How to design a menu that increases profits: the star dish myth against cash-register arithmetic
A menu increases profits when you rebuild it around CONTRIBUTION MARGIN IN DOLLARS, not around food cost percentage and not around the chef's preference. The sequence that works is short and unglamorous: a standard recipe per dish, per-portion costing with measured yield loss, popularity crossed against unit margin, removal of the bottom 15-20% that fails to cover its own opportunity cost, and relocation of the four highest-margin dishes into the zones where the eye actually stops. The myth says the star dish is the best seller; register arithmetic says the star dish is the one leaving the most dollars per unit times its rotation, and those two rarely coincide. With per-dish food cost capped at 32% and yield loss measured in real service, the redesign moves gross margin three to seven points in sixty days without touching a single menu price.
There is probably one dish costing you money every time it leaves the kitchen, and it is not the cheapest item on your menu: it is the one you defend because it is your signature. Open its standard recipe, add the trim loss on the protein, the portion your cook actually plates when the pass is slammed, and the waste on the side dish, and a theoretical 29% food cost lands in the books at 38%. At eighty covers a month, that gap runs into four figures a year. That is where this guide starts, and not with the typography of your printed menu.
The industry carries a sizing problem and a method problem at the same time. Menus grew by accretion — the dish a regular requested, the special that never left, the dessert a supplier pushed — until they became inventories of sixty references that force forty purchased SKUs, poor storeroom rotation, and three weeks of training before a new cook can produce unsupervised. Profitability is not lost at the table; it is lost in purchasing and in the operation that menu imposes. Margin recovered by cutting dead references usually beats margin gained by raising prices.
Diego F. Parra frames it at Masterestaurant with a sentence chefs dislike: the menu is a financial document before it is a culinary one. It determines what you buy, how much capital sits in the walk-in, how many hands each service needs, what average check you can sustain, and what remains after the plate is paid for. Designing a menu that increases profits means deciding, with numbers, what stays and what goes — and that decision runs on two columns: margin dollars per dish and units sold per period.
Side-by-side comparison
| Redesign by food cost percentage (myth) | Redesign by contribution margin (reality) | |
|---|---|---|
| Metric that ranks the menu | ✕Food cost % per dish; the 22% item wins | ✓Margin dollars per dish; the item leaving $180 wins even at 34% |
| Gross margin effect at 60 days | ✕Zero to 2 points, often negative once average check drops | ✓Three to 7 points with the same guests and no price increase |
| Resulting menu size | ✕Stays at 45-60 references because nobody cuts anything | ✓Drops to 24-32 references; the bottom 15-20% leaves |
| Unique SKUs purchased | ✕38-45 SKUs with uneven rotation and 6-9% storeroom waste | ✓22-28 SKUs cross-used between dishes, waste under 4% |
| Training time for a new cook | ✕18-21 days before unsupervised production | ✓7-9 days; the standard recipe is the manual |
| Risk when moving prices | ✕High: an 8% across-the-board hike costs 11% of units on elastic dishes | ✓Low: 4-6% only on low-elasticity items, retaining 97% of units |
| What remains documented | ✕A freshly printed menu and no traceability | ✓Auditable standard recipes, per-portion costing and a menu engineering matrix |
Step 1: write a standard recipe for every dish, with weights and yield
The deliverable of step one is a card per dish with exact weights, measured trim yield and a plating photo; without it, nothing that follows is comparable. Weigh the whole cut, clean it, weigh it again and write down the percentage lost to fat, bone or trim, because that number—not the invoice price—sets your portion cost. A tenderloin bought at 210 pesos a kilo with 22% trim loss actually costs 269 pesos per served kilo. Verify it this way: ask the cook to plate three units of the same dish without looking at the card, weigh all three, and if the spread between them passes 8% you don't have a recipe yet, you have a habit. Cash leaks out in those extra 60 grams on a rushed Saturday. Contribution margin in pesos is selling price minus portion cost, and that subtraction outranks any percentage. A 90-peso dish at 25% food cost leaves 67 pesos; a 320-peso dish at 34% leaves 211 pesos, three times the cash for the same menu real estate and nearly the same kitchen time.
