SKU reduction and short menus: the before and after numbers

Your supplier did not set that 38% food cost; your 96-item menu did

A 96-item menu in a 70-seat dining room drags between 380 and 1,050 purchase references behind it, and every reference gets paid for with supplier minimums, occupied walk-in space and an expiration date already running. When SKU reduction cuts that menu down to 22 sellable lines, food cost gives up 2 to 5 percentage points within the first 90 days and perishable waste drops between 25% and 40%, because each ingredient leaving the menu takes with it a purchase, a storage slot, a prep task and a risk. The decision is concrete: pull the last 60 days of sales, sort by units sold and flag every line under 8 portions per week. That bottom block, usually 30% of the menu, is the one buying inventory you end up throwing away. A SKU is every distinct reference coming through the back door, not every pretty name on the menu: fresh salmon and smoked salmon are two SKUs, 35% cream and 18% cream are two SKUs, even if the chef writes them the same way on the standard recipe.

A SKU is not a dish, and mixing them up costs you two points of margin

That distinction rules everything, since cost is created at purchase and not on the menu. With food-away-from-home prices climbing 3.8% in 2025 (USDA Economic Research Service, Food Price Outlook 2025) and Colombian menu prices adjusting 9.8% since February (ACODRES 2025), each extra SKU multiplies your exposure to that inflation. Run the real count this week: export active references from your purchasing software, not from the menu, and set that number beside the 22 lines that actually sell. The gap between both figures is your hidden storeroom. You calculate that the mushroom risotto costs 4.20 USD, price it at 16 USD and record a theoretical food cost of 26%. Flawless on paper. What the paper never records is that this risotto moves seven portions a week, that porcini arrives in a 1 kg case with a 9-day shelf life and that you discard 60% of every case: load that waste onto the dish and real cost climbs toward 61%.

Portion costing lies until waste is charged to the dish that produces it

Repeat the exercise across your thirty long-tail dishes and the margin your system promised but never delivered finally shows up. The operational decision fits in one column: add a real waste field per reference to your spec sheet, measured for two weeks with a scale instead of memory, and recost. Dishes crossing 45% real cost leave the menu the following Monday. Long menus split demand across too many lines, so no single dish reaches the volume that moves a supplier's price. Concentrate sales into 22 dishes and that volume turns into discounts of 5 to 12 points on your core references, without switching suppliers or raising your voice in a negotiation. The mechanics are plain arithmetic: 40 monthly kilos of one cut carry weight in a quote, 4 kilos carry none. Diego F. Parra keeps insisting at Masterestaurant that a short menu gets designed from the shopping list toward the menu and never the other way around, because savings are born in the order, not on the plate.

Concentrating the sales mix is the one purchasing lever that needs no hard negotiation

Start by identifying your six core references, the ones appearing in more than three dishes, and renegotiate only those with the consolidated volume your short menu just handed you. Here sits the paradox a short menu resolves: operators keep long menus so they will not lose sales, and that long menu is precisely what makes it impossible to sustain the costing that would protect those sales. With 22 spec sheets, a head chef reviews weights and purchase prices every quarter in a single afternoon. With 96, nobody reviews them, portion costing goes stale six months after it was written, and input prices keep moving 4.1% a year (USDA Economic Research Service, Food Price Outlook 2024). The outcome is a menu that believes it costs 29% and actually costs 38%. Set a rule that admits no excuse: no line enters the menu without a costed spec sheet, and every sheet gets recosted the first Monday of each quarter using real invoice prices.

What would happen if you cut 40 dishes next Tuesday?

Assume you pull the 40 slowest-moving lines. Week one: the kitchen stops ordering 120 to 300 references and the weekly purchase order falls between 15% and 22% while sales stay flat, because those lines carried under 8% of total units.

Week two: the uncomfortable effect arrives, three regulars ask for their dish and you lose between 1% and 3% of that segment's check. Week six: ticket times drop because the line works with less mise en place, the kitchen shift covers with one person fewer per service, and labor savings stack on top of food savings. You trade 2% of check for 3 points of food cost and a lighter payroll. That trade pays and it holds, provided you measure both sides across the full 90 days. Ranges shift with your structure, so locate yourself before applying them. Small restaurant, up to 60 seats and a single kitchen shift: the food cost drop lands at the low end, 2 to 3 points, because your volume negotiates little discount, though waste pays you back hardest, close to 40%.

How to read these numbers in YOUR operation: small, midsize, group?

Midsize operation, 80 to 150 seats across two services: this is where the full 5 points show up, since consolidated volume triggers those 5 to 12 point discounts on core references.

Groups of three or more locations with centralized purchasing: food savings moderate to 2 or 3 points, as much of the volume discount was already captured, and the gain migrates to the storeroom and to labor, where AI-assisted scheduling trims another 8% to 12% (TimeForge 2025). The ranges in this piece combine three sources of different nature and it helps to know which is which. Input and menu inflation figures come from the USDA Economic Research Service (Food Price Outlook 2024 and 2025) and from ACODRES 2025 for Colombia, public and auditable series. The labor savings figure for assisted scheduling comes from TimeForge 2025, a software vendor, so read it as the upper bound of a favorable case rather than a sector average.

Where these benchmarks come from and how far they reach?

Food cost and waste ranges reflect restaurant operations consulting practice and shift with category, climate and purchasing structure: seafood kitchens with short cold chains land at the high end, dry-pantry kitchens at the low end.

Check them against your own invoices before you move a single price. The long menu scatters demand across too many lines, so no dish reaches the volume that earns a better purchase price; the short menu concentrates the sales mix and that volume becomes 5 to 12 points of supplier discount. On a long menu waste stays invisible because it dissolves into the monthly total cost; assign waste to the dish that produces it and the bottom 30% of the menu flips from contributing margin to destroying it. A short menu makes the standard recipe real: 22 spec sheets get audited every quarter, 96 never get audited at all and the cost per portion goes stale six months after somebody wrote it.

Where the difference actually sits?

Demand elasticity changes sign with fewer options: cut the alternatives and price sensitivity on the anchor dish drops, which lets you lift 6-8% without losing units.

Guest decision time falls from 6-9 minutes to 2-3 minutes, and across a two-turn service that frees tables and lifts revenue per seat without touching a single price. The hidden cost of a long menu is not food, it is hours: 112-140 weekly mise en place hours against 68-84, a gap worth 1,800 to 3,400 USD monthly in payroll for a mid-size operation.