Food cost percentage and plate cost tell you whether a single dish is priced correctly. They don't tell you what to actually do with a 40-item menu once every dish has a number attached to it. Menu engineering answers that second question — it plots every item by how much profit it makes and how often it sells, then sorts the whole menu into four categories that each call for a different move: keep, reprice, reposition, or cut.
This tool runs the classic menu engineering matrix on whatever items you enter — no spreadsheet, no signup, no POS export required to get a first read.
How the matrix is calculated
Two numbers decide where a dish lands. Contribution margin is menu price minus item cost — the dollars each sale actually contributes toward labor, rent, and profit once the ingredient cost is covered. Popularityis that item's share of total units sold across the whole menu for the period you're analyzing.
An item counts as high-margin if its contribution margin is at or above the sales-weighted average across every item on the menu — not a simple average, but one weighted by how many of each dish actually sold, so a high-cost item that barely sells doesn't skew the line. An item counts as high-popularity if its share of units sold is at or above an equal share of the menu (for a 10-item menu, that's 10% or more of total units sold). Cross those two lines and every item falls into one of four quadrants.
The four quadrants
| Quadrant | Profile | What to do |
|---|---|---|
| Star | High profit, high popularity | Protect it — keep the recipe, price, and placement exactly as is. |
| Plowhorse | High popularity, low profit | Reprice or re-cost carefully; it drives real volume even at a thin margin. |
| Puzzle | High profit, low popularity | Reposition on the menu, rename, or re-describe it to drive more orders. |
| Dog | Low profit, low popularity | Candidate to rework or cut from the menu entirely. |
Puzzles and dogs get confused with each other more often than any other pair — both sell slowly, so both feel like weak performers. The difference is what happens if you fix the popularity problem: a puzzle is already profitable per sale and just needs more orders, while a dog needs a cost or price fix before it's even worth promoting harder.
Worked example
Take five items from a casual-dining menu over a month: Grilled Salmon (250 sold, $8 cost, $26 price), Classic Burger (300 sold, $5.50 cost, $14 price), Ribeye Steak (60 sold, $11 cost, $36 price), Soup of the Day (50 sold, $3.50 cost, $8 price), and Caesar Salad (180 sold, $4.80 cost, $11 price) — the same five items loaded into the calculator above.
Total units sold across the five items is 840, so the popularity threshold (an equal 1-in-5 share) is 20%. The sales-weighted average contribution margin works out to roughly $11.78. Against those two lines: Grilled Salmon (30% popularity, $18 margin) is a Star. Classic Burger (36% popularity, $8.50 margin) and Caesar Salad (21% popularity, $6.20 margin) are both Plowhorses — high volume carrying a below-average margin. Ribeye Steak (7% popularity, $25 margin) is a Puzzle — the highest margin on the menu, but ordered far less often than an even share. Soup of the Day (6% popularity, $4.50 margin) is a Dog on both counts.
Five items, five different stories — and none of them visible from a single blended food cost percentage for the whole menu.
Common mistakes when running this analysis
- Using too short a sales window.A single slow week or one unusually busy weekend can distort popularity numbers. A 60- to 90-day window is the commonly cited minimum for a read that isn't dominated by one-off noise.
- Comparing contribution margin dollars across very different price points without context. A $25 steak with a $10 margin and a $9 appetizer with a $6 margin can both be worth keeping — the matrix already accounts for this by comparing each item against the weighted average, not a flat dollar target.
- Treating every dog the same way.Some dogs are new items that haven't had time to build sales yet, not genuinely weak performers — check how long an item has actually been on the menu before cutting it.
- Only counting food cost, never labor. A dish with a great contribution margin on paper that also requires disproportionate prep time can still be a net drag — this matrix is a food-cost-driven first pass, not the full picture.
- Running it once and never again. A menu mix that was accurate in summer can be stale by winter, especially for a seasonal concept — see below.
Why seasonal businesses need to re-run this more often
Most menu engineering guidance treats the matrix as a quarterly or even annual exercise. That's a reasonable cadence for a restaurant with flat, year-round demand — it's the wrong cadence for a seasonal concept. A patio restaurant's summer menu mix and its winter menu mix aren't the same population sample; they're two different menus being ordered by two different customer bases at two different volumes. Re-running the matrix at the start of each season, rather than once a year, catches items that are genuinely seasonal dogs (correctly slow in winter, fine to keep) instead of misreading them as year-round underperformers worth cutting.
What to do after you have the four quadrants
The matrix tells you the category; the next tool tells you the fix. For a plowhorse that needs a margin fix without losing volume, run the numbers through the menu price optimizer to see what price would hit your target food cost without guessing. If the margin problem traces back to ingredient cost rather than price, the recipe & plate cost calculatorshows exactly which ingredient is driving the cost up. And since popularity is a volume metric, pair a puzzle's repositioning plan with the table turnover & revenue per seat calculator if the real constraint is limited seats during peak hours rather than low demand for the dish itself.
See understanding food cost percentage benchmarks for why a single blended food cost number can hide exactly this kind of item-by-item variation — the matrix above is the tool that surfaces it.