Menu engineering: how to find your Stars, Plowhorses, Puzzles and Dogs
· 9 min read
Menu engineering is a simple way to answer a hard question: which dishes are earning their place on the menu? It was set out by Michael Kasavana and Donald Smith at Michigan State University in 1982, and it is still the method most restaurant consultants use. You need two numbers for every dish, and about half an hour with a spreadsheet.
The two numbers
- Popularity: how many of this dish you sold, as a share of everything sold in its section. This is called the menu mix.
- Contribution margin: the menu price minus the food cost of one plate. This is the cash each sale contributes towards wages, rent and profit. If you have not costed your dishes yet, start with how to calculate food cost percentage.
The four groups
| Group | Popularity | Profit per plate | In short |
|---|---|---|---|
| Star | High | High | Your best dishes. Protect them. |
| Plowhorse | High | Low | Guests love them, but they earn little. |
| Puzzle | Low | High | Earn well, but too few people order them. |
| Dog | Low | Low | Neither. Rework or remove. |
Where to draw the lines
Both lines are relative to the rest of the section, which is what makes the method useful: it compares dishes with their real alternatives.
- Popular means selling at least 70% of an equal share. With 6 dishes in a section, an equal share is 1 ÷ 6 = 16.7%, so the line is 0.7 × 16.7% = 11.7% of sales.
- Profitable means a contribution margin at or above the section's average, weighted by how many of each dish you sold.
The 70% rule is deliberate. If the line were a full equal share, around half of your dishes would always be called unpopular, even on a well-balanced menu.
A worked example
Here are the mains from one month at a small curry house. Six dishes, 600 plates sold in total:
| Dish | Sold | Mix % | Price | Food cost | Margin | Group |
|---|---|---|---|---|---|---|
| Chicken curry | 180 | 30.0% | $14.00 | $4.20 | $9.80 | Plowhorse |
| Veg biryani | 150 | 25.0% | $12.00 | $3.00 | $9.00 | Plowhorse |
| Butter chicken | 140 | 23.3% | $16.00 | $4.80 | $11.20 | Star |
| Lamb shank | 60 | 10.0% | $24.00 | $9.60 | $14.40 | Puzzle |
| Paneer tikka masala | 40 | 6.7% | $15.00 | $3.75 | $11.25 | Puzzle |
| Fish curry | 30 | 5.0% | $19.00 | $8.55 | $10.45 | Dog |
The popularity line is 11.7%, so the top three are popular and the bottom three are not. For the margin line, multiply each margin by the number sold, add them up and divide by 600: (1,764 + 1,350 + 1,568 + 864 + 450 + 313.50) ÷ 600 = $10.52. Dishes earning $10.52 or more per plate count as profitable.
Notice two things. First, the best-selling dish is not the Star: chicken curry sells most but earns below average. Second, fish curry is a Dog by only 7 cents. Borderline cases like that deserve judgement, not automatic removal.
What to do with each group
Stars: make them easy to find
- Keep them at the top of their section and give them the best description and photo.
- Do not change the recipe or portion without a good reason.
- A small price rise is usually safe. Guests order Stars because they want them, not because they are cheap.
Plowhorses: earn more from each one
- Raise the price a little, often 5–10%. Popular dishes can usually absorb it.
- Re-cost the recipe. Can a cheaper side, a smaller garnish or a better supplier price lift the margin without anyone noticing?
- Pair them with high-margin add-ons: a bread, a drink, a side.
Puzzles: get them noticed
- Move them up the section. On a phone menu, a dish near the bottom of a long list is simply seen by fewer people.
- Rewrite the description so it says what makes the dish worth ordering.
- Add a photo or a badge such as "Chef's pick", and ask staff to recommend it.
- If it still does not move, check the price: it may be too high for what guests think it is.
Dogs: rework or remove
- Ask why it is there. Some dishes have a job: the only vegan main, the kids' option, the dish your regulars would miss.
- If it has no job, take it off. Mark it unavailable for two weeks first and see whether anyone asks for it.
- If you want to keep it, rework it into a Puzzle by raising its margin, or into a Plowhorse by making it more appealing.
Common mistakes
- Comparing across sections. A coffee and a steak should never be judged against each other. Run the analysis per section.
- Too little data. Use at least four weeks of sales. A dish that sold 5 times can swing groups on a single busy table.
- Ignoring seasons. A soup that is a Dog in summer may be a Star in winter.
- Changing everything at once. Make a few changes, wait a few weeks, then measure again, so you know what worked.
- Treating food cost percentage as profit. A high-cost dish can still be your best earner, as the food cost guide shows.
How often to do it
Every three months is a good rhythm for most restaurants, and always after a big supplier price change or a menu rewrite. The first time takes the longest; after that it is mostly updating numbers.