Step 2: cost by portion and calculate contribution margin IN PESOS
Charge only ingredients to the plate: payroll and rent belong to the break-even calculation, following the costing doctrine Diego F. Parra applies at Masterestaurant, where 32% works as a CEILING per dish and never as a target to chase downward. Labor cost gets managed separately, inside its 25% to 35% of revenue band (U.S. Bureau of Labor Statistics). Deliverable: one sheet with four columns—price, portion cost, margin in pesos, food cost percentage—and every dish sorted by the third column, not the fourth. You need two axes and ninety days of per-dish sales to classify the menu: units sold on the horizontal axis, margin in pesos on the vertical one, and the median of each as the cut line. Dishes landing upper right get protected and placed where the eye lands first. Popular low-margin items get their recipe reworked or their price raised with a visible change on the plate.
Step 3: cross popularity against margin and sort the menu into four boxes
High-margin slow movers are rescued through name, description and position, the cheapest lever available. And the bottom-left box gets cut without mourning. The figure that settles the argument: killing a dead reference frees ingredients, cold storage and training time, and that saving usually weighs more on the result than raising every price by 4%. Deliverable: the printed matrix with each dish in its quadrant and the decision written beside it. A sixty-item menu that forces you to buy forty different ingredients isn't variety, it's frozen inventory plus three weeks of training per new cook. Cut until each ingredient works across two or three dishes, because that's where the margin nobody books shows up: fewer purchases, less spoilage from slow rotation, fewer errors at peak service. Match the format to what guests actually want now: 37% look for quick bites instead of large meals (Circana, 2024), 51% replace meals with snacks (Technomic, 2023) and 53% had dessert in the past day (Technomic), so a menu with shareable starters and one strong dessert bills better than one carrying four redundant mains.
Step 4: cut references and size the menu to what your kitchen can sustain
Deliverable: the list of dropped references, the list of ingredients you stop buying, and the new storeroom SKU count. Price gets set from your target margin in pesos and from what the guest accepts paying in that category, never by multiplying cost by a fixed factor. Raise first where guests compare least: sides, seasonal starters, desserts, drinks. Behind the bar the ranges are known—pour cost runs near 25% on bottled beer and 20% on draft (Toast, 2024), while wine moves between 35% and 45% (BackBar)—so a badly calibrated wine list eats the cash your kitchen earned. Steer clear of aggressive dynamic pricing: 52% of consumers read it as gouging and 36% would order less often (Capterra, 2024). And whenever you raise a price, change something visible on the plate that same day. Deliverable: the new menu showing old price, new price and expected margin per dish on a single sheet.
Step 6: write and lay out the menu so the eye reaches the profitable dishes
Layout is the last lever, and it works only when the previous five are done. Put the profitable-quadrant dishes in the first third of each block and in the upper right corner of the page, drop the currency symbol and set prices right after the description, never in a column that invites reading top to bottom. Descriptions sell when they name origin, technique and texture in under twenty words. Some demand signals are worth stating in writing: 61% of consumers look for natural items on the menu (Nation's Restaurant News, 2024), 47% ate globally influenced food in the past week (Datassential, 2025) and 25% avoid major allergens (FARE), which makes clear allergen marking mandatory. Deliverable: a full-size printed proof, read by someone outside the business in under ninety seconds. The costliest mistake is costing from the invoice instead of from real trim loss: a theoretical 29% food cost shows up in the books as 38% and, if the dish sells eighty times a month, the annual hole reaches four figures.
