Menu by taste and tradition vs Masterestaurant menu engineering

The taste-driven menu: when the chef decides alone

A menu designed by the chef to showcase technique is not the same as a menu designed to generate margin. In 25 years of consulting across more than 8,400 restaurants in 43 countries, Diego F. Parra has found the same pattern time and again: chefs build their menus around what they know how to cook, what they love, or what they believe guests value. The result is an inflated offer — 60, 80, even 120 items — where 70% of sales come from just 20% of the dishes. The best-selling items are rarely the highest-margin ones. That means working twice as hard to earn half as much. A taste-driven menu is not a failure of intention; it is a failure of method. Without contribution margin and popularity metrics, the restaurant owner operates blind with their most powerful asset. Menu engineering classifies every dish into four categories using two variables: contribution margin (selling price minus food cost) and popularity (unit volume sold over a period).

What menu engineering actually measures?

A dish with high margin and high sales is a star; high margin and low sales, a question mark; low margin and high sales, a plow horse or margin trap;

low margin and low sales, a dog. That decision grid — popularized by Michael Kasavana and Donald Smith in 1982 — tells the operator exactly what to promote, what to adjust in price or food cost, what to redesign, and what to eliminate. The historical problem was execution: a manual analysis of 80 dishes takes between 4 and 8 hours per period. In 2026, a system connected to the point of sale updates every classification in real time, with no spreadsheets and no data-entry errors. Eliminating dog dishes does not reduce sales — it concentrates effort. In analyses conducted using the Masterestaurant Method, Diego F. Parra has documented cases where removing 6 to 10 dog dishes reduced raw material waste by 18% to 27% and increased total gross margin by 3 to 5 percentage points in the first quarter after the adjustment.

The hidden cost of dog dishes

The mechanism is direct: fewer active SKUs means more concentrated purchasing, lower spoilage from variety, faster inventory turnover, and simpler kitchen briefs. A restaurant with 90 menu items can have up to 340 active ingredients; one with 45 well-selected dishes operates with fewer than 160, while retaining 85% of its sales volume. Operational complexity is not an indicator of quality — it is a source of silent revenue leakage. A dish's food cost is calculated by dividing ingredient cost by selling price; the operating threshold at Masterestaurant is ≤32% per dish as the maximum acceptable, not as the target. Most chefs who design menus by taste do not know the real food cost of each item: they work with global averages that hide the dispersion underneath. It is common to find dishes at 18% food cost living alongside dishes at 47% on the same menu, with the reported average at 31% — which sounds acceptable until the income statement does not add up.

Food cost per dish: the metric most often falsified

Diego F. Parra has calculated that a restaurant with an average ticket of 280 MXN and 200 covers per day loses between 1,400 and 2,800 MXN daily when two or three plow-horse dishes concentrate 30% of sales volume. What is not measured leaks out. Artificial intelligence turns menu engineering from a periodic exercise into a continuous monitoring system. An AI module connected to the point of sale can recalculate every dish's classification — star, question mark, plow horse, dog — every 24 to 48 hours, with no manual intervention. If a dish that spent three months as a question mark suddenly starts selling strongly in a given week, the system flags it as a star candidate and alerts the operator to promote it or adjust its menu position before the momentum fades. In pilots conducted with Masterestaurant program restaurants in Mexico and Colombia, this feedback cycle reduced the response time to menu opportunities from 30 days to fewer than 72 hours.

AI applied: from monthly analysis to real-time traffic lights

Decision speed is now a measurable competitive advantage. The Masterestaurant Method applies menu engineering in four concrete steps. First, data collection: real food cost per item plus sales volume over the last 90 days. Second, four-quadrant classification using the restaurant's own thresholds — not industry averages, because cost structure varies by concept, location, and ticket size. Third, an action plan by category: stars are featured prominently in the menu and in marketing; question marks are tested with a price or presentation adjustment over 30 days; plow horses are optimized on food cost before being removed; dogs exit in the next print cycle. Fourth, results validation at 60 days with a new classification run. This iterative cycle — data, decision, action, measurement — is what separates a redesigned menu from one that was simply reprinted with new aesthetics. Menu psychology delivers an 8% to 15% margin increase without changing a single price, according to studies from the Cornell University School of Hotel Administration.

Menu position and visual language: the margin that lives outside food cost

The diner's eye follows predictable paths: on two-page menus, initial focus lands on the upper-right quadrant; on single-page menus, at the center. Masterestaurant incorporates these principles into every redesign: stars occupy anchor positions; question marks with potential are placed next to high-price items to anchor value perception; plow horses are moved to the back or removed from the visual menu even if still available. Removing the currency symbol reduces price sensitivity by up to 12% in laboratory studies. Every layout decision is a margin lever. What does not show up in the food cost does show up in the register at the end of the day. A restaurant that applies menu engineering through the Masterestaurant Method with AI monitoring can expect measurable results within 90 days: a reduction in average food cost of 2 to 4 percentage points, a decrease in raw material waste of 15% to 25%, and an increase in average ticket of 6% to 12% when stars are well positioned.

Expected results: what changes in the first 90 days

Diego F. Parra's project database shows that restaurants with menus of more than 70 dishes that reduced to 35–45 well-classified items saw operating profit increases of 9% to 18% in the first semester, without raising prices. The menu is not the document the chef likes to write; it is the most powerful financial instrument a restaurant owns. Treating it as such is the difference between surviving and scaling. The difference between a taste-based menu and an engineered menu isn't visible in the physical menu — it's visible in the month-end margin. I've analyzed restaurants where eliminating 8 dog dishes reduced raw material waste by 22% and increased total margin by 4 percentage points. They weren't selling less — they were selling differently, with more focus. That's what menu engineering does when applied correctly. AI revolutionizes the process by making it continuous rather than periodic.

Why menu engineering multiplies your margin without raising sales?

A manual monthly menu engineering analysis takes hours. An AI system connected to your POS can update each dish's classification in real time:

if a puzzle dish started selling strongly this week, it's already a star candidate — and you should immediately redirect marketing toward it. That decision speed is the difference between capturing a trend and burying it.