Profitable menu: criteria to build it, and what changes in the till before and after

Step 1 · Write the standard recipe for every dish before touching a single price

The standard recipe is the first criterion of a profitable menu and it comes BEFORE costing, because a cost calculated on an imaginary portion cannot support any decision. The deliverable is concrete: one spec sheet per dish with the weight of every input, yield of the cut, expected waste and plating procedure, signed off by the head chef. You verify it by weighing three services of the same dish on different days: if the gram deviation between them exceeds 5%, the sheet exists on paper but was never implanted in the kitchen. In the MIPYME strengthening programs SATE Institute runs with gastronomic cooperatives, a broken chain between recipe and costing produces an error of 12% to 18% in cost per portion, and that error travels untouched all the way to the printed price. With food inputs up 35% since 2019 according to the National Restaurant Association, no operator has room for that margin of error anymore.

Step 2 · Cost per portion against this month's invoices, with a 32% food cost ceiling

Costing per portion is done against the current month's invoices, never against last year's supplier list, and the maximum food cost per dish is 32%: a ceiling, not a recommended target. One point deserves to be explicit because the mistake repeats in every cost audit: payroll, rent and utilities are NOT charged to the plate, they belong to the venue's break-even; loading them onto the dish inflates the price and kills rotation. The deliverable is a sheet with four columns per dish — input cost, selling price, food cost percentage and contribution margin in currency — refreshed the day the month closes. You verify it by comparing the theoretical food cost on that sheet against the real food cost your inventory returns. Diego F. Parra, founder of Masterestaurant, holds that a gap wider than three points between the two is never a pricing problem but a portion, waste or theft problem, and chasing it with price increases only widens the hole.

Step 3 · Classify the menu with menu engineering: margin against rotation, four quadrants

Menu engineering crosses two axes and only two: contribution margin in currency per dish against units sold over a ninety-day window. That crossing yields four quadrants, and each one calls for a different move. High margin, high rotation: protect it, place it in the hot zone of the menu, leave the recipe alone. High margin, low rotation: rescue it through name, description and placement, since labelling a dish 'Most popular' or 'Chef's favourite' lifts orders by 13% to 20% according to NeatMenu. Low margin, high rotation: redesign the recipe or raise the price in small increments. Low margin, low rotation: it goes. The deliverable is the full matrix with all four quadrants populated and one written decision per line. You verify it by repeating the count ninety days later: if no dish moved quadrant, the decisions never reached the kitchen. Physical redesign is the last criterion and it depends on the previous three, because moving a dish around without knowing its margin is expensive decoration.

Step 4 · Redesign the physical menu and mirror the decision in the QR, no exceptions

Four concrete decisions apply here: star-quadrant dishes go top right on the printed menu and inside the first two screens of the QR; prices are written without currency symbols or decorative decimals; every rescuable dish carries a twelve-to-twenty-word description naming the input that justifies its price; and the menu shrinks to a workable number of lines per section. The point almost nobody honours is the mirror: the printed menu and the digital one must be the SAME menu, same order, same descriptions. A QR that still serves the old menu while the dining room runs the new one turns four weeks of work into an inconsistency the guest reads as sloppiness. Raising every price on the menu by the same percentage is the shortcut that destroys the most cash, because demand elasticity is not uniform within a single menu. Follow the chain to its end.

What happens if you raise every price instead of applying these four criteria?

You lift everything 10%: the high-rotation, low-margin dishes that carry lunch traffic lose somewhere between 8% and 15% of units sold;

the average check rises on paper while volume falls, and since the venue's fixed costs did not move a cent, break-even moves further away rather than closer. Meanwhile the high-margin, low-rotation dishes, the ones that could absorb the increase without losing a single guest, kept selling exactly the same. Large US chains raised menu prices 42% between 2020 and 2025, nearly double the 22% of general inflation, according to One Haus, and a good share of that margin was surrendered in lost visit frequency. The most repeated mistake is not arithmetic: it is sequence. Costing without a spec sheet tops the list, and three others cost just as much. Copying the menu of the competitor down the street who looks like they are doing well, with no knowledge of their cost structure or their purchase volume.

