Two dishes on your menu can sell the exact same number of orders every week and still be worth completely different amounts to your business. One quietly pads your margin every time it sells. The other is bleeding you dry one plate at a time — and you'd never know it just by looking at the sales report, because sales volume alone tells you nothing about profit.
That blind spot is exactly what menu engineering was built to fix. Developed in the early 1980s by Michael Kasavana and Donald Smith at Michigan State University, the technique classifies every item on your menu into one of four categories based on two numbers: how much profit it generates per sale, and how often people order it. Get those two numbers for every dish, plot them on a simple grid, and your entire menu turns into a decision-making tool instead of a static list of food.
Why Popularity Isn't Profitability
Most restaurant owners already track which dishes sell best. Fewer track which dishes actually make money — and the two lists rarely match.
Here's a common trap: a restaurant's best-selling burger might have a 32% food cost, which sounds fine against the 28–35% industry benchmark. But if the burger sells for $14 and costs $4.48 in ingredients, it only contributes $9.52 to covering labor, rent, and profit. Meanwhile, a slow-moving short rib special priced at $28 with a 38% food cost — technically "worse" by the percentage — contributes $17.36 per plate. Sell 200 burgers and 40 short ribs in a week, and the burger generates $1,904 in contribution margin while the short rib generates $694. The burger wins on total dollars even though its percentage looks weaker, which is exactly why menu engineering never looks at food cost percentage alone.
This is the core insight: food cost percentage tells you about a single dish, but contribution margin tells you about your bank account.
The Two Numbers That Matter
Before you can build the matrix, you need two figures for every item on your menu.
1. Contribution margin. This is simply the menu price minus the direct food cost of that dish:
Contribution Margin = Menu Price − Food Cost Per Plate
A $16 pasta dish that costs $5.20 in ingredients has a contribution margin of $10.80. This is the number that actually pays your overhead and lands in your pocket — not the food cost percentage.
2. Popularity (sales mix). This measures how often an item sells relative to everything else on the menu, usually expressed as a percentage of total items sold in a category over a set period (a week or a month is typical):
Popularity % = (Number Sold ÷ Total Items Sold in Category) × 100
A common rule of thumb: an item is "popular" if it sells at or above 70% of what you'd expect from an even split across your menu. If you have 10 entrees, an even split would put each at 10% of sales — so anything above roughly 7% clears the popularity bar.
To get the food cost numbers right in the first place, you need standardized recipes with exact quantities, current supplier pricing, and a habit of recalculating whenever a key ingredient's price moves. Restaurants that only revisit food costs once a quarter are usually working from stale numbers — with today's volatility in commodity pricing, weekly or biweekly recalculation catches problems before they eat a month of margin.
The Four Quadrants: Stars, Plowhorses, Puzzles, and Dogs
Once you have contribution margin and popularity for every item, plot them on a two-by-two grid. High contribution margin on one axis, high popularity on the other. Every dish lands in one of four quadrants, and each quadrant has a specific playbook.
Stars: High Profit, High Popularity
These are your best dishes by every measure — customers love them, and they make good money. Protect these at all costs. Don't mess with the recipe, don't quietly swap in a cheaper ingredient to save a few cents, and don't bury them on the menu. Feature Stars prominently, use them in photography and marketing, and treat any proposed change to them with real caution, since they're doing the heaviest lifting for your bottom line.
Plowhorses: High Popularity, Low Profit
Plowhorses are the dishes everyone orders that don't actually make you much money — often loss-leaders like a classic cheeseburger, a simple pasta, or a fried appetizer that anchors the menu. You can't remove them without upsetting regulars, but you can quietly improve their economics: a modest price increase (small enough that customers barely notice), tighter portion control, a cheaper side swap, or renegotiating with your supplier for that specific ingredient. Because these items sell in volume, even a $0.50–$1.00 price nudge compounds fast across hundreds of orders a week.
Puzzles: High Profit, Low Popularity
Puzzles are hidden gems — dishes that make great money whenever someone orders them, but not enough people do. The fix is almost never to cut the price (that would only shrink the margin that makes them valuable in the first place). Instead, work on visibility and framing: reposition the item higher on the menu (top-right of a page and the first or last items in a list get read most), add a more evocative description, have servers verbally recommend it, or feature it as a limited-time special to build awareness. If repositioning doesn't move the needle after a few weeks, the item may really belong in the Dogs category.
Dogs: Low Profit, Low Popularity
Dogs cost you money in two ways: low margin per sale, and the ongoing expense of carrying inventory, prep time, and menu real estate for something few people order. Unless an item serves a strategic purpose — a vegan option that keeps a whole dining party at your table, or a dish with sentimental or signature value — Dogs are candidates for removal. Cutting them also simplifies your kitchen, reduces waste from ingredients that only exist for one recipe, and frees up menu space to feature a Star or reposition a Puzzle.
Building Your First Menu Matrix
You don't need specialized software to run this analysis the first time — a spreadsheet works fine.
- List every item in a menu category (entrees, appetizers, and desserts should generally be analyzed separately, since comparing a $4 appetizer to a $30 entree on the same grid produces meaningless results).
- Pull food cost per plate from your recipe costing — ingredient quantities multiplied by current supplier prices.
- Calculate contribution margin for each item (price minus food cost).
- Pull sales counts for each item over the same period, ideally 4–8 weeks to smooth out day-to-day noise.
- Calculate popularity percentage for each item within its category.
- Find your averages: the average contribution margin across the category, and the even-split popularity threshold (roughly 70% of 1 ÷ number of items).
- Plot each item against those two averages — above or below on each axis determines the quadrant.
- Assign an action to each item based on its quadrant, and set a date to revisit the whole exercise — quarterly at minimum, monthly if your ingredient costs are volatile.
The single biggest mistake operators make here is doing this exercise once and never repeating it. Ingredient prices move, customer preferences shift with the season, and a Star from last year can quietly slide into Plowhorse territory without anyone noticing until margins tighten.
Beyond the Matrix: Menu Psychology That Supports the Data
Menu engineering isn't only about the math — where and how an item is presented changes what people order, independent of price. A few well-documented patterns:
- Eyes go to the top-right and the first/last items in a list first. That's prime real estate for Stars and repositioned Puzzles.
- Removing dollar signs and ".00" from prices ("14" instead of "$14.00") reduces the psychological friction of spending, a widely tested effect in menu design research.
- Anchoring with a high-priced item (even one that rarely sells) makes the items below it look more reasonably priced by comparison.
- Descriptive language sells. Menu items with specific, sensory descriptions ("slow-braised" instead of just "beef") consistently outsell generically described versions of the same dish.
None of this replaces the underlying cost math — it amplifies it. A Puzzle item repositioned to a high-visibility spot with a better description is the fastest lever most restaurants have to improve margin without changing a single ingredient.
Keep the Books That Make the Math Possible
Menu engineering only works if your underlying numbers are trustworthy — accurate ingredient costs, clean sales-by-item data, and a real picture of what each dish actually costs to produce. That discipline starts with solid bookkeeping: separating COGS by category, reconciling supplier invoices against recipe costs, and keeping records that you can actually query when it's time to run this analysis again next quarter.
Beancount.io offers plain-text accounting that's transparent, version-controlled, and easy to query for exactly this kind of item-level analysis — no black boxes, no vendor lock-in, and no waiting on a report export to know where your margin is going. Get started for free and see why restaurant operators and finance-minded owners are switching to plain-text accounting for the visibility it gives them into decisions like these.