The four mistakes that ruin this work, and how to avoid them
The second is chasing the percentage down to 24% by shrinking portions or downgrading ingredients, and the guest catches that before the accountant does. The third is pulling the dish that carries the identity of the place just because its margin is middling: rework the recipe there, don't kill it. The fourth is doing all of this once and filing it away, when purchase prices move every quarter. Working rule: recost your ten highest-volume dishes every ninety days and the full menu twice a year, with a written record of what changed and why. You'll know the job landed when you can answer five questions with a document rather than an opinion. One: does a card with weight, trim loss and photo exist for 100% of active dishes? Two: is the peso margin of every dish written down, with no active item above 32% food cost? Three: does each dish have an assigned quadrant and a decision taken?
Closing checklist: how to know the new menu came out right
Four: did the storeroom ingredient count drop against last month? Five: do theoretical and actual food cost differ by less than two points at the close of the following month? If that gap stays above four points, the problem isn't the menu but portioning or purchasing, and that's where you should look. Set a date for the first measurement: thirty days after printing, with per-dish sales in hand. Percentage food cost compares dishes at different price points, which is exactly why it misleads: a $90 plate at 25% leaves $67, and a $320 plate at 34% leaves $211. Favor the first because it costs less and you have optimized a ratio while neglecting cash. The Masterestaurant rule keeps 32% as a CEILING per dish, never as a target to chase downward, because every point below 26% tends to be paid for with a smaller portion or a worse ingredient, and guests notice that before your accountant does.
Seven differences that move the register
A standard recipe is not paperwork; it is the only instrument that makes two services comparable. With no written gram weight, no measured trim yield and no plating photo, the same dish leaves at 180 grams on Tuesday and 240 on Saturday, and those 60 grams on a protein at $420 per kilo are $25 per unit nobody books. Multiply by 300 monthly units and you have $7,500 a month evaporating as unbudgeted generosity. Demand elasticity is not uniform across one menu, and that detail decides the increase. Dishes with public price anchors — burgers, margherita pizza, caesar salad — tolerate almost nothing above 4%, while a signature plate with no comparable nearby absorbs 8% to 12% with no measurable unit loss. Raising everything equally is the fastest way to lose the items that could have carried an increase and to punish the ones that could not. Opportunity cost on the pass outweighs food cost in a saturated kitchen.
Seven differences that move the register — in practice
A dish taking 14 minutes and leaving $150 of margin yields less per cook-hour than one taking 5 minutes and leaving $95, and at Saturday peak that gap sets your table turns. That is why the menu engineering matrix in this guide crosses three axes instead of two: margin, popularity and pass minutes. Dishes that hurt profitability rarely show up in the sales report, precisely because they sell so little. The damage sits in the inventory they force you to hold: two exclusive refrigerated ingredients with a seven-day shelf life, purchased to serve eight orders a month. The real loss there is not the plate margin, it is storeroom waste and locked-up capital, which in mid-sized kitchens runs 6% to 9% of purchase value. Average check moves through architecture, not through floor pressure. When the menu carries three shareable starters placed well, two high-margin sides and a quick-execution dessert, the check rises 8% to 14% with no server suggesting anything, because the reading order already induced the decision.
Seven differences that move the register — key points
Verbal upselling works, yet it is fragile: it depends on the server remembering, and front-of-house turnover in this industry runs above 70% a year. A 26-reference menu is not a poor menu; it is a menu your kitchen can execute identically every day. I was wrong about this for years, defending variety as a selling argument, until measurement across several operations showed 80% of revenue coming from 12 to 16 dishes while the rest added only purchasing complexity. Cutting does not shrink revenue; it frees purchasing, walk-in space and cook hours so the dishes that do sell come out better.
