Profitable menu: criteria to build it — the traditional method, its limits and four honest alternatives
Verdict: when you build a profitable menu the deciding criterion is NOT food cost percentage dish by dish, it is contribution margin in money multiplied by how often each reference sells; the traditional method (portion costing plus the one-third rule) holds up to roughly 35 references and until your average check flattens, and past that point it falls short because it ignores demand elasticity and what position on the panel does to sales.
For 2026 the order we recommend at Masterestaurant runs like this: standard recipes for the whole menu first, then the menu engineering matrix of margin against rotation, then pricing and design decisions, and technology only at the end. Buy software before writing standard recipes and you have paid for a dashboard that displays expensive garbage.
A 92-seat restaurant in Bogotá billed 41 % more than the previous year and closed December with less cash in the bank. The menu carried 58 dishes, none recosted since protein prices moved, and the three best sellers left 9,200 pesos of margin while the fourth one, the one nobody ordered, left 31,400. That is what a menu that grows without criteria looks like.
Building a profitable menu is not a percentage exercise. It is a portfolio decision: you have a limited amount of guest attention, a kitchen with physical bottlenecks, and demand that responds differently to every price. The standard recipe gives you cost; menu engineering tells you what to do with it.
Below I compare the traditional method against four alternatives that people actually use in this trade — a spreadsheet with the Kasavana-Smith matrix, recipe-costing software, project-based menu consulting and the Masterestaurant framework — with real cost, learning curve and the operation profile each one serves.
Side-by-side comparison
| Traditional method (costing + one-third rule) | Masterestaurant framework (margin × rotation) | |
|---|---|---|
| Decision criterion on a dish | ✕Theoretical food cost ≤ 30 %; above that, raise price or drop the dish | ✓Contribution margin in money × units sold per month; food cost ≤ 32 % is a ceiling, not a target |
| Implementation cost (single-unit restaurant) | ✕0 to 400 USD, usually done in Excel with no outside help | ✓600 to 1,800 USD covering kitchen hours to weigh recipes plus 3 analysis sessions |
| Time to the first useful decision | ✕2 to 3 weeks costing dish by dish before anything shows | ✓8 days: the 12 references worth 70 % of sales get costed first |
| References the menu holds without breaking | ✕Up to 35 dishes; beyond that costing maintenance gets abandoned | ✓60 to 90 dishes with quarterly recosting because work is ranked by sales weight |
| Measured effect on average check | ✕Rises through price, not mix: 3 to 5 % and then it stalls | ✓7 to 12 % across two quarters by reordering mix, position and price anchors |
| What it does with demand elasticity | ✕Ignores it, assuming an 8 % price rise means 8 % more revenue | ✓Measures it per reference against sales history before any price moves |
| Printed menu versus QR | ✕Everything goes digital to save printing and the printed menu disappears | ✓PRINTED menu governs service pace and suggestive selling, QR complements delivery, accessibility and price updates |
| Recosting when an input jumps 15 % | ✕Reactive, whenever the accountant flags it 60 days later | ✓Threshold alerts on the 15 critical inputs, review within 72 hours |
The criterion that decides a profitable menu is not the percentage
A profitable menu is built on contribution margin in currency multiplied by turnover, and food cost percentage stays as a secondary control, never as the deciding criterion. The Bogotá case says it plainly: 92 seats, revenue up 41 %, 58 dishes on the menu, and less cash in the bank in December than the year before, because the three best sellers left 9,200 pesos per unit while the fourth dish, the one nobody ordered, left 31,400. A ceviche running 34 % food cost that sells 240 times a month contributes more money than a pasta at 22 % that sells 40. The arithmetic is elementary and still the industry keeps costing by portion and applying the one-third rule, a method that works while inputs barely move. It stopped working: US retail beef hit USD 5.98 per pound in May 2025, an all-time high according to the US Bureau of Labor Statistics via CBS News.
When the original option falls short?
Portion costing with the one-third rule runs out when an input's volatility exceeds the margin of error the formula tolerates, and the giveaway is easy to check:
if any ingredient in your three star dishes rose more than 15 % since the last recipe costing, your menu is already pricing against costs that no longer exist. Public numbers confirm it. A dozen Grade A eggs went from USD 2.04 in August 2023 to USD 4.95 in January 2025 per the BLS, and ground beef marked USD 6.12 per pound in June 2025 (BLS via NPR). No fixed percentage survives a 142 % jump in a base input. The second symptom belongs to the portfolio: going from thirty to fifty-eight references without retiring a single one. The one-third rule has no way to tell you what to cut, because it measures dishes one at a time and a menu is a set.
