Profitable menu: the criteria to build it (before and after you measure dish by dish)

Step 1 — Build the recipe cost sheet and calculate margin in DOLLARS, not percentage

The deliverable of this first step is a sheet with three columns per dish: ingredient cost, selling price and contribution margin in dollars, and it is not done until all 61 dishes on the menu have the three filled in without a single cell estimated by eye. The criterion that orders everything else is easy to say and hard to accept: a dish with 31% food cost that leaves 14 dollars is worth more than one with 22% that leaves 5, because payroll and rent are paid with dollars, not with percentage points. Set the 32% maximum cost per dish as your purchasing control ceiling, and verify the step by checking that the weighted sum of mix margins covers your monthly fixed costs; if it does not, the problem is not your meat supplier. The Kasavana and Smith matrix crosses contribution margin with popularity and remains the most honest instrument available for deciding what stays on the menu, and that cross is the second deliverable: every dish classified into one of four quadrants, with a cutoff date and real sales from at least 90 days.

Step 2 — Cross margin with popularity on the menu engineering matrix

Pull the sales report from your POS, sort by units sold and flag the six dishes that concentrate the bulk of the mix. In the case that opens this guide, the top five represented 41% of orders and left less margin than a low-rotation soup. You know the step is done right when you can point, without hesitating, at which dish is a star, which a plowhorse, which a puzzle and which a dog. Redesigning means a new recipe card: revised portions, recalculated garnish, confirmed supplier and recalibrated price, dish by dish, and the deliverable is six signed cards with the target margin written at the top. You do not sell your menu, you sell those six dishes that concentrate between 38% and 52% of orders; if they all sit in the low-margin quadrant, no across-the-board price increase saves you. TouchBistro reported that 47% of restaurants raised menu prices during the last six months of 2024, and some of them landed back in the same hole three months later because they moved the price without touching the recipe.

Step 3 — Redesign one by one the six dishes that rule your mix

Full-service menu inflation runs at +0.2% monthly according to the National Restaurant Association: that pace does not offset a badly designed dish. Here the cut is decided with the purchasing list in hand: count how many pantry references sustain each dish and pull first the ones dragging exclusive low-rotation inputs, because that is where the waste no supplier negotiation can fix actually lives. When 180 references sustain 61 dishes, food cost variance becomes structural. Aggregated menu design research places the healthy range between 7 and 15 items per category to avoid decision paralysis, and the average guest spends 109 seconds reading the menu according to NeatMenu: nine appetizers nobody has ordered in eight months are stealing attention from the ones that actually pay. The deliverable is a kill list signed by the kitchen and by the books, with the reference savings calculated.

Step 5 — Time the plating during peak service before confirming the new menu

Time every surviving dish in real service, between 8:00 and 9:30 p.m., and write the seconds next to the margin: that third number is the one almost nobody measures and the one that decides whether the menu works in real life or only on the spreadsheet. A high-margin dish that jams the line at 8:40 costs far more than its recipe card admits, because it delays the following tables and drags the rotation of the whole night. Consider the contrast with drive-thru: the industry went from 6 min 13 s in 2022 down to 5 min 29 s in 2024 according to Intouch Insight, and it got there by cutting operational complexity, not by raising prices. The step is verified when no top-selling dish exceeds the target time you set for your line.

Step 6 — Write the names and lay out the page with selling criteria, not aesthetics

A descriptive name is worth measurable money: Cornell's Food & Brand Lab documented that guests pay on average 12% more for a dish with a worked description, so the deliverable of this step is rewritten copy for your high-margin dishes, with ingredient, technique and origin. Place those dishes in the upper zone of each category and strip the decorative boxes around the cheap ones. Diego F. Parra insists at Masterestaurant that a menu is a financial instrument disguised as a graphic piece, and the designer should receive the menu engineering matrix before opening the file. Check the allergens as well: sesame has been the ninth major allergen requiring mandatory labeling in the United States since 2023 under the FDA's FASTER Act. The most repeated mistake is raising every price 5% and calling it menu engineering, when what belongs here is redesigning six dishes and leaving the rest alone.

The mistakes that sink the exercise and how to avoid them

The second is cutting by the chef's perception instead of by POS data. The third, and the most expensive, is adding new dishes without retiring any: the menu grows by accumulation and the cost hides inside the waste. And here I make a concession: for years I recommended short menus on principle, and I got the nuance wrong, because the number of dishes matters less than the number of pantry references that sustain them. A 40-dish menu with 60 shared inputs runs better than a 24-dish one with 110 exclusive inputs. Avoid them with one rule: a dish comes in, a dish goes out, and the pantry does not grow. The menu is ready when you can answer five questions without opening a file: which dish leaves the most dollars, which six concentrate the mix, how many pantry references sustain the operation, which dish is slowest at peak hour and when the next review happens.

Closing checklist — how to know the menu came out right

Verify as well that no dish exceeds 32% cost, that the weighted mix margin covers fixed costs and that no category runs past 15 items. Book the review at 90 days with the POS report open, because the mix drifts on its own with the season. What happens if you skip that date? Limited-service inputs climb at +0.3% monthly according to the National Restaurant Association, and within a year that silent drift hands you back exactly the menu you just dismantled, dish by dish. CRITERION 1 — Margin outranks percentage. A dish at 31% food cost leaving 14 dollars beats one at 22% leaving 5. Percentage controls purchasing; dollars pay payroll and rent. The classic Kasavana and Smith menu engineering matrix crosses contribution margin with popularity, and it remains the most honest instrument available for deciding what stays. CRITERION 2 — Sales mix weighs more than the printed page. You don't sell your menu: you sell the six dishes that concentrate between 38% and 52% of all orders.

The seven criteria that decide whether the menu makes money

If those six sit in the low-margin quadrant, no across-the-board price hike saves you; the right move is to redesign those six, one at a time, with a fresh spec sheet. CRITERION 3 — The pantry is part of the design. Every new dish demanding an exclusive input adds invisible waste. House rule: no ingredient enters the menu unless at least three live recipes use it. That criterion alone drops 180 pantry references below 95, and food cost variance falls with them. CRITERION 4 — Line time is a cost. A high-margin dish eating four minutes of grill at peak blocks three other tickets. Time plating seconds per dish during one Friday service; you will find two or three of your matrix stars are operational brakes in disguise. CRITERION 5 — Pricing psychology is architecture, not a trick. Guests don't compare against the market, they compare against what sits in front of them: a deliberate high anchor in each section pushes the choice toward the dish you actually want to sell.

The seven criteria that decide whether the menu makes money — in practice

Dropping the currency symbol, avoiding a right-aligned price column and tucking the figure right after the description are measurable decisions, not aesthetics. CRITERION 6 — Paper rules the dining room; the QR complements it. I got this wrong for years, recommending full digitization to save on printing. Paper controls service pace, carries the menu narrative and enables the server's suggestive selling; the QR solves delivery, accessibility, same-day price changes and click analytics. They travel together, each in its role, and whoever kills the printed menu loses control of the experience. CRITERION 7 — Review is monthly or it doesn't exist. A menu frozen for six months accumulates cost drift silently. With AI dashboards reading the POS, the alert lands the same day a dish crosses its ceiling, and the call to reformulate, reprice or retire gets made on fresh data instead of on a feeling at accounting close.