SKU Reduction and Short Menus: the before and after numbers

How many dishes are genuinely surplus on a 42-item menu?

The surplus sits in the bottom quartile of absolute contribution margin, usually 14 to 18 dishes on a 42-item menu, and pulling them cuts waste by 3 to 6 points of food cost.

Here is the arithmetic almost nobody runs: every dish drags along two to five exclusive ingredients, so 42 dishes averaging three exclusives each means roughly 126 slow-moving SKUs sleeping in the walk-in. The National Restaurant Association put 2026 industry food waste between 4% and 10% of total purchases, and that range is not spread evenly, it concentrates in references turning less than once a week. On monthly purchases of 60,000 USD, moving from 9% waste to 5% frees 2,400 USD every month without touching a single sale price. What these figures together should trigger is simple: rank the menu by dollars of margin per unit sold before anyone argues about cuisine. A slow-moving SKU costs far more than its purchase price, and that gap is what pulls real food cost away from the theoretical number on your spreadsheet.

The cost of carrying a SKU nobody orders

Take an exclusive ingredient at 40 USD a case, bought only because one dish demands it: a minimum order that forces you to overbuy, a stretch of refrigerated shelf, an expiry date that eats 15% or 20% of the case, and three minutes of prep your cook no longer has at the peak. Multiply that by the 126 exclusive references behind a 42-dish menu and you see why food cost refuses to drop below 36% even after two price rises this year. With menu inflation running at +3.5% year over year as of May 2025 (National Restaurant Association / Restaurant Business), raising prices is no longer the available lever. The lever lives in inventory, not on the price board. ABSOLUTE margin in dollars per unit sold outranks the percentage, and mixing them up is the mistake that turns sound cuts into destructive ones. A dish at 28% food cost priced at 14 USD returns 10.08 USD of gross margin per unit; another at 35% food cost priced at 32 USD returns 20.80 USD.

Percentage margin versus absolute margin: the 28% trap

Classic menu engineering, sorted by popularity and percentage, would flag the second one for removal, and you would be deleting the dish that pays the rent. At Masterestaurant, Diego F. Parra insists on measuring every dish against the three scarce resources it consumes, walk-in space, cook minutes and inventory lines, versus the dollars it gives back. Once those four numbers share one sheet, the shortlist writes itself, and it usually includes something you happen to love. Sort by dollars per unit, never by percentage, before you cross anything out. There is a real tension here and I would rather state it: a short menu cuts cost but also cuts coverage, and a 24-dish list with no vegetarian option, no gluten-free option and no celebration plate loses the whole table, not one cover. The evidence carries weight, since among guests with food allergies 36% always return to the same venue against 17% of those without them (Food Allergy and Foodservice study, PMC), which means covering restrictions buys loyalty rather than mere tolerance.

Cutting without losing the party of eight: the occasion axis

The answer is not to cut less. Cut by occasion axis instead: define the consumption blocks your house serves, shared starter, individual main, celebration, restricted option, and inside each block keep whichever dish shows the best absolute margin. That way the savings on waste are not paid for with lost tickets, which is exactly what happens when the cut is made by staring at the menu instead of the sales mix. Your reduced menu turns fragile when 24 dishes hang from three suppliers, because one stockout on a Thursday no longer removes a single plate, it removes eight, and that means shutting half the kitchen mid-service. The rule I apply when validating a cut: no ingredient family should carry more than two dishes on the pass, so a supplier failure has a known ceiling. The risk is far from theoretical with protein, where the US cattle herd sits near 86 million head, the lowest since the 1950s according to the USDA (2025), with everything that implies for the cost of a beef plate.

A badly built short menu concentrates supply risk

And with over a million foodservice locations chasing the same supply (National Restaurant Association, 2025 Forecast), emergency replenishment is expensive. Spread your ingredient families while you design the short menu, not after the first stockout teaches you. The operational payoff of a short menu is collected in service time, and that is where the chef-owner feels it long before the P&L shows it. Fewer dishes mean less mise en place, fewer preparations dumped at closing and a learning curve a new cook finishes in two weeks instead of two months. When the kitchen carries 24 dishes built on shared ingredient families, one stock, one braise and one vegetable base feed several plates at once, and that overlap is what pushes food cost down without touching the sale price. A warning belongs here: overlap too much and the menu flattens, which a regular notices by the third visit. Two or three dishes per ingredient family is the point where efficiency has not yet eaten the personality of the restaurant.

What a short menu frees up on the pass?

Measure pass minutes before and after the cut, because that number tells you whether the design landed.

Should the average check drop after the cut, the fault was not cutting but what you cut, and the fix is surgical rather than a full reversal. Say you removed the two highest-priced dishes because they carried 35% food cost: you lost the menu's high anchors, the guest reframed spending downward, the check fell 9%, and you gained three points of food cost while losing more gross margin than you gained. The repair sequence is to restore one high anchor, not two, measure four weeks and compare absolute margin per cover, never percentage. Presentation pushes in the same direction, since removing the dollar sign from the menu lifted per-person spending by 8.15% in the Cornell University, School of Hotel Administration study (2009). A short menu is a pricing instrument as much as a cost instrument, and whoever treats it only as the latter leaves money on the table.

The 3 figures worth tattooing on your arm

Three numbers govern this decision and each one demands a concrete action this week. First: 4% to 10% of purchases disappear as waste (National Restaurant Association, 2026); action, pull the list of references turning less than once a week and mark which dish holds each one alive. Second: +3.5% menu inflation year over year as of May 2025 (National Restaurant Association / Restaurant Business); action, stop expecting the sale price to fix food cost and work the SKU count instead, which is the variable you actually control. Third: +8.15% in per-person spending when the dollar sign disappears (Cornell University, 2009); action, when you redesign the 24-dish menu, drop the symbol and keep both high-price anchors. Start with the first one today, because the slow-rotation list fits into one afternoon and it decides which 18 dishes leave. Cutting by instinct removes the dishes the chef considers weak; cutting by marginal profitability per dish removes those consuming scarce resources without returning contribution margin, and they are almost never the same list.

Five differences that decide the outcome

That distinction explains why two restaurants that both drop 18 dishes end up with opposite results. A badly built short menu concentrates risk: if your 24 dishes lean on three suppliers and one fails on a Thursday, the kitchen shuts. A well-designed reduced menu spreads input families so no single stockout pulls more than two dishes off the pass. Classic menu engineering sorts by popularity and margin percentage; real portion costing sorts by ABSOLUTE margin in dollars per unit sold. A dish at 28% food cost selling twelve units a week contributes less cash than one at 34% selling ninety, and that arithmetic should outrank any personal preference. Price psychology behaves differently across 24 mental pages than across 42: with fewer options the anchoring effect of the priciest dish sharpens, guests decide faster, and table turns rise without touching a price. Demand elasticity for a removed dish is neither zero nor one: between 55% and 70% of guests who ordered the withdrawn item migrate to another dish on the same menu when the occasion block stays covered.

Five differences that decide the outcome — in practice

That recapture rate is the single number deciding whether your cut was profitable or suicidal.