accumulated error. Export twelve months of units sold per item from your POS and paste them beside the food cost column.

Step 2: cross the cost with units sold from your POS, which is where the money shows up

That cross is the step separating a profitable menu from a pretty one, because your restaurant's aggregate food cost tells you absolutely nothing: a global 33% can hide four dishes running at 48% offset by beverages at 18%, and alcohol shows up as the highest-margin category for 46% of operators surveyed by Technomic for Nation's Restaurant News in 2024. The deliverable here is a four-quadrant matrix with contribution margin in currency on one axis and units on the other. Verification is brutal and fast: dishes in the high-rotation, low-margin quadrant should account for less than 15% of your total units. If they account for 71%, as in the Bogotá case opening this guide, your menu is working against your till. Bring the menu down to a range of 18 to 24 items and calculate the average food cost WEIGHTED by units sold, not the simple average.

Step 3: cut down to 18-24 items with a weighted average food cost under 32%

The difference between the two is the trap almost everyone falls into: a simple average treats the dish selling 400 units a month exactly like the one selling nine, which is why a menu that averages 30% on paper bills at a real 36%. With 46 items you hold dead inventory, product waste on a fifteen-day rotation and a kitchen that cannot standardize anything. The absolute ceiling for food cost per dish is 32%, and that number is a MAXIMUM, not a target. The deliverable is the final signed list with the weighted figure calculated at the bottom. You verify it by recalculating that figure with real units from the month after the cut, which almost always rise on the dishes that survived. Set each price starting from the contribution margin that dish needs to deliver, and only then look at what the competition charges. The reverse sequence —market price first, costing later— is what produces the March scene where you discover twelve dishes selling at a loss.

Step 4: set prices from contribution margin, not from what the place next door charges

A dish with a portion cost of 9.800 pesos and a 30% food cost target sells at 32.700, and if your neighborhood market will not carry that price the decision is not to lower the price: it is to change the recipe, the gram weight or the side until the cost comes down. I got this wrong for years, recommending you round prices down for psychological reasons. The deliverable is your price list with the margin in currency noted beside each one. Verify it by multiplying margin times projected units and contrasting the total against your monthly break-even. Design the menu only once the numbers are closed, and use it to steer the eye toward the high-margin dishes you already identified in the matrix. At Masterestaurant we treat the menu as a financial instrument with a visual format: the number first, the lettering after.

Step 5: the menu as a financial instrument with a visual format, in that order

That means placing the four or five highest contribution margin dishes in the top third of each block, dropping the repeated currency symbols, avoiding the right-aligned price column that invites people to compare downward, and reserving the highlighted box for the dish that earns most, not for the chef's favorite. Diego F. Parra has spent twenty years running this routine in restaurants across 43 countries and the order never changes. The deliverable is the approved final artwork. Verification takes thirty seconds: hand the menu to someone outside the business and ask which three dishes they remember. The most expensive mistake is printing. A batch of menus on 300-gram paper armor-plates your price structure for the eight or twelve months that batch lasts, and with input costs moving nearly 10% a year you are signing off on losses in advance; use sleeves with an interchangeable price sheet or a QR code.

Step 6: the four mistakes that wreck the execution of this guide

The second mistake is loading payroll, rent and utilities onto the plate cost: those belong in your break-even, never in food cost, and doing it inflates the price until it leaves the market. The third is measuring food cost only at the global level. The fourth, quieter one, is forgetting allergen declaration: the FDA identifies eight food groups responsible for 90% of food allergies, and a menu without that labeling is a legal exposure that costs nothing to close. Check all four before sending anything to production. Your menu came out right when you can answer these six questions with a number rather than an impression. One: how many items does it carry? It should sit between 18 and 24. Two: what is the average food cost weighted by units sold? It must land below 32%. Three: does any individual dish exceed 32%? If so, either the recipe changes or its presence gets justified by traffic volume.

Step 7: how to know the menu came out right, closing checklist

Four: are the five highest contribution margin dishes sitting in the preferred sight zone? Five: does the costing sheet carry a date from this month? Six: is there a declaration covering the eight FDA allergens? Put the quarterly review on your calendar right now, with a date on it, because this routine runs four times a year and not once for the business plan. Tomorrow, start with step one: your three best sellers, costed against this week's invoice. The first difference is one of sequence, and it costs the most money. An improvised menu starts with graphic design and ends with the number; the correct method starts with per-portion costing and only then allows anyone to open a design file. Reverse that order and the print run armor-plates your error for the eight or twelve months that batch lasts, and you pay for that armor every single day in food cost points.

The five differences that decide the margin

The second is granularity. Measuring the restaurant's global food cost decides nothing: an aggregate 33% can hide four dishes at 48% offset by beverages at 18%. Marginal profitability per dish, crossed with POS unit sales, is the only thing that tells you what to cut. Without that cross, every menu decision is a well-meant opinion. The third difference is called demand elasticity and almost nobody works it. Some dishes take a 3,000-peso increase without losing a single unit, and others lose 20% of sales over a 1,000-peso move. You only learn which is which by testing, in four-week windows, one variable at a time. I got this wrong for years, recommending flat 8% increases across the whole menu, until data from three clients showed me that a flat rise destroys the workhorses and leaves the dishes nobody orders untouched. The fourth sits in the carrier.

The five differences that decide the margin — in practice

A well-built physical menu lets the server walk the guest through the reading, suggest the pairing and close the sale with a finger on the paper. The QR does none of that: the guest opens their phone, sees a WhatsApp notification and mentally leaves the restaurant for ninety seconds. Which is why the house rule is BOTH, with distinct roles, not one replacing the other. The fifth is cadence. Restaurant menu design revised when the paper runs out is design that reacts; design audited every 90 days with a control figure per step is design that steers. The gap between reacting and steering, measured in a 200-million-a-month operation, runs between 6 and 9 million in monthly profit.