SATE Institute Menu Pruning Analysis 2026: which dishes to remove from the menu to gain profitability when most of the sector does not run menu engineering: impact and employment

Finding 1 — Which dish should leave the menu first?

Out goes the dish that yields fewer DOLLARS per unit sold than the median of your own menu and that also sells below 70% of what an even split across all items would give it.

That double filter, not the food cost column, drives the pruning. Very few restaurants practice menu engineering with quality and most never have, so a large share of menus on the market rest on a hunch. Meanwhile the National Restaurant Association places median limited-service food cost at 32.4% of selling price (Restaurant Operations Report 2025), a benchmark useful for catching purchasing anomalies, never for deciding removals. Sort by dollars. Then look at the mix.

Finding 2 — Why the food cost percentage lies when you decide what to cut

A low percentage can hide a poor dish, and that trap is the expensive one. Sorted by the percentage column, the first one falls, and the first one is precisely what pays payroll. If your venue serves 900 covers a month and you remove the wrong dish, the till loses roughly USD 5,400 monthly while the menu's average percentage does not shift a single point. The indicator looks better. The bank looks worse. I sorted menus by that column for years myself, until break-even stopped adding up.

Finding 3 — Median contribution margin, not the average

Use the MEDIAN when you calculate the cut line, because a single expensive dish destroys the average. With twenty-eight items and a ribeye yielding USD 31 per unit, average margin climbs enough to condemn half a menu that actually performs; the median absorbs that spike without flinching. The procedure fits on one sheet: price minus recipe cost per portion, column in dollars, sort high to low, mark the middle value. Everything below goes on watch, not to the guillotine. Datassential projects that more than 40% of U.S. menus will feature protein prominently by 2029, so the temptation to keep expensive low-margin plates will grow rather than ease. Cutting at the median guards against that temptation.

Finding 4 — Sales mix corrects what margin cannot see

A high-margin dish nobody orders has no claim to its slot either, which is where the second filter comes in. With 28 items listed, an even split gives each one 3.57% of the mix; the 70% threshold sets the line at 2.5%. Below that floor, the item eats menu space, ties up raw material and forces a slow-rotating purchase. Before signing the removal, check placement: where the dish sits in its category also drives orders regardless of price, and Cornell (Wansink) documented a +27% sales lift when the dish carries a descriptive name. Move the item to the head of its category, write it properly and let a full four-week cycle run. If it stays under 2.5%, then yes.

Finding 5 — What happens if you cut the menu in half

Picture the exercise taken all the way: twenty-eight items down to fourteen. Purchasing concentrates, inventory turns double on the same references, waste falls and food cost variance—that gap between theoretical recipe cost and the real cost inventory reveals—finally becomes traceable, because with fourteen recipes you can physically count what left. The kitchen gains seconds per ticket. And there is a real risk worth stating instead of dressing up: you lose niche guests, perhaps 3% or 4% of traffic, especially if you removed the only plant-based option. Watch the figure from Plant Based Foods Association and Datassential (2024), which recorded 48.4% of U.S. restaurants offering a plant-based alternative. A short menu missing that slot is not discipline, it is a hole.

Finding 6 — A shorter menu also raises price, and that is a tension

Two uncomfortable truths live side by side here. Raising prices is the sector's dominant reflex: TouchBistro recorded in 2024 that 47% of operators did so in the six months before its cutoff, Toast measured 42% in 2023, and Oysterlink compiled that 93% of U.S. quick-service restaurants adjusted rates during 2024. Yet menu inflation keeps moderating, at +0.2% monthly in full service and +0.3% in limited service according to the National Restaurant Association via Restaurant Business (2026), which means margin no longer arrives through price. The bridge between both ideas is pruning: removing the low-margin, poor-mix dish lifts average check without touching a single rate, because the guest spreads spending across fewer options and every one of them performs.

Finding 7 — What Masterestaurant contributes on top of public data

The figures in this analysis come from verifiable third parties; the reading order is ours. Diego F. Parra and the technical team at Masterestaurant S.A.S. hold a criterion that contradicts the usual manual: menu engineering does not start with the four-quadrant matrix but with a dollar column and a physical inventory count, because without real recipe cost the matrix classifies noise very elegantly. The mistake that shows up again and again in small-business menus is loading payroll, rent and utilities onto the dish to work out its cost; that does not belong to the dish, it belongs to break-even, and mixing them inflates apparent food cost to 55% and condemns healthy plates. Maximum tolerable food cost per dish is 32%, and it is a ceiling, not a target.

Finding 8 — Why this is an employment indicator, not a kitchen detail

An oversized menu degrades the credit scoring of a small gastronomic business, and from there it jumps to payroll. The mechanism runs straight: slow-rotating inventory tied up, food cost variance never tracked, financial statements showing fragility where there is only measurement disorder. For SATE Institute that reading crosses SDG target 12.3, aimed at reducing waste, with the decent work target of SDG 8, because capital sleeping in a walk-in hires nobody. And the starting figure deserves repeating: 60% of restaurants do no menu engineering, according to Oracle NetSuite. Open your last quarter's sales report, pull the dollar margin per unit column and mark the median. That single line will tell you which dishes you are subsidizing.

Finding 9 — Four differences between profitable pruning and blind cutting

Percentage versus currency: that is where most operators lose the thread. A plate running 38% food cost that leaves USD 9 per unit contributes more to break-even than one at 26% leaving USD 3, yet the first is the one that usually falls when an owner sorts the menu by the food cost column. The sector reference from the National Restaurant Association Restaurant Operations Report 2025 detects purchasing or portioning anomalies; it is not an elimination criterion. Prune by percentage and you end up with a cheap menu that cannot cover payroll. Second, position gets corrected before anything gets cut. NeatMenu documented in 2026 that the first entrée in each category captures 33% of orders independent of price, so a dish buried at the bottom of its block arrives at quarterly review with an artificially depressed mix. Moving it costs one reprint. Removing it costs a margin you never recover. Before dropping any reference, run two cycles with that dish at the top and its copy rewritten.

Finding 10 — Four differences between profitable pruning and blind cutting — in practice

Third, the food cost variance nobody invoices. When a menu drags slow-moving references along, the exclusive input behind each one gets purchased, stored, spoiled and discarded without accounting ever separating it from general food cost. SDG target 12.3 and the IDB Group's #SinDesperdicio initiative put regional numbers on that loss; for the individual operator it is simply cash evaporating between the walk-in and the dish pit. Pruning well is, measured this way, an environmental intervention with immediate financial return. And the fourth carries macro consequence: scoring. A restaurant with an oversized menu shows low inventory turnover, immobilized working capital and margins that swing quarter to quarter, a profile commercial banks with MSME portfolios read as high risk. Diego F. Parra presses this point whenever a Masterestaurant diagnostic reaches a credit committee: menu pruning does not merely change this month's EBITDA, it changes the financial picture that MSME presents when it requests working capital next year.