Spreadsheet with the Kasavana-Smith matrix: for the owner who already tracks sales per dish
The Kasavana-Smith matrix crosses popularity against contribution margin and sorts every reference into four quadrants, with an entry cost close to zero if you already export per-dish sales from your point of sale. It fits the chef-owner of a single location, somewhere between 20 and 60 references, with the discipline to update recipe costings every quarter. Real effort runs 6 to 10 hours the first time and roughly 2 hours per later review, nearly all of it spent squaring standard recipes rather than doing the math. What you gain: a map showing which dish sells heavily and leaves little, the number one candidate for recipe or price rework. What you lose: the sheet captures neither cannibalization between similar dishes nor the griddle bottleneck at the Friday peak, and that is precisely where decisions get made that the sheet cannot see. Recipe costing software wired to inventory recalculates dish cost every time a purchase price changes, and that is its one decisive advantage over the spreadsheet, though it is enough to justify the tool in a multi-unit.
Recipe costing software: for operations with live inventory and several locations
The profile is clear: two locations or more, centralized purchasing, someone responsible for inventory five days a week. With fresh fish at USD 9.18 per pound in 2024 according to the USDA Economic Research Service, and beef at USD 6.51 against chicken at USD 2.99, a surf-and-turf menu can drift several points of theoretical cost within a month before anyone notices at closing. The switching cost does not sit in the license. It sits in loading recipes and sustaining counts: 40 to 80 hours of implementation, and if counting is abandoned the system hands you false costs with two decimals, which is worse than having nothing. Hiring project-based menu consulting makes sense when the pending decision is structural —closing lines, redesigning the whole menu, repositioning prices against a new competitor— and not when what is missing is costing routine. Diego F. Parra insists on an order the industry usually reverses: first you decide what restaurant you want to be, then you pick the tool that measures.
Project-based menu consulting: when the menu must be rebuilt, not tuned
A typical project runs four to eight weeks, covers sales analysis, price testing and physical menu redesign, and its biggest benefit is rarely the food cost saving: it is retiring dead references and reallocating the diner's attention. One figure no menu should ignore: the first entrée listed in its category carries a 33 % chance of being ordered, regardless of price, according to NeatMenu. That ordering is decided once and pays every single day. The Masterestaurant framework sets price by demand elasticity and then validates it against contribution margin, reversing the sequence of the traditional method, which prices by formula and hopes the market accepts. The difference gets paid in units: raising a dish 10 % when demand is elastic can cost you 18 % of sales and leave you worse off than before, even as the food cost percentage improves on the sheet. According to Sheryl E. Kimes, professor emerita at Cornell's School of Hotel Administration, revenue management in restaurants is played in the combination of price and duration of table use, not in price alone.
The Masterestaurant framework: price by elasticity, validated against margin
Some categories hide margin out of habit: average pour cost on alcoholic beverages sits near 20 %, with liquor around 15 % and wine between 35 % and 45 % (BackBar). And 74 % of operators say global flavors let them charge more (Datassential). Suppose you apply a blanket 6 % increase across the menu and change nothing else. Inelastic dishes —the house specialty, the plate people come looking for— absorb the increase with no unit loss and lift your margin. The elastic ones, typically those competing with three restaurants on the same block, lose between 8 % and 15 % of covers, and since they tend to be the high-turnover items, total cash falls even as average check rises. A month later you see average check climbing and sales flat, the most confusing photograph in this trade. Year-over-year menu prices at full-service restaurants peaked at 9.0 % in 2022 according to the National Restaurant Association with BLS data, and plenty of operators copied that flat number across every line.
What happens if you raise prices without touching menu order?
The alternative is boring and it works: raise hard where nobody compares, freeze where they do, retire what does not turn.
Stay with portion costing and the one-third rule if your menu holds fewer than twenty references, one location, stable inputs, and you already know by heart what each dish leaves. At that scale menu engineering buys precision you will never use, and those 10 setup hours pay better in purchasing or on the shift. Do not switch during high season either: redoing recipe costings and reordering the menu in December pulls the chef's head away exactly when the kitchen tolerates no distraction. And if your real problem is waste rather than price, no matrix will solve it. The Bogotá case was not fixed with software: it was fixed by retiring 21 dishes and moving the one with 31,400 pesos of margin to the first line of its category.
When NOT to switch methods?
Track per-dish sales in your point of sale for one month. If the top 20 % of references fails to explain at least 60 % of cash, there is your work.
