Menu engineering calculator · 100% free · no signup
Menu Engineering Matrix Calculator
Plot every item as a Star, Plowhorse, Puzzle, or Dog on contribution margin and popularity. The framework comes from Donald Smith at Cornell, 1982. Still the single best menu decision tool an operator owns (Cornell SHA). The math is the easy part. The discipline of running it every quarter is the entire game.
Classified results
Every item with its quadrant assignment, contribution margin, and dollar contribution. Sorted by contribution dollars, descending.
| Item | Food cost | Price | CM $ | CM % | Units | Contribution $ | Class |
|---|
Why this matters
Menu engineering is the second biggest move an operator makes.
Schedule discipline is the first. After that, nothing moves the P&L like a properly engineered menu. Same revenue, same labor, same rent — and two or three points of margin between a menu built on data and a menu built on what the chef likes.
Most operators run menu engineering by gut. They know the wings sell. They know the lamb does not. They have a vague sense the burger is profitable and the salad is not. What they do not have is a quadrant. Without a quadrant, you cannot tell a plowhorse from a star. Both sell. Only one pays rent. Confusing them is how a restaurant stays open while the bank account never grows.
The menu is the only marketing asset most restaurants have. Every customer reads it. Every server uses it. Every dish on it is either pulling weight or being subsidized by another dish.
Felix at Flaco's Cafe is a friend of mine. His parents came straight from Cuba and cooked the menu themselves. The food was incredible. Service was warm in a way you cannot fake. They gave plates away to staff every shift, comped regulars, sent food out for the AC tech and the priest. Nobody tracked the cost of any of it. Flaco's closed. Not because the food was wrong. Because nobody was running the math on the generosity. A matrix would have caught that the comp-heavy plates were dogs by contribution while sitting in the highest-traffic slots on the menu. Discipline is not being cheap. It is knowing what you are giving away, so you can choose to do it on purpose.
The framework comes from Donald Smith at Cornell, 1982. Forty-plus years later it is still the standard because the underlying logic has not moved. Two axes. Four quadrants. Specific actions per quadrant. The math is high school algebra. Running it every quarter is what separates the operators who renew leases from the ones who quietly do not.
The quadrant guide. What to do with each one.
Classification is the easy half. The hard half is acting on it. One specific play per quadrant. No mystery, no committee.
Protect the price. Hero on the menu.
- Top-third menu placement with a box, photo, or "chef pick" callout.
- Server training: lead with the star when describing the section.
- Hold price until you see meaningful resistance. Stars are inelastic at small bumps.
- Watch food cost monthly. A star with creeping food cost becomes a plowhorse fast.
Raise price or rebuild the recipe.
- Raise price 5–10%. Plowhorses are popular. Demand holds at small increases.
- Rebuild the recipe to cut food cost. Smaller portion, swap a premium ingredient that does not change the dish identity.
- Tighten plate spec. Walk the line. Plowhorses often suffer from over-portioning.
- Demote on the menu. Plowhorses do not need a hero spot. The volume is already there.
Better placement. Server training. Photograph it.
- Move to a hero menu spot. Top of the section, boxed callout, photo on the table tent.
- Train servers to recommend the puzzle when asked "what is good." High CM means every conversion is worth more.
- Photograph for socials. Puzzles are often photogenic items that customers do not know exist.
- Consider featuring as a daily special for two weeks to seed demand.
Remove. Replace with a star variant.
- Audit for strategic role first. Is this the only vegetarian, kids, or allergy-safe item? If yes, rework the recipe.
- If no strategic role and two consecutive quarters as a dog, remove it.
- Replace the menu slot with a star variant in the same category.
- Last resort before deletion: price drop with rebrand. New name, new description, repositioned.
Real scenarios
Three operators, three menu engineering wins
Same matrix, different fixes. The play depends on the quadrant the item lives in.
The filet that was popular and unprofitable
Steakhouse, filet at $45, $19 food cost, 280 units a month. CM 58%. Owner saw the unit count and assumed it was a star.
Inside the actual menu, the filet was a plowhorse. Other steaks ran 70% CM at similar price points. The filet was popular but bleeding 12 points of margin against the rest of the section. Three years of "raise prices later" added up to roughly $12K a year of left-on-the-table contribution. The owner had read the popularity column first and stopped reading.
Price moved $45 to $48. Cut from 8oz to a 6oz center cut, upgraded the demi-glace presentation to read as a premium plate. CM rose to 67%. Volume held at 270 units.
Monthly contribution moved from $7,280 to $7,830. Same item, same customer, same plate cost percentage. About $600/month from one matrix read. The math was the easy part. Pulling the trigger on a popular item was the hard part.