Myth against reality, criterion by criterion
The myth: profitability means cutting food costWhat the industry repeats
- Ranks dishes by cost percentage and attacks the most expensive first, ignoring how many units it moves
- Raises prices across the board and discovers next month that unit losses outran the increase
- Confuses the best seller with the most profitable dish, then protects it on the menu for years
- Costs from invoice price, with no trim loss, no service waste and no measurement of the portion leaving the pass
- Treats the redesign as a graphic project — type, photography, placement — with the arithmetic still unresolved underneath
The reality: you rebuild around margin dollars and rotationMasterestaurant
- Ranks by unit contribution margin times units sold over 90 days, which is what actually pays payroll
- Adjusts price only where demand elasticity allows it: signature plates with no external price reference nearby
- Removes without nostalgia the bottom 15-20% that fails to cover its opportunity cost on the pass
- Costs per portion using yield measured across three real services, never the yield a supplier promises
- Treats restaurant menu design as the closing step: matrix first, layout second
Side-by-side comparison
| Redesign by food cost percentage (myth) | Redesign by contribution margin (reality) | |
|---|---|---|
| Metric that ranks the menu | ✕Food cost % per dish; the 22% item wins | ✓Margin dollars per dish; the item leaving $180 wins even at 34% |
| Gross margin effect at 60 days | ✕Zero to 2 points, often negative once average check drops | ✓Three to 7 points with the same guests and no price increase |
| Resulting menu size | ✕Stays at 45-60 references because nobody cuts anything | ✓Drops to 24-32 references; the bottom 15-20% leaves |
| Unique SKUs purchased | ✕38-45 SKUs with uneven rotation and 6-9% storeroom waste | ✓22-28 SKUs cross-used between dishes, waste under 4% |
| Training time for a new cook | ✕18-21 days before unsupervised production | ✓7-9 days; the standard recipe is the manual |
| Risk when moving prices | ✕High: an 8% across-the-board hike costs 11% of units on elastic dishes | ✓Low: 4-6% only on low-elasticity items, retaining 97% of units |
| What remains documented | ✕A freshly printed menu and no traceability | ✓Auditable standard recipes, per-portion costing and a menu engineering matrix |
The numbers that decide
“We ran 54 dishes and believed variety protected us. Once we costed per portion with real trim loss, 11 dishes sat above 32% food cost and together they made up 6% of sales. We cut those 11, left the menu at 29, raised prices 6% on the four signature plates only, and moved the sides into the top block. In 74 days gross margin went from 61.4% to 67.1%, average check rose from $412 to $468, and purchased SKUs fell from 41 to 26. We did not lose guests: we lost useless work.”
Six steps, each with a deliverable and a numeric checkpoint
Before touching the menu you need three files: the POS export of units sold per dish over the last 90 days, last month's purchase invoices with price per unit of measure, and a physical count of walk-in and dry storage. Without those three, any costing is an estimate wearing the face of a fact. COMMON ERROR: using the supplier's list price instead of the price actually paid net of discounts and freight, a gap that averages 4% to 9%. CHECKPOINT: you hold one table with 100% of sold references and their current purchase price, and monthly purchases reconcile to the books within 2%.
Record exact gram weight per component, the yield after trimming — weighing the ingredient before and after, on three separate days — and one plating photo that serves as the pattern. Yield loss gets measured in operation; it is never copied from a table. COMMON ERROR: accepting the yield a supplier promises; on beef cuts the gap between a promised 82% and a real 71% in a rushed kitchen is worth $18 to $40 per portion. CHECKPOINT: 100% of active dishes carry a card with weight, yield and photo, and three random plated samples deviate no more than 8% from the written weight.
For every dish: total ingredient cost including yield loss, divided by portions produced, subtracted from the pre-tax selling price. What remains is the unit contribution margin, the figure that decides everything downstream. Payroll, rent and utilities do NOT load onto the plate; they belong to the monthly break-even. COMMON ERROR: allocating cook wages per dish and concluding that everything loses money. CHECKPOINT: every dish carries three numbers beside it — cost per portion, food cost percentage and margin in dollars — and none exceeds the 32% ceiling without a signed decision note.
Cross unit margin against units sold, then add the axis almost nobody uses: execution minutes on the pass. Four groups emerge — high margin and high volume, high margin and low volume, high volume and thin margin, and the ones doing neither — and inside each group you will see which items jam the kitchen. COMMON ERROR: averaging a period that covers an atypical season. CHECKPOINT: every dish sits in a quadrant with its margin figure and its pass time clocked across at least five real tickets.