The traditional method asks what the dish costs; menu engineering asks what the dish leaves and how many times it goes out. A ceviche at 34 % food cost selling 240 times a month leaves more money than a pasta at 22 % selling 40. According to Sheryl E. Kimes, professor emerita at the Cornell School of Hotel Administration and a reference in restaurant revenue management, revenue in restaurants is won on the combination of price and how long the table is occupied, never on price in isolation, and that observation transfers straight to the menu. Traditional pricing runs on a formula; the Masterestaurant framework prices against demand elasticity and then validates with margin. Raise an elastic dish 10 % and you can lose 18 % of its units, which leaves you worse off than before.
Where the criteria really diverge?
Menu design is a printing matter for the traditional approach; we treat it as part of the cost system, because moving a reference from the center of the panel to the edge changes its units sold before you touch the price.
The traditional method recosts when it hurts; the framework recosts by threshold on the critical inputs, which in an average menu number 15 rather than 300. And on printed menu versus QR, Masterestaurant recommends BOTH with separate roles: the printed menu governs service pace, narrative and suggestive selling, while the QR handles delivery, accessibility, price changes and analytics. Dropping print to save 380 USD a year costs average-check points worth far more than that.
Alternative by alternative: cost, curve and verdict
Traditional method: portion costing and the one-third ruleThe original option
- Each dish gets a standard recipe and the cost is divided by 0.30 to set the price.
- It is cheap, it can be learned in an afternoon and it depends on no software vendor.
- Real LIMIT: it treats every dish as equal and mistakes a low percentage for money earned.
- It breaks past 35 references, or when the same cook weighing recipes also has to cover service.
- It also breaks with seasonal menus, since nobody sustains recosting 58 dishes every six weeks.
Real alternatives once the traditional method falls shortMasterestaurant
- Spreadsheet with the Kasavana-Smith matrix: near-zero cost, 6 to 10 hours of learning, for the owner who already exports per-dish sales from the POS.
- Recipe-costing software (Apicbase, MarketMan, Fudo and similar): 40 to 250 USD a month, 3 to 5 weeks of learning, for two or more units with formal inventory.
- Project-based menu consulting: 1,500 to 6,000 USD, results in 4 to 8 weeks, for relaunches and openings.
- Masterestaurant framework: mid-range cost, 8-day curve, for the operator who wants the criteria to stay in the house.
- None of the four works without standard recipes first, and that is the toll nobody skips.
Side-by-side comparison
| Traditional method (costing + one-third rule) | Masterestaurant framework (margin × rotation) | |
|---|---|---|
| Decision criterion on a dish | ✕Theoretical food cost ≤ 30 %; above that, raise price or drop the dish | ✓Contribution margin in money × units sold per month; food cost ≤ 32 % is a ceiling, not a target |
| Implementation cost (single-unit restaurant) | ✕0 to 400 USD, usually done in Excel with no outside help | ✓600 to 1,800 USD covering kitchen hours to weigh recipes plus 3 analysis sessions |
| Time to the first useful decision | ✕2 to 3 weeks costing dish by dish before anything shows | ✓8 days: the 12 references worth 70 % of sales get costed first |
| References the menu holds without breaking | ✕Up to 35 dishes; beyond that costing maintenance gets abandoned | ✓60 to 90 dishes with quarterly recosting because work is ranked by sales weight |
| Measured effect on average check | ✕Rises through price, not mix: 3 to 5 % and then it stalls | ✓7 to 12 % across two quarters by reordering mix, position and price anchors |
| What it does with demand elasticity | ✕Ignores it, assuming an 8 % price rise means 8 % more revenue | ✓Measures it per reference against sales history before any price moves |
| Printed menu versus QR | ✕Everything goes digital to save printing and the printed menu disappears | ✓PRINTED menu governs service pace and suggestive selling, QR complements delivery, accessibility and price updates |
| Recosting when an input jumps 15 % | ✕Reactive, whenever the accountant flags it 60 days later | ✓Threshold alerts on the 15 critical inputs, review within 72 hours |
The figures behind the decision
“We had 58 dishes and we were convinced payroll was the problem. We weighed the 12 references worth 71 % of sales and found the tenderloin running at 38 % food cost since we changed supplier in March, eight months earlier. We cut 19 dishes, moved four to the center panel of the printed menu and kept the QR for delivery and price changes. Average check went from 71,400 to 79,100 pesos in eleven weeks and global food cost dropped from 36.4 % to 30.8 % without touching a single recipe that was already working.”
