Which menu items to cut for profitability: the numbers that decide it in 2026

Which dish should leave the menu first?

The first dish to go is the one whose contribution margin in dollars falls below the menu median and that also sells less than 70% of what an equal split would give it, measured across 90 days of comparable trading.

That crossover matters because it separates the cheap dish that sells constantly from the expensive dish nobody orders, and those are different problems with different fixes. Discipline pays: Cornell documented that menu engineering applied with method lifts profitability by roughly 10%, and Oracle NetSuite places the range between 10% and 15% when the exercise is sustained over time. With a 22 USD average check and 4,500 covers a month, that 10% is 9,900 USD a year currently draining out through the tail of your menu. Use the median, never the average: one 40 USD plate distorts any mean. Between 65% and 75% of units sold in a full-service restaurant come from 20% to 25% of the menu references, and that proportion repeats with almost geological stubbornness in kitchens of any size.

GROUP 1 · Sales concentration is not an opinion

The case opening this piece shows it without mercy: 19 dishes out of 84 moved 71% of units, while 65 references split the remainder and dragged 38 exclusive ingredients through the walk-in with their weekly shrink and their idle capital. A long tail diversifies nothing, it occupies. Every dead line eats refrigerated space, mise en place minutes and, above all, the cook's attention during peak service. Group takeaway: if your menu runs past 60 lines, count on roughly 35 that probably fail to pay for the inventory count they cause, and those 35 are your Monday work list. Every additional dish on the menu drags an average of 1.7 exclusive ingredients behind it, and that starts a cost chain no food cost software will ever show you on a single screen. ReFED and the Food Waste Reduction Alliance put food waste in full-service restaurants between 4% and 10% of food purchases, and the avoidable portion of that shrink concentrates precisely in the slow-turning ingredients that exist only for two or three tail dishes.

GROUP 2 · What holding a dead reference really costs

If you buy 38,000 USD in food monthly, we are talking about 1,520 to 3,800 USD hitting the bin every month. Add the weekly count, the walk-in shelf and the supplier returns. Group takeaway: pulling 20 low-rotation references frees around 34 ingredients and cuts avoidable waste before you touch a single recipe. Some 62% of operators still decide what stays on the menu by looking at each dish's food cost percentage, and that habit destroys margin with perverse elegance, because the percentage rewards cheap low-impact items and punishes exactly what pays the payroll. Run the numbers yourself: a salad at 25% food cost on 12 USD leaves 9 USD; a steak at 38% on 34 USD leaves 21.08 USD. The percentage says the salad is the better dish. The till says you need 2.3 salads to match one steak, and your kitchen does not have infinite shifts.

GROUP 3 · The mistake of deciding by food cost percentage

Diego F. Parra works this cut at Masterestaurant with a rule that unsettles chefs: the menu is not a catalogue of what you know how to cook, it is a financial instrument where every line pays rent. Group takeaway: re-sort your analysis by dollar margin before touching anything. Menu prices at large U.S. chains climbed 42% between 2020 and 2025, nearly double the 22% general inflation of that period according to the One Haus analysis of rising check averages. That figure carries an uncomfortable reading for the independent operator: if the chains pushed price that aggressively and you did not, your 2020 recipes are costed against ingredient prices that no longer exist and your real margin is thinner than your spreadsheet claims. Before deleting a dish, check whether the problem is the dish or the frozen price. A simple counterfactual: raise prices 8% on your twelve star dishes and lose 4% of the units, and total contribution still climbs, because unit margin grows faster than volume falls.

GROUP 4 · Pricing, inflation and dishes that no longer work

Group takeaway: re-cost before you cut. Not every slow-moving dish is a bad dish, and I got this wrong for years, cutting by volume without first checking whether the trouble sat in the menu presentation rather than in the recipe. Cornell's menu design research measures up to 30% more sales for a dish paired with a photograph, with roughly 6.5% incremental lift per dish when the photo is professional, and digital channels amplify that effect: upselling through QR or app ordering raises average order value by 20% to 30% according to aggregated digital ordering provider data. A dish with high margin and low mix deserves 30 days of repositioning on the page, a photo and a server suggestion before the axe. Group takeaway: rescue the high-margin, low-mix dishes first, and delete the low-margin, low-mix ones without ceremony. Cutting the menu pays for itself on the payroll line before it does on the purchasing line, and that is the effect almost nobody models when pitching the project to a partner.

GROUP 6 · What happens in the kitchen when the menu shrinks

Base hourly pay in U.S. restaurants rose 4% to 14.20 USD an hour in 2024 according to the 7shifts workforce report, so every minute of mise en place costs more each year. Fewer references means fewer parallel preparations, less training for the new cook, fewer blown ticket times at the nine o'clock peak. If your kitchen preps 84 spec sheets today and drops to 55, an eight-hour cook recovers close to 50 minutes daily: about 21 USD per shift, 630 USD a month per person. Group takeaway: count the labour saving in the same table where you count the shrink, or you will undersell the project. First: 70% of the equal split. Divide 100 by your number of dishes, multiply by 0.7, and any reference selling below that threshold across 90 days joins the candidate list. Monday action: export last quarter's sales mix and flag the rows.

The 3 numbers worth tattooing on yourself

Second: the median contribution margin in dollars. Rank your dishes by what each one leaves in cash, not in percentage, and draw the line at the midpoint; whatever sits below it and also sells poorly goes. Action: re-cost your ten oldest recipes using this month's invoices. Third: 10% additional profitability, the return Cornell attributes to disciplined menu engineering and that Oracle NetSuite stretches to 15% when the exercise repeats every quarter. Action: book the next review 90 days out and refuse to move it. GROUP 1 · Sales concentration. On the average full-service menu, somewhere between 65% and 75% of units sold sit in 20-25% of the lines. A long tail is not variety, it is sleeping capital: every item down there demands ingredients, walk-in space, and a slot in the cook's head during service. Takeaway: if your menu runs past 60 lines, odds are overwhelming that 35 of them fail to cover the inventory count they generate.

The numbers, grouped by the decision each one triggers

GROUP 2 · What dead lines really cost. Each additional menu item brings roughly 1.7 exclusive ingredients along with it, and food waste in full-service restaurants runs between 4% and 10% of food purchases per ReFED and the Food Waste Reduction Alliance. An ingredient feeding a single item that moves 12 units a month is overwhelmingly likely to end up in the bin. Takeaway: count orphan ingredients rather than items, because that is where the money evaporates. GROUP 3 · Aggregate margin impact. Operators who reconcentrate sales into the top margin third report gross margin gains of 2 to 4 points without touching prices, since the mix shifts on its own once the price-cannibalising options vanish. The arithmetic is plain, and most operators skip it: moving 100 guests from a 4-dollar margin plate to a 9-dollar one is worth 500 dollars a month, more than any campaign.

The numbers, grouped by the decision each one triggers — in practice

GROUP 4 · The elasticity nobody tests. Before deleting a beloved thin-margin plate, price is the cheap route: in entree categories, increases of 8-12% typically move volume less than 5% when real attachment exists. Raise the price, watch demand hold, and what you had was never a profitability problem but a pricing one. Takeaway: test price for 30 days before touching the menu, because deletion is irreversible in the guest's mind and a price move is not. GROUP 5 · The risk of cutting blind. Pulling an item with identified repeat purchase above 12% among your regulars costs more in lost frequency than it saves in walk-in space. I got this wrong for years: I recommended aggressive matrix-driven cuts without checking per-guest recurrence, and in two cases regular frequency dropped enough to wipe out the margin gain. Takeaway: cross the matrix with repeat data before signing the deletion.