Which menu items to cut for profitability: myth vs reality

Chapter 1 — The ratio doesn't pay payroll: contribution margin in dollars does

Cut with money, not with percentage. A dish carrying 34% food cost that leaves eleven dollars of margin per unit and turns forty times a day brings in 440 dollars daily; three virtuous 24% dishes leaving four dollars each and turning six times add up to 72, and that 368-dollar daily gap decides whether the month closes with cash or with excuses. For years I defended 32% as if it were a target, and I was wrong there: with food and labor costs running 35% above 2019 levels according to the National Restaurant Association (2024), the ratio stopped being the control variable and became a traffic light, worth a glance while the real decision comes from multiplying unit margin by units sold. Sort the menu by that column, highest to lowest, and look at the bottom quartile without affection. Before pulling a dish, test whether the market tolerates it more expensive.

Chapter 2 — Elasticity decides whether the dish dies or simply gets repriced

Inelastic demand —house signature, no close substitute within three blocks— absorbs a 6% adjustment without losing units, and in a market where away-from-home food prices climbed 4,1% in 2024 and 3,8% in 2025 according to the USDA Economic Research Service, that move doesn't even read as gouging: it reads as average. In Colombia the sector moved menu prices 9,8% since February 2025 to sustain 98.000 jobs, per ACODRES. The test runs fourteen days with the new price and a daily unit count. If units drop under 4%, you just rescued a dish that was heading for the cemetery. If they fall 15%, the dish was elastic and its place on the menu rested on price, not on flavor. A dish nobody photographs, nobody searches and nobody mentions in reviews isn't identity, it's operating expense in disguise.

Chapter 3 — A dish invisible in the local digital engine is OpEx with a pretty name

Your Google Business Profile listing, your ranking inside Rappi or Uber Eats and the answer an AI assistant hands whoever types «where to eat near me» now weigh as much as the corner window, and reputation there converts straight into cash: each additional star in the rating moves between 5% and 9% of revenue, according to Michael Luca's work at Harvard Business School. So cross two lists: the ten dishes with the lowest absolute margin and the ten that never show up in a customer photo or a comment. Whatever sits in both columns doesn't get debated, it gets retired, and the line space it frees goes to a dish that actually exists outside the kitchen. Operation size completely changes what pruning means. Under 500 thousand dollars a year the cut is survival: going from 78 to 40 SKUs frees immobilized working capital and lowers waste, and food cost stays the most honest traffic light there because no analyst is watching anything else.

Chapter 4 — By revenue band: the same cut, five different effects

Between 500 thousand and 1 million the lever is purchasing: consolidating 210 inputs into 120 improves volume pricing with two suppliers instead of five. Above 1 million the gain sits on the line, with shorter ticket times and faster table turns. Above 5 million the cut pays for itself in shift scheduling, where AI-assisted assignment reduces labor cost by 8% to 12% with forecast accuracy above 90%, according to TimeForge (2025). Above 10 million, every SKU removed replicates that saving across each unit in the chain. Above 5 million dollars —the celebrity-chef house, the large-format themed venue, 300 seats— a wide menu is part of the show and doesn't get trimmed with the same scissors. The cost structure there is different: cellar, mise en place across twelve stations, a specialized team whose departure costs 150% of salary in replacement according to StaffedUp (2025), and a bar carrying the margin of the business, since alcohol is named the highest-margin category by 46% of operators surveyed by Technomic for Nation's Restaurant News (2024).

Chapter 5 — High end: when the long menu is the product and you still must prune

Pruning in this band doesn't attack the number of visible dishes, it attacks exclusive inputs: the ingredient that enters one preparation only, arrives by the case and gets half thrown out. Reduce single-use inputs, keep the theater, and margin appears without the guest noticing one line fewer. Do not charge payroll, rent or utilities to dish cost: that account belongs to break-even, and confusing them makes you kill profitable dishes. At Masterestaurant the first thing we review, with Diego F. Parra leading the costing audit, is whether the P&L pushes fixed expenses onto each recipe, because a dish with a real 30% food cost can show up at 52% once rent gets spread over it, and that ghost has killed dishes that carried the weekend register. Dish cost is inputs plus attributable waste plus packaging, period. Fixed costs get covered by the sum of the month's contribution margins.