How to build the profitable menu in four steps
Do not start with all 58. Rank your last 90 days of sales and weigh, on a scale and with real trim loss included, the ones adding up to 70 % of revenue: in an average menu that means 12 to 15. Every card carries gram weight, portion cost, cleaning loss and cut yield. Without this, everything else is opinion formatted as a table.
Cross each reference's contribution margin in money with its units sold and four groups appear: stars, cash cows, puzzles and dogs. Dishes that hurt profitability sit in the low-low quadrant, and dropping them is not always the answer; sometimes redesigning the portion, or swapping an expensive garnish for a 900-peso one the guest values just as much, is enough.
Before raising a price, look at what happened the last time you moved it. If the dish lost a larger percentage of units than it gained in price, it is elastic and stays untouched: you redesign it instead. Set a high anchor in each family, keep food cost under 32 % as a ceiling, and confirm the venue's break-even is still covered by the new mix, since payroll and rent never load onto the plate.
Move the stars to the center panel of the PRINTED menu, which is where you control service pace and suggestive selling, and keep the QR for delivery, accessibility and price updates. Then set alerts on the 15 critical inputs: when one rises 8 %, you recost within 72 hours rather than two months later, once the quarter is already gone.
And with AI?
Optimize menu engineering, descriptions and the photos that sell most. Diego F. Parra is an expert in AI applied to restaurants.
Free tools to apply this now
Ecosystem tools to sustain the menu
Criteria stay in the house when somebody writes them into a board that gets read every Monday. These three pieces of the Masterestaurant ecosystem answer the three questions that surface while building a profitable menu: what business model holds that menu, how far it can grow on the current mix, and whether cash survives the transition.
Diego F. Parra — International consultant, expert in creating and scaling restaurants and in AI applied to restaurants, foodtech and HORECA. Methodology applied in 8.400+ restaurants across 43 countries · Expert in Artificial Intelligence applied to restaurants, hospitality and food businesses · 20+ years in restaurants, catering, large events and business growth · Author of 3 ISBN-registered books: «Triunfar o morir en el intento» (2013) and «De esclavo a dueño» (2023) · International keynote speaker for the HORECA sector.
Frequently asked questions about profitable menu criteria
How many dishes should a profitable menu have?
How many dishes should a profitable menu have?
Between 24 and 40 references for a single-unit full-service restaurant. Above 40 inventory balloons, waste climbs and portion costing stops staying current. Under 20 usually caps average check, except in specialized formats built around one product.
Does food cost have to be 30 % on every dish?
Does food cost have to be 30 % on every dish?
No. The 32 % figure is a per-dish ceiling, not a uniform target. What you optimize is contribution margin in money multiplied by how often the dish goes out; a 34 % dish with high rotation can leave more cash than a 22 % one almost nobody orders.
How often should the menu be recosted?
How often should the menu be recosted?
Quarterly for the whole menu and by threshold on the 15 critical inputs: if one rises 8 % or more, recost within 72 hours. With food-away-from-home inflation at 2.1 % annually per the BLS, waiting six months eats your margin before you notice.
Should the printed menu be dropped in favor of QR only?
Should the printed menu be dropped in favor of QR only?
No. Masterestaurant recommends keeping both with distinct roles: the printed menu controls service pace, menu narrative and suggestive selling, while the QR handles delivery, accessibility, price changes and analytics. Cutting print saves on the printer and costs you check.
Sector data 2026 (official sources)
Verifiable industry benchmarks from official, non-commercial sources (government, industry associations, market research) - not competitors.
| Metric | Benchmark 2026 | Source |
|---|---|---|
| Cruce de ventas: servicio completo supera al limitado | El servicio completo superó al servicio limitado en ventas en 2024 | USDA Economic Research Service |
| Caída de tráfico en casual dining (marzo 2024) | -4,1% en casual dining; -5,7% en fine dining | Technomic / Black Box Intelligence |
| Pour cost promedio de bebidas alcohólicas (bar) | ~20% (licor ~15%, cerveza de barril ~20%, vino 35-45%) | BackBar (guía de la industria) |
| Costo de vertido del vino como % de su venta | 35% a 45% | BackBar / Restaurant365 (guía de la industria) |
| Tamaño óptimo de menú por categoría | 7 a 15 ítems por categoría (para evitar parálisis de decisión) | Investigación de diseño de menú (agregada) |
| QSR que subieron precios en 2024 | 93% de los restaurantes de servicio rápido | Oysterlink (recopilación) |
Related content
Grow your restaurant with the Masterestaurant method
Applied in +8.400 restaurants across 43 countries.