How menu engineering works
Menu engineering sorts every dish on a menu into one of four boxes using two numbers you already have: how many of it sold, and how many dollars it leaves behind after food cost. This page explains where the two thresholds come from, why one of them is weighted, and works a small menu all the way through.
Open the free menu engineering calculator
The two axes
The method is Kasavana and Smith’s, published at Michigan State University in 1982. It has outlasted a lot of fancier ideas because both of its inputs come off a sales report and a recipe card, and neither requires a consultant to interpret.
Popularity is a dish’s share of covers. If every item on a ten-item menu sold equally, each would take 10%. The convention is to call an item popular once it clears 70% of that equal share — so 7% on a ten-item menu, 5.6% on a twelve-item one. The 70% is a convention rather than a law of nature, which is why a calculator worth using prints the threshold it applied instead of asking you to take the sorting on trust.
Contribution margin is menu price minus plate food cost, in dollars. Not as a percentage. This is the axis operators most often get wrong, and the next section is about why.
Why not food cost percentage?
Because food cost percentage and contribution margin disagree constantly, and only one of them pays rent.
| Dish | Price | Plate cost | Food cost % | Contributes |
|---|---|---|---|---|
| Ribeye | $34.00 | $13.60 | 40% | $20.40 |
| Pasta | $9.00 | $1.80 | 20% | $7.20 |
The pasta wins by twenty points on the number the industry quotes, and loses by $13.20 on the number that reaches the bank. A kitchen that ranks its menu by food cost percentage will promote the pasta, sell more of it, sell fewer steaks, and end the month with a better food cost percentage and less money. This is the single most common way menu analysis backfires.
Why the margin bar is weighted
The bar every dish is measured against is the menu’s weighted average contribution margin: total contribution margin divided by total covers. Not the average of the per-dish margins.
The difference is not academic. An unweighted average lets a $40 plate that sold twice pull the bar up as hard as the $14 plate that sold four hundred times. The bar is supposed to describe what the menu actually earns per cover, and only the weighted version does that. A menu with one rarely-ordered expensive special can have almost every other dish misclassified by this alone.
The four boxes
- Star — sells well, contributes above average.
- Plowhorse — sells well, contributes below average. Volume is doing the work margin should be. Because a plowhorse lands on more plates than anything else, it is where a given price move is worth the most in absolute dollars.
- Puzzle — contributes above average, does not sell. The margin is already there; what is missing is covers.
- Dog — below the bar on both. Still takes prep, storage, and a line on the menu.
The names are the industry’s and they are blunter than they need to be. A “dog” can be the dish the owner is proudest of, the one a regular comes in for, or the only vegetarian main on the page. The matrix measures two things and is silent about every other reason a dish earns its place.
A worked menu
Four dishes over four weeks, 1,000 covers. Total sales $16,000, total food cost $5,100, so total contribution margin is $10,900 and the weighted average is $10.90 a cover. Four items, so the equal share is 25% and the popularity bar is 25% × 0.7 = 17.5%.
| Dish | Sold | Menu mix | Price | Plate cost | CM | Class |
|---|---|---|---|---|---|---|
| Smash burger | 500 | 50.0% | $14.00 | $5.00 | $9.00 | Plowhorse |
| Ribeye | 100 | 10.0% | $34.00 | $14.00 | $20.00 | Puzzle |
| House salad | 300 | 30.0% | $16.00 | $3.00 | $13.00 | Star |
| Soup of the day | 100 | 10.0% | $8.00 | $3.00 | $5.00 | Dog |
The burger clears the popularity bar easily and contributes $1.90 a plate below the menu average, so it is a plowhorse. The ribeye contributes nearly twice the average and sells at 10%, below the 17.5% bar, so it is a puzzle. Note that the ribeye also has the worst food cost percentage on the menu at 41%, and is the best contributor on it by a wide margin.
What the matrix does not tell you
It does not tell you what to charge, what to cut, or how to lay out a menu. Those depend on the room, the clientele, what the kitchen can execute on a Saturday, what your suppliers will do to a price next quarter, and a dozen things that are not in four columns of numbers. The matrix gives you a defensible starting point and takes the guesswork out of which dishes are worth arguing about.
Common questions
What is a plowhorse on a menu?
A plowhorse is a dish that sells well but contributes below the menu average in dollars. Volume is doing the work that margin should be. Because a plowhorse lands on more plates than anything else on the menu, a given price move on one is worth more in absolute dollars than the same move anywhere else — which is also why it carries the most risk if the price change moves the mix.
How many dishes do you need for menu engineering to work?
At least two, because both thresholds are relative: an item is measured against the rest of the menu, not against an outside standard. In practice the matrix gets more useful as the menu gets longer, because the popularity bar falls as items are added and the weighted margin bar starts to describe a real mix rather than two or three dishes.
Does menu engineering work for a cafe, food truck or bar?
Yes, wherever you can put a plate cost, a price and a unit count against each item. The method has nothing restaurant-specific in it — it is a portfolio read. For a bar the equivalent per-drink arithmetic, including what an over-pour costs across a year, is a separate calculator on this site.
Should you just remove the dogs from a menu?
That is not a question the matrix answers. It measures two things — covers and contribution margin — and is silent about every other reason a dish earns its place: a regular who comes in for it, the only vegetarian main on the page, an ingredient shared with three other dishes, or the thing that keeps a supplier relationship alive. The matrix tells you which dishes are worth arguing about, not how the argument should end.