The puzzle that nobody had a slot for
Daily-special scallop dish, server-mentioned only. 90 units a month at 73% CM when offered. Never on the printed menu.
High CM, decent volume despite zero print real estate. Classic puzzle profile (Cornell 1982). The kitchen liked making it. FOH liked selling it. Customers who got it loved it. The question was whether it would scale with a permanent slot.
Permanent menu, top of the entree section, photo on the table tent. Server training: lead with the scallop on Tuesdays and Thursdays. Recipe locked at the same plate cost.
Volume jumped to 240 units a month. CM held at 73%. Top contribution-dollar item on the menu within a quarter. The thesis confirmed itself: the dish was not unpopular. It was unpromoted.
The crab cake nobody wanted to remove
Crab cake appetizer, $24 menu, $11 food cost, 38 units a month. CM 54%. Held a top-third appetizer slot for two years on sentiment from the original menu.
Below average on both axes. Two consecutive quarters in the dog quadrant. The slot was working at about $494/month of contribution while the rest of the apps section averaged closer to $1,800.
Removed. Replaced with a shrimp cocktail at $18 menu, $5 food cost. Same prominent slot. Same category, same customer expectation.
Shrimp cocktail: 180 units a month, 72% CM, immediate star. Same menu real estate now contributing $2,340 a month vs the old $494. Replacing one dog with one star paid for an entire prep cook shift. The sentimental hold cost about $22K over the two years it held the spot.
FAQ
Common questions
01 What is menu engineering?
Plot every menu item on two axes: contribution margin (menu price minus food cost in dollars) and popularity (units sold). The result is a 2×2 matrix that tells you which items pay rent and which ones are eating it. The framework comes from Donald Smith at Cornell in 1982 (Cornell SHA). Forty-plus years later it is still the standard, because the math has not changed.
02 Stars, plowhorses, puzzles, dogs — what are they?
Star: high CM, high popularity. Hero on the menu, protect price, train upsells. Plowhorse: low CM, high popularity. Volume drivers that bleed profit. Reprice or rebuild the recipe. Puzzle: high CM, low popularity. Money makers nobody orders. Reposition, photograph, train servers to suggest. Dog: low CM, low popularity. Remove or rework.
03 How often should I run it?
Quarterly minimum. Monthly during a launch, after a price change, or when food costs spike. The 90-day cadence aligns with the seasonal product cycle most distributors run, and gives you four data points a year to tell trend from noise.
04 What is a healthy contribution margin?
Full-service: 65–75% CM (food cost 25–35% of menu price, per Toast). QSR/fast-casual 70–80%. Fine dining 70%+ floor. Beverages run higher: beer 75–80%, wine 65–75%, cocktails 80–85%. The matrix uses your menu average as the threshold, not an industry number — comparison stays apples to apples inside your own concept.
05 How do I fix a plowhorse?
Three levers in order of speed. Raise price 5–10% on the item alone. Plowhorses are popular by definition, demand is inelastic at small bumps. Rebuild the recipe: smaller portion, swap a premium ingredient nobody tastes, tighten plate spec. Reposition to a less prominent spot. Plowhorses get demoted, not deleted.
06 Should I remove dogs immediately?
Not always. Some dogs do strategic work. The vegetarian item nobody orders may be why a four-top chose you. The kids plate may be the reason a family came in at all. Audit each dog for strategic role before cutting. Two consecutive quarters in the dog quadrant with no strategic purpose, then remove. Replace with a star variant in the same category.
07 What about beverages?
Run them in a separate matrix. Beverage CM runs 15–25 points higher than food. Mixing them in pulls the food average up and makes real food stars look like dogs. Two parallel matrices, same logic, different averages.
08 I have 200 items. Where do I start?
Two approaches. Run the matrix by category — apps, entrees, desserts separately. Within-category averages are more useful than aggregate. Then focus the analysis on the top 80% of revenue. Pareto holds in restaurants: roughly 20% of items drive 80% of contribution. Engineering the long tail is rarely worth the work.
09 Volume or margin — which matters more?
Neither alone. The matrix exists because operators kept making the wrong call from one variable. A 280-unit/month plowhorse at 58% CM contributes more dollars than a 40-unit puzzle at 72% CM. But the puzzle has higher margin per cover and is under-promoted. Most operators undervalue puzzles because they read the popularity column first. The matrix forces both reads at once.
10 Does this calculator save my data?
No. Every row, every input, every classification runs in your browser and disappears the moment you close the tab. No signup, no email, no account, no analytics on your numbers.
Cited sources