Engineer your menu
One row per menu item from the same period (a typical month works). Sample menu loaded; overwrite it with yours.
| Menu item | Menu price ($) | Portion cost ($) | Units sold | Margin | Class | Remove |
|---|
Email yourself this analysis
The full classification with benchmarks, in your inbox for the next menu meeting.
What menu engineering actually is
Menu engineering scores every item on two axes: how much money it makes you each time it sells (contribution margin, meaning price minus portion cost) and how often it sells (popularity, meaning its share of total items sold). Cross the two and every item lands in one of four quadrants. The names come from the classic Kasavana and Smith framework, and the point is not the labels; it is that each quadrant has a different correct move.
STAR
High margin, high popularity. Protect the recipe, keep quality consistent, give it prime menu real estate, and do not discount it.
PLOWHORSE
Popular but thin margin. Raise the price in small steps, re-engineer the plate cost with yields and portioning, or pair it with high-margin sides.
PUZZLE
Great margin, few takers. Rename it, reposition it on the menu, have servers speak to it, or move it into a feature slot before touching the price.
DOG
Low margin, low sales. Rework it into something the kitchen already preps, or cut it and free the inventory and menu space.
How the math works, with the sample menu
The margin benchmark is your menu's weighted average margin: total margin dollars divided by total items sold. The popularity benchmark follows the 70 percent rule: with 7 items, average share is 14.3 percent, so the bar sits at 10 percent. In the sample above, the fish tacos clear both bars and land as the Star. The burger sells hard but makes $8.86 against a $9.91 benchmark, a classic Plowhorse. Here is the subtle part: because the burger's own volume drags the benchmark up with any price move, $13.25 is not enough; it takes $13.50 before its margin clears the bar it just raised. Try it in the tool above. That second-order math is exactly what the matrix does for you, and finding the one-move items like that is the whole discipline.
Two warnings from years of doing this. First, the matrix is only as accurate as your portion costs; engineer a menu on stale costs and you will promote the wrong items. Cost your recipes properly first, with the recipe cost calculator or, at scale, with recipe costing software that reprices as invoices land. Second, re-run the matrix after every meaningful vendor price change; a Star with a 20 percent protein increase can turn into a Plowhorse in a week, and you want to see it happen the same week, not at month end. And when a quadrant calls for a price move, make it with a method: how to price a menu walks through all three.
Common questions
What is menu engineering?
Analyzing every menu item by contribution margin and popularity, then repricing, repositioning, reworking or removing items based on their quadrant: Star, Plowhorse, Puzzle or Dog.
How do I find my portion costs?
Cost each recipe from pack prices with the free recipe cost calculator, or let the platform keep them current automatically from your invoices.
What counts as popular?
The classic rule: 70 percent of an equal share. Ten items means an equal share is 10 percent, so anything above 7 percent of items sold counts as popular.
How often should I re-engineer the menu?
Quarterly at minimum, and any week your vendor prices move meaningfully. If your costs update themselves, the matrix is always current.
Is this tool free?
Yes. No signup, runs in your browser, nothing stored unless you email yourself the results.
Engineering a menu on live costs beats engineering it once
Culvana keeps every portion cost current by reading your vendor invoices the day they arrive, so this matrix stays true all year. Book a demo and we'll run your real menu through it live, or see the whole picture in the savings calculator. $199 a location, everything included.