The four mistakes that show up every time this guide gets executed

Confusing the best-selling dish with the most profitable one, when margin in currency usually sits on the second or third dish of the sales ranking. And updating the costing once a year, with inputs that move quarter by quarter. For years I argued that the central problem of a menu was price, until costing data showed something else: price is usually set reasonably well, and what is broken is the portion, the waste and the absence of standardisation in the kitchen. That fourth mistake dies with one rule: cost monthly, not annually. A restaurant with 14 employees that pulls its food cost from 38% down to 31% frees seven points of sales turned into operating cash, and that is enough to sustain formal payroll through the weak quarter. SATE Institute works this file with Masterestaurant S.A.S. as technology partner under the Twin Ecosystem Model: the institute sets the agenda, measures impact and runs the strengthening programs; Masterestaurant contributes the platform that turns a written recipe into an auditable cost per portion.

The MIPYME program case: from 38% to 31% food cost and what it means on payroll

For a program officer at the IDB Group or the World Bank the translation is direct: sustained formal employment is SDG 8 measured on payroll, not in speeches, and the waste avoided is target 12.3, the same one the IDB's #SinDesperdicio initiative pursues. It matters because ECLAC documents that mortality among Latin American microenterprises exceeds 50% before year four, and in food service the menu is where that risk materialises at every lunch service. The menu is properly built when these six boxes are ticked and not one minute earlier. One: a signed spec sheet exists for 100% of the menu lines, with no exceptions of the 'the chef does that one from memory' variety. Two: no dish exceeds 32% food cost, and those that did were either removed or redesigned. Three: the gap between theoretical food cost and real inventory sits under three points. Four: the menu engineering matrix has all four quadrants populated with a written decision per dish.

Closing checklist: how to know the menu came out right

Five: the printed menu and the QR display the same order, the same names and the same prices, verified by scanning the code from a table in the dining room. Six: the next costing date is already on the calendar, thirty days out. If any box fails, the menu is not profitable yet; it is a pretty menu waiting for month fourteen. The first difference is sequence, and almost nobody respects it: the standard recipe comes BEFORE costing, and costing comes before any decision on price or design. A restaurant costing without documented grammage is calculating on an imaginary portion; the typical error that broken chain produces runs 12% to 18% on cost per portion, according to the strengthening exercises SATE Institute runs with gastronomic cooperatives. Diego F. Parra, founder of Masterestaurant, insists the spec sheet is not kitchen bureaucracy but the only document that makes a dish cost the same on Tuesday as on Saturday, with the head cook or with the stand-in.

Four decisions that separate a menu that earns from one that bleeds

Second, demand elasticity is not uniform within a single menu, and treating it as if it were costs money. Anchor dishes, the ones diners use to judge whether a place is expensive, tolerate increases poorly; drinks, sides and desserts absorb adjustments of 8% to 15% with no measurable drop in units sold. The practical consequence is that price adjustment becomes surgical, line by line, rather than a flat percentage across the whole card, which is the fastest way to destroy average ticket while believing you are defending it. Third: withdrawing a low-margin, low-turnover dish frees more cash than selling more of the star. It sounds counterintuitive and it is plain arithmetic. Every active reference consumes cold storage, tied-up inventory, training time and mise en place complexity; when that reference represents under 1.5% of units sold, its carrying cost exceeds any margin it contributes. Here sits the real tension of the trade: the chef defends variety as identity, the finance side attacks it as overhead.

Four decisions that separate a menu that earns from one that bleeds — in practice

The bridge is seasonal rotation, which preserves creativity in a 4 to 6 dish off-menu format without loading the permanent operation. The fourth difference touches format, and at Masterestaurant it admits no nuance: the PHYSICAL menu and the QR menu coexist, each with its job. The printed card controls the experience (service pacing, menu narrative, suggestive selling, hospitality); the QR complements it (delivery, accessibility, price updates within 48 hours, consultation analytics). Restaurants that scrapped paper between 2021 and 2023 reported average ticket drops in the order of 6% to 9%, because they lost the instrument the server sells with. Whoever removes the physical card is not digitising: they amputate the suggestive-selling tool and then blame the market for the drop.