Real cost of a combo or promotion: how we recovered 5.1 EBITDA points by fixing the capital leak with the Restaurant Model Canvas and the Standard Recipe Generator

The symptom: record sales, flat bank account

This family trattoria's flagship combo —pasta, starter and drink at USD 16.90— was selling like never before while EBITDA didn't budge an inch, because the traditional method costed it by adding up the plates' food cost «by eye» and subtracting a fixed percentage. At an estimated 30% food cost, on paper it «made money». In practice each unit pushed the month's Prime Cost up. The venue: 14 tables, 28 covers, 9 employees, average check of USD 21.40, 11 years of operation, dining room at 72% and in-house delivery at 28% (per the case file). The owner launched the combo to fight the corner chain on price. Full-service margin runs from 3% to 8% (WhippleWood CPAs, 2026): with that thin a cushion, a mis-costed combo doesn't shave profit, it erases it. «By eye» food cost lies because it ignores three costs that do happen: assembly waste, the dedicated labor minutes, and check cannibalization.

Why does «by eye» food cost lie?

A combo isn't a discount on existing plates: it's a new recipe with its own spec sheet.

Measuring the real combo at this trattoria we found an effective food cost of 34% —not 30%— from assembly portions and plating waste (per the case audit). The median server wage in the U.S. is USD 16.23/hour with tips (U.S. Bureau of Labor Statistics, May 2024): each combo consumed 2.3 extra minutes of floor and kitchen that nobody counted. And cannibalization hurt more: 41% of guests who used to order à la carte at a USD 24 check migrated to the USD 16.90 combo (per the case), giving away USD 7 of revenue per table without cutting a single fixed cost. The question that turned the case wasn't «what food cost does the combo have?» but «what contribution margin does it add per unit and at what volume?».

The right question: contribution margin per unit

The traditional method asks the first and hides the loss; the Masterestaurant method asks the second and exposes it. We rebuilt the combo's spec sheet as a new product: price USD 16.90, real input cost USD 5.75 (34%), assembly waste USD 0.60, delivery packaging USD 0.45 on 28% of orders. Gross contribution margin came out at USD 10.10 per unit, not the USD 11.83 that the «by eye» math suggested. At 640 combos a month, that USD 1.73-per-unit gap drained USD 1,107 in invisible monthly leakage. The 2026 cost pressure makes it worse: all food is up 3.2% and beef 7.5% this year (USDA ERS, 2026). The tool we applied was the Masterestaurant method's per-product costing sheet (inside herramientas_restaurantes.html), which treats each combo as a recipe with its own food cost, its waste, its labor minutes, and its channel allocation.

The tool: Masterestaurant per-product costing sheet

We loaded the three plates, the assembly waste and the real timed minutes in kitchen and floor; the system separated the combo's recurring OpEx (inputs, labor, packaging) from the promotion's CapEx (point-of-sale material, signage). Diego F. Parra insists on a principle I see fail over and over: price isn't set by looking at the competition, it's set backward from the target margin. Reference commercial rent in Los Angeles ran about USD 53 per sq ft/year in 2025 (Pepperlot, 2025): no combo can be given away when the fixed structure weighs that much. The result in 90 days was an EBITDA that finally moved, after three concrete method actions. First, the combo was raised to USD 18.50 and its composition changed: the pricier pasta was swapped for one at 28% food cost and the drink shifted to a higher-margin option —relevant because non-alcoholic beverages rise 5.7% in 2026 (USDA ERS, 2026)—.

Action and measurable result in 90 days

Second, the combo was limited to off-peak hours to curb dining-room cannibalization. Third, assembly waste was attacked with pre-built stations. Contribution margin per combo rose from USD 10.10 to USD 13.90 (per the case), check cannibalization fell from 41% to 22% and monthly Prime Cost dropped 2.4 points. Preventing waste pays: every USD 1 invested returns USD 7 of future benefit, a 600% ROI (ReFED). The transferable lesson is that a combo must be costed as a new product, not as a sum of discounted plates, and the first step depends on the size of your operation. Small independent (1 location): this week, time the real assembly minutes of your best-selling combo and add them to the cost; the «real» food cost is almost always 3-5 points above the estimate. Mid-sized (2-5 locations): this week, build the per-product costing sheet for each promotion and measure cannibalization by comparing average check before and after launch; fast-casual margin runs from 4% to 10% (WhippleWood CPAs, 2026), with no cushion to spare.

Transferable lessons

Multi-site group: this week, standardize the combo spec sheet across sites and audit food-cost variance by location; with the cattle herd at a 75-year low and beef up 7.5% (USDA ERS, 2026), a 2-point deviation across 6 sites is a leak of thousands a month. This case shouldn't be read as a guaranteed result: there are at least three contexts where I wouldn't expect the same improvement. First, in a business with dominant delivery (60%+ of the channel), the combo faces platform commissions and packaging that change the whole margin equation; here delivery was in-house and only 28%, a favorable scenario not everyone has. Second, in markets with very high price elasticity —where clientele migrates over USD 1— raising the combo from USD 16.90 to USD 18.50 can collapse volume instead of protecting margin; it worked because there were 11 years of brand and loyal clientele.

Limits of this case

Third, in operations without channel-level check data, isolating a cannibalization diagnosis is impossible. Remember that 26% of new restaurants close or change owners in the first year (Cornell University): costing discipline lowers the risk, it doesn't eliminate it. The traditional method asks «what's its food cost?»; the Masterestaurant method asks «what contribution margin does it add per unit sold, and at what volume?». The first question hides the loss; the second exposes it. The traditional method treats the combo as a sum of dishes; Masterestaurant treats it as a new product with its own card, its own waste and its own labor minutes. A combo isn't a discount: it's a different recipe. The traditional method sets price by looking at competitors; Masterestaurant sets it backward from the target margin and cost structure, distinguishing what is recurring OpEx from the combo and what is CapEx of the promotion (point-of-sale material, delivery packaging).