Chapter 6 — The mistake of loading rent and payroll onto the dish before deciding

Whoever blends both layers ends up with a short menu and no anchor, which is a worse business than a long, disorderly one. Picture the worst case so you know what you're risking. You pull 23 of 78 SKUs choosing by food cost, and two high-margin dishes go with them, dishes that turned slowly but brought a full table: cross-selling collapses, average ticket drops a dollar and a half, and the purchasing saving —real, measurable, maybe 3% of cost of goods— gets eaten by falling traffic within six weeks. That's the risk of pruning with the wrong ratio, and it's why sequence matters more than nerve: contribution margin in dollars times turns first, then the fourteen-day elasticity test, then the digital review, and only then the cut. Discounting also works as a rescue tool before it works as a farewell, like the 15% saving McDonald's states for its Extra Value Meal combos (2025), built around dishes that already carry margin.

Chapter 7 — The pruning calendar: four cuts a year, no emotional decisions

Set a date or the menu will fatten again. Every quarter pull the report of contribution margin per unit multiplied by units sold, flag the bottom quartile, and from that group remove three SKUs and intervene two more with a new price or a shorter recipe. Twelve exits a year keep a menu between 32 and 45 SKUs, the range where the kitchen dispatches without a bottleneck and purchasing keeps negotiating power. The first quarter always hurts because dishes with history walk out; by the fourth it's administrative routine. Then add an entry rule: a new dish comes in only if another leaves the same day, with costing signed before printing. That single lock —one in, one out— keeps you from waking up four years later with 78 SKUs, 210 inputs and a storeroom where dead inventory passes for variety. Food cost is a ratio, contribution margin is money.

Chapter 8 — Six differences that move EBITDA

A dish at 34% food cost leaving 11 dollars per unit and turning 40 times a day delivers 440 dollars of daily margin; three dishes at 24% leaving 4 dollars and turning six times deliver 72. The arithmetic leaves no room for argument, and still most menus get pruned with the ratio. I got this wrong for years: I defended 32% as a target when, according to the National Restaurant Association (2024), with costs 35% above 2019 the ratio is no longer the control variable, it is barely the warning light. Demand elasticity decides whether a dish gets deleted or repriced. An inelastic dish, the house signature with no close substitute in the neighborhood, absorbs a 6% move without losing units, and that move can turn a deletion candidate into a profitable item. An elastic one sheds 20% of units on the same move. Testing before cutting costs three weeks of data and prevents destroying demand.

Chapter 9 — Six differences that move EBITDA — in practice

Shared ingredient footprint reorders the whole list. When two dishes share eight of their twelve ingredients, deleting one shrinks neither inventory nor purchasing leverage, it only narrows choice for the guest. A dish with three exclusive ingredients that spoil in four days, meanwhile, is a silent waste engine even when its percentage margin looks respectable on the per-portion costing sheet. Local visibility is an asset even though no balance sheet records it. According to Harvard Business School (Michael Luca), each additional star in review rating moves between 5% and 9% of revenue; if three of your five-star reviews name a dish, that dish is buying reputation. Deleting it over middling margin means selling a digital asset to save pennies of portion cost. Channel changes the verdict dish by dish. An item that travels badly in delivery, fried food going soggy, plating falling apart, can stay excellent in the dining room and ruinous on Rappi or Uber Eats, where commission eats 15 to 30 points of the price.

Chapter 10 — Six differences that move EBITDA — key points

The right call is rarely deletion: pull it from the app menus, keep it on the printed menu, and price by channel. The entry threshold matters more than the cut itself. Without a written rule, no dish enters unless it clears the menu's median contribution margin and shares at least three ingredients, the menu re-inflates within two seasons. The cut is an event; the threshold is the system, and only the system holds the result.