Out go dishes with thin margin, low rotation and an exclusive ingredient, in that order of priority. Then raise prices 4% to 6% only on plates without an external price anchor, leaving untouched anything guests compare elsewhere. Diego F. Parra insists at Masterestaurant that increases happen dish by dish and with criteria, never across the board. COMMON ERROR: cutting the plate a regular always orders and communicating it badly. CHECKPOINT: the menu drops to 24-32 references, unique SKUs fall below 30, and after 30 days total units have not dropped more than 3%.
Now, and only now, comes restaurant menu design: short blocks of five to seven dishes, the highest-margin items at the top right of each block, no aligned price column inviting a vertical scan, and high-margin sides visible beside the mains. Measure at 60 days. COMMON ERROR: changing the menu, the prices and the supplier on the same day, then being unable to attribute the outcome. CHECKPOINT: gross margin rises three points or more against baseline, average check rises 6% or more, and consolidated monthly food cost lands under 32%.
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
What holds the redesign in place
A redesign collapses when the old menu creeps back for lack of control. Three pieces hold the result: an explicit business model so you know which guest that menu serves, a growth dashboard measuring what happened after the change, and cash control confirming the margin you gained reached the bank instead of parking in the storeroom.
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.
Questions that arrive every week
How many dishes should a profitable restaurant menu have?
How many dishes should a profitable restaurant menu have?
Between 24 and 32 references in full service, and 10 to 16 in fast formats. The criterion is not aesthetic: 80% of revenue should concentrate in 12 to 16 dishes, and the kitchen must execute them identically at Saturday peak. Above 40 references, storeroom waste climbs to the 6-9% range.
Is raising prices the fastest way to increase profits?
Is raising prices the fastest way to increase profits?
It is the fastest and the most expensive when done across the board. An 8% increase on the whole menu can cost you 11% of units on dishes with public price anchors. Raise 4-6% only where demand elasticity allows — signature plates without a close comparable — and recover the rest by cutting references that fail to cover their cost.
How do I decide which dishes to cut from my menu?
How do I decide which dishes to cut from my menu?
Rank by contribution margin in dollars times 90-day units, then flag those also requiring an exclusive ingredient. That cross identifies the 15-20% that leaves. A thin-margin dish sharing ingredients with the rest of the menu can stay; one forcing two refrigerated purchases for eight monthly orders cannot.
Is 32% food cost a target or a limit?
Is 32% food cost a target or a limit?
It is a CEILING per dish, never a goal to chase downward. Below 26% you usually find a shrunken portion or a cheaper ingredient, and guests detect that before accounting does. Payroll, rent and utilities never load onto the plate: they belong to the monthly break-even, where real profitability is decided.
Sector data 2026 (official sources)
Verifiable industry benchmarks from official, non-commercial sources (government, industry associations, market research) - not competitors.
| Metric | Benchmark 2026 | Source |
|---|---|---|
| Comensales que eligieron platos con etiquetas descriptivas en el estudio | 56% | Cornell University Food & Brand Lab (Wansink) — Descriptive Menu Labels' Effect on Sales |
| Ventas proyectadas de la industria de restaurantes y foodservice (EE. UU.) | USD 1,5 billones en 2025 | National Restaurant Association — State of the Restaurant Industry 2025 |
| Empleo total de la industria de restaurantes (EE. UU.) | 15,9 millones de personas en 2025 | National Restaurant Association — 2025 Forecast |
| Nuevos empleos que suma la industria de restaurantes (EE. UU.) | +200.000 empleos en 2025 | National Restaurant Association — 2025 Forecast |
| Locales de restaurantes y foodservice (EE. UU.) | Más de 1 millón de locales | National Restaurant Association — 2025 Forecast |
| Utilidad antes de impuestos en servicio completo (mediana) | 2,8% de las ventas en 2024 | National Restaurant Association — Restaurant Operations Report 2024/25 |
Related content
Grow your restaurant with the Masterestaurant method
Applied in +8.400 restaurants across 43 countries.