Profitable Menu: Criteria to Build It — Traditional Method vs Masterestaurant Method

Which criterion actually decides what stays on the menu?

What decides is contribution margin in dollars multiplied by weekly turnover, never food cost percentage on its own.

A steak carrying 42% raw material cost that leaves $14 per plate and goes out four minutes after the ticket prints beats a pasta at 24% that leaves $6.20 and ties up the pan for eleven minutes during the Friday peak. Write the arithmetic and the argument ends: forty steaks bring $560; forty plates of pasta, $248. The percentage favors the pasta, the register favors the steak. I got this wrong for years, chasing the percentage as if it were the whole trade, and I built menus that were flawless in the spreadsheet yet made less money than the messy menu of the competitor across the street. Menu engineering applied with this criterion raises average profitability by roughly 10%, according to a Cornell University study on menu engineering.

The hard 2026 trend: recost every four weeks, not every six months

Recosting frequency is the trend defending the most margin this year, and its measurable signal sits in input costs: in the United States food is up 35% since 2019 and labor another 35%, according to the National Restaurant Association. On that slope, a menu priced back in 2023 with a flat three-times multiplier on the recipe arrives in 2026 selling protein dishes below true cost. Operators who moved from reviewing prices twice a year to reviewing them every four weeks are holding two or three points of margin their neighbors lose without noticing. What should you do by size of operation? Single location: one sheet with the ten recipes that carry 70% of sales, recosted the first Monday of every month. Three units or more: automated reading of the supplier invoice that updates the spec sheet with nobody typing. Large United States chains raised menu prices 42% between 2020 and 2025, nearly double the 22% general inflation of that period, according to the One Haus analysis of rising check averages.

Raising prices is not the only lever, but the market already pulled it

That figure serves the independent operator in two opposite ways, both true: tolerance runs higher than you assume, and there is a ceiling where the guest stops coming even when the price is technically correct. The tension resolves through composition, not through a flat percentage. Raise the dish that already earns and turns, freeze the one that turns but earns little, retire the one that neither earns nor turns. A surgical 8% adjustment on your five best sellers moves more profit than an even 4% across the whole menu, because the even move also rewards the dishes that were already losing money. Labeling a dish as «most popular» or «chef's favorite» lifts its orders between 13% and 20%, according to NeatMenu in its 2026 menu psychology report. That number turns menu design into a cash lever, provided the label sits on the dish you WANT to sell —high margin, proven turnover— and not on the cook's sentimental favorite.

A menu is read, not admired: the signal written on the card

Another signal from the guest side: in the United States, 38% define a healthy food first of all as a «good source of protein», according to the International Food Information Council in 2025. Put both signals together and the recommendation writes itself: flag your highest-contribution protein dish explicitly. And cap the labels at three or four per menu, because when everything is highlighted nothing is, and the effect evaporates. A kitchen bleeds money from too many references long before it bleeds from low prices, and the cutoff is simple: any dish below 2% of quarterly sales is being financed by the others. Each extra reference drags idle inventory, waste, one more spec sheet nobody updates and station minutes during the peak. The Masterestaurant method that Diego F. Parra applies sorts the menu into four quadrants —high contribution with high turnover, high with low, low with high, low with low— and treats each differently: the first is protected and highlighted, the second is repositioned on the card, the third is redesigned to cut cost without touching price, the fourth leaves with no negotiation.

How many dishes can a menu carry before it loses money?

Trimming twenty references down to twelve usually frees between 10 and 15 points of tied-up inventory. Drinks and add-ons hold the margin once protein no longer can, and matcha is this year's living example:

delivery orders grew 34% during 2025 in the United States, according to Grubhub's Delivered report. A drink with a standardized build and controlled ingredient cost returns contribution percentages no main course can match, and it takes zero hot-station minutes at the peak. Say your menu sells 600 mains a week and barely 12% come with a house drink; if that attachment rises to 30% through a trained suggestion at the moment the order is taken, 108 extra drinks come in weekly. At $5 of contribution each that is $540 a week, without buying a gram of protein or hiring anyone. Adopt three things now and watch the rest. First, a living spec sheet with cost per portion updated monthly, because without it everything else is opinion.

2026 horizon: what to adopt now and what to keep watching

Second, the contribution and turnover matrix reviewed each quarter with point-of-sale data, not with the head cook's intuition. Third, active review management, since every additional star in your rating moves between 5% and 9% of revenue, according to Michael Luca's Harvard Business School research on Yelp. Leave AI demand forecasting under observation, because today it demands two years of clean history almost no independent has, and keep self-order kiosks there too, since their payback only closes above a certain volume. The horizon rule I apply is blunt: nothing enters the operation unless it can be measured in next month's register. The trend you can ignore without guilt is the visual redesign of the menu with AI-generated photography and QR-code digital cards. Not because the technology is bad, but because it solves a problem the average restaurant does not have while leaving untouched the one it does.

The overrated trend: the digital menu and the AI-generated photo

Industry net margin runs between 3% and 6%, according to the National Restaurant Association, and no pretty image recovers a mispriced point of food cost on your three best sellers: that point comes back only through recosting. A perfect photograph over a badly built menu speeds up the sale of the wrong dish, which is precisely the worst outcome available. Build the contribution matrix this week, with ninety days of sales data and the real cost from yesterday's invoice. Aesthetics come later, once you know what you are selling. The traditional method asks «what does this dish cost?». The Masterestaurant method asks «how many dollars does this dish leave per station minute it occupies during the Friday peak?», and that second question rewrites the whole menu. A steak at 42% food cost leaving $14 of margin and firing in four minutes beats a pasta at 24% leaving $6.20 and blocking the pan for eleven, whatever the percentage says.

The differences that actually move cash

I got this wrong for years: I chased the food cost percentage as if it were the only number in the trade, and built menus that looked flawless on the spreadsheet while leaving less money than the messy menu across the street. The percentage is a traffic light, not a destination; the destination is total monthly contribution margin, which comes from unit margin multiplied by units sold and nothing else. The second difference is calendar. Recosting once a year while food commodity inflation has run in double digits across stretches of the last three years, per FAO price index readings, is driving by mirror: you know where cost was, not where it is. AI in the back office solved exactly that, because reading 300 invoices and refreshing 40 recipe cards stopped costing an administrator two days and now costs twelve minutes of human review. And one more criterion almost nobody argues about: reference count.

The differences that actually move cash — in practice

A 70-dish menu is not generous, it is expensive. Every extra line drags exclusive ingredients, walk-in space, mise en place time and spoilage probability, which is why the quarterly cut of the worst 15% usually moves margin more than any price increase. Guests do not miss what they cannot find; they miss what arrives badly cooked.