Plate profitability: the food cost myth and the alternatives that really decide your menu

Why the green percentages on your menu lie?

Food cost percentage lies when you read it without the units-sold column, and that is exactly how an entire menu in green coexists with a cash register that never closes.

Take two real items from a restaurant in Chapinero: a ceviche costing 34% of its price, leaving $19,000 of contribution per portion and selling 180 times a month, against a pasta at 22% that leaves $8,400 and sells 95 times. The ceviche contributes $3,420,000 monthly; the pasta, $798,000. That is a 4.3x gap in favor of the dish with the WORSE percentage. Percentage is useful for negotiating with a supplier and for setting price, never for deciding which dish the house pushes. Payroll and rent get paid with accumulated pesos of contribution, not with pretty ratios on a spreadsheet.

When the original approach falls short?

Percentage-based food cost control falls short the moment two dishes with similar cost rotate very differently, and the giveaway is blunt:

if your three best-percentage dishes add up to less than 15% of monthly orders, you are optimizing what nobody orders. That is the symptom. It shows up too when average check rises and profit does not move, or when you shaved two points of cost across the whole menu and the register stayed flat. At Masterestaurant, Diego F. Parra insists on reading the POS units report across 90 days before touching a single recipe, because most cost adjustments made in Colombia land on dishes representing a tiny share of orders and move nothing at all.

Alternative 1: classic menu engineering, the Kasavana-Smith matrix

Crossing contribution margin against popularity in a four-quadrant matrix remains the highest return per peso invested, and it costs between 0 and 150 USD if you build it in a spreadsheet using standard recipes you already have. The real learning curve runs three weeks: one to standardize recipes, one to pull 90 days of units from the POS, one to read the quadrants without fooling yourself. Who is it for? The chef-owner of a single location with 20 to 45 menu references and recipes already written. Its limit is serious and worth stating: the matrix assumes historical popularity holds, and in a house publishing three Reels a week that popularity shifts month to month. Add limited-time offers and the distortion grows, since Technomic 2024 measured that 52% of consumers consider an LTO important when choosing a restaurant.

Alternative 2: contribution margin per station minute

When the kitchen has a physical bottleneck, pesos of contribution stop being the right unit and must be divided by the minutes each dish occupies the critical station. A steak leaving $23,000 while blocking the only griddle for seven minutes yields $3,285 per minute; egg arepas leaving $9,500 in two minutes yield $4,750. At peak hour, the second one wins. Software costs zero, yet the fieldwork does not: you have to time the pass of every reference across two full services, and the curve runs four to six weeks until the medians settle. This is aimed at kitchens with one griddle, one oven or one fryer, where the most profitable dish in pesos stalls four others behind it. If your kitchen has slack during peak, this method gives you nothing and takes your time.

Alternative 3: profitability weighted by content traction

Here sits the axis almost nobody calculates and where the money is today: weighting each dish's contribution against the traction that dish actually gets from the house's own content. The mechanic is measuring, per reference, how many orders arrived in the two weeks following a Reel or TikTok featuring it, then comparing against its baseline. The cost is production time, four to eight hours monthly, plus tagging discipline in the POS. It fits the chef-owner already publishing consistently with at least six months of history. The risk, stated without ornament, is that the most photogenic dish rarely matches the best margin, and pushing the pretty one without checking cost raises your orders while lowering your profit. Calculate contribution in pesos first, decide what gets filmed second.

What happens if you push the wrong dish for a quarter?

Suppose you promote your best-percentage dish on social, the one costing 21% and leaving $7,900 per portion, and you manage to double it from 95 to 190 monthly units.

You gain an extra $750,500 a month. Now suppose that push cannibalizes just 40 units of the ceviche that left $19,000: you lose $760,000. You closed the quarter with more plates served, more dishwashing, more kitchen wear and $28,500 less in the till. That is the whole trap, and it is reversible only if you measure. The underlying tension is real: the dish carrying your margin usually carries the highest cost, because guests pay for protein, not for flour. The way out is not picking a side but arranging the menu in two layers, the margin dishes you push and the low-cost ones that hold up average check without the spotlight.

The market figure that changes how you read your protein

The protein category behaves unevenly across fronts, and that weighs on menu decisions. Nation's Restaurant News reported in 2024 that chicken already accounts for 37% of quick-service food spending in the United States, two points above two years earlier; in parallel, SeafoodSource and Technomic measured a drop in seafood menu penetration across US restaurants that same 2024, and Supermarket Perimeter documented that 59% of seafood is eaten at home against 41% in restaurants. Translated to your menu: cheap protein gains volume, expensive protein loses menu presence yet keeps a high unit margin. Neither reading gets resolved with a food cost percentage. It gets resolved by looking at which mix of units and pesos of contribution delivers the profit you need to close the month.

When NOT to switch methods?

Sometimes sticking with food cost percentage and nothing else is the right call, and honesty helps here.

If your menu holds fewer than twelve references, if you have not standardized recipes and if you have been open less than six months, any of the three alternatives will cost you three to six weeks of work to hand back a number that is not yet stable. Standardize first. It also pays to leave everything untouched during high season, because changing prices or menu mix in December adds noise to the baseline and later you will not know what moved what. And if your operating margin already sits above 12% with the current menu, redesigning the method returns less than renegotiating two suppliers. Open the POS tomorrow, pull units and pesos of contribution per reference for the last 90 days, and sort that list from highest to lowest. That column tells you in ten minutes whether you need a new method.

Three honest alternatives once food cost runs out of road

ALTERNATIVE 1 — Classic menu engineering (Kasavana-Smith matrix). Cost: 0 to 150 USD if you build it in a spreadsheet; roughly three weeks to competence. Who it suits: 20 to 45 references, one location, an owner who already keeps standard recipes. Its limit: it assumes historical popularity holds, and in a house publishing three Reels a week popularity shifts month to month. ALTERNATIVE 2 — Contribution margin per station-minute. Cost: zero in software, but somebody must time the pass on every dish across two full services; four to six weeks to competence. Who it suits: kitchens with a physical bottleneck —one grill, one oven— where the highest-cash dish blocks four others from leaving. Its limit: pointless if the kitchen has slack. ALTERNATIVE 3 — Portion costing with elasticity testing and pricing psychology. Cost: 300 to 900 USD for an initial pricing study, or free if you A/B test on your own menu across two months; six to ten weeks to competence.

Three honest alternatives once food cost runs out of road — in practice

Who it suits: operations with an average check above 18 USD or brands mid-repositioning. Its limit: it needs volume, at minimum 900 tickets a month, for the test to mean anything. ON THE PRINTED MENU AND THE QR CODE — neither replaces the other. Keep the PHYSICAL menu always, because that is where you control service pace, menu narrative and suggestive selling; the QR is a complement for delivery, accessibility, price updates and analytics on what guests browse without ordering. The verdict is BOTH, each in its role, and operators who scrap the printed card to save on printing usually lose more in average check than they saved in paper. WHAT NO ALTERNATIVE FIXES — if the standard recipe does not exist or lives in a cook's head, every figure above is tidy fiction. Grammage first, model second.