Menu Pricing That Recovered 9.4 EBITDA Points: how we cleaned up a trattoria that billed well and lost money using the Restaurant Model Canvas and the Standard Recipe Generator

The diagnosis: strong sales, nothing left over

This fourteen-table trattoria billed well and still closed the year with a 2.8% EBITDA, dining room full most nights. «I bill more than ever and it's never enough»: the owner's line sums up half the industry, and it described neither a demand problem nor a marketing one. It described pricing set by gut feel. With an $18 average check and the dining room generating 68% of sales, every dish reached the table at a number decided out of habit, with no real food cost behind it. Prime cost, food plus labor, already brushed 71%. To size just the labor weight: full-service cost hit a median of 36.5% of sales in 2024 (National Restaurant Association, Restaurant Operations Data Abstract 2025). Without a price built to absorb that figure, the business works for everyone except its owner. Before calculating anything, the traditional method had already decided the price: dish cost times three, or simply whatever the place across the street charges, with the margin assumed along the way.

The root error: pricing before costing

We reverse that order. Real food cost and the required contribution margin come first; the price shows up afterward, as the output of the calculation. Measuring recipe by recipe in this trattoria surfaced an 8.9-point gap between theoretical and actual cost: money evaporating in production that no one could see, per the case data. Where traditional costing applies one percentage to the whole menu, ours measures every dish on its own. The sector benchmark places labor cost between 25% and 35% of revenue depending on format (U.S. Bureau of Labor Statistics), and in QSR it climbed 6.3% in 2024 on minimum-wage hikes (National Restaurant Association, 2024). Ignoring that number when pricing dooms EBITDA to anemia. No 'decent' price added up in this trattoria, and the reason has a name: prime cost, the variable traditional pricing simply ignores. With food plus labor at 71%, nothing was left for rent, utilities, insurance, let alone the owner.

The forgotten constraint: prime cost as the ceiling

We work toward a target prime cost of 60-62% as a prior ceiling, and from there calculate the contribution margin each dish must deliver before the menu gets touched. The fixed costs a price has to cover aren't abstract: electricity averages $2.90 per square foot a year and natural gas $0.85 (Toast, Average Restaurant Electricity Bill 2025); a business owner's policy runs about $3,000 annually, with general liability near $900 and property coverage at $740 (MoneyGeek, Restaurant Business Insurance Cost 2025). Food waste adds another $72,000 a year on average (The Restaurant HQ, 2025). A gut-set price never accounts for that cost floor, and that is exactly where EBITDA disappears. Raising every price at once would have scared off customers, so the fix was surgical: menu engineering across the existing menu, dish by dish. We classified each recipe by popularity and real contribution margin, then repositioned accordingly.

The action: menu engineering, not blanket hikes

Star dishes, high demand and strong margin, were protected and made more visible on the menu; the ones that sold well but left little were redesigned in portion and recipe to recover food cost points without touching their price. Only three dishes went up, by $1.40 on average, a nearly invisible move on an $18 check, as the case shows. That step, paired with standardizing portions and yields, closed the 8.9-point gap between theoretical and actual cost. In parallel we rescheduled shifts against the real dining-room demand curve. Recall that median full-service labor runs 36.5% and limited-service 31.7% (National Restaurant Association, Restaurant Operations Data Abstract 2025). Price stopped being a hunch. Five months in, EBITDA had climbed from 2.8% to 11.4% without losing traffic or lowering the average check, per the case. Aggregate food cost gave up 6.1 points as the theoretical-actual gap closed, and prime cost fell from 71% to 63% by combining standardized recipes with demand-matched shifts, per that same record.

The measurable result: from 2.8% to double digits

None of it came from selling more: it came from pricing with method on the same fourteen tables. For the owner the shift showed up in the register, not on a spreadsheet: a margin that used to be swallowed by any surprise now leaves a cushion for reinvestment. Worth the industry context: opening a restaurant today costs a median of $375,000, or $113 per square foot (Rezku, 2025), capital only recovered with healthy margins. And sector costs will keep pressing through 2026 despite resilient demand (Bloomberg Línea), which turns disciplined pricing into a defense, not a luxury. Masterestaurant's costing and menu-engineering system, available in the restaurant tools ecosystem, was the central piece of the whole process. It was applied in three concrete steps. First, we loaded each recipe with real portion weights and current purchase prices, and that exposed the 8.9-point gap between theoretical and actual food cost, per this case's numbers.

The tool used and how it was applied

Next, the tool calculated contribution margin per dish and placed each recipe on the popularity-margin matrix, flagging what to protect, what to redesign, and, in a few cases, what to pull from the menu altogether. Finally it set the target prime cost at 62% as a constraint and returned the minimum viable price for each dish, instead of the owner guessing it by eye. That's the heart of the method: food cost and contribution get decided first, and price is merely the output of that calculation. With sector labor cost running 25% to 35% of sales (Toast, Restaurant Payroll Guide), the tool forces the price to absorb that figure before the menu goes to print. The first step changes with the size of the operation, but the lesson travels the same. If you run a single dining room under twenty tables: cost your five best-sellers by hand this week, using real portion weights and today's purchase prices; a multi-point gap almost always turns up, the way this trattoria's 8.9 points did.

Transferable lessons by the size of your operation

If you manage one or two locations with on-site managers: put per-dish food cost into your system and set a target prime cost of 60-62% as a rule before any menu change. Labor alone already runs a median 36.5% in full service (National Restaurant Association, 2025). If you lead a multi-unit group: standardize recipes and the menu-engineering matrix across locations, and audit the theoretical-actual gap by site every month; waste averages $72,000 a year per restaurant (The Restaurant HQ, 2025), and at that scale it's serious capital. In all three, price gets calculated. It doesn't get copied. This result isn't universal, and it's worth saying so to avoid survivorship bias: I got this wrong for years myself. I assumed fixing food cost was always enough on its own. First, it worked because hidden margin existed inside an operation with solid demand; in a business with weak traffic or a bad location, fixing pricing won't create customers, and EBITDA won't jump to double digits from better costing alone.

Limits of this case: where I wouldn't expect the same

Second, this trattoria was dining-room dominant (68% of sales), where the check and value perception tolerate a fine adjustment; in a pure delivery model, platform commissions, often 20-30%, and price wars compress the margin and call for a different strategy, not just menu engineering. Third, we're talking about a full-service format; in high-volume, low-ticket QSR, where labor rose 6.3% in 2024 (National Restaurant Association, 2024), the main lever is usually productivity and speed, not menu price. The method stays the same. What changes is the size of the result. Setting the sale price is where the traditional method starts; in ours, that same number is the consequence. We calculate the real food cost and the contribution margin each dish must yield first, and only then does a price land on the menu. A single theoretical food cost governs the entire menu under the traditional method, with no distinction dish to dish.

What separates pricing by habit from pricing by method?

We measure recipe by recipe, and in this trattoria that measurement uncovered an 8.9-point gap between theory and reality: money vanishing in the kitchen that no one was tracking.

Prime cost doesn't even enter the traditional equation. With food plus labor at 71%, no reasonable sale price left room for EBITDA. Charging a bit more wouldn't have closed that gap either. We set the target prime cost as a prior constraint, before touching a single line on the menu. Fear of losing customers freezes prices for years under the traditional method. Menu engineering, instead, lifts the ticket exactly where it doesn't hurt, at high-margin star dishes, while protecting the perceived value of everything else on the menu.