On any menu, the dishes that sell best are not necessarily the ones that earn the most. Menu engineering starts from that observation and offers a simple method: rank every dish on two criteria, how popular it is and how much margin it leaves. The result fits in a four-quadrant matrix, and each quadrant calls for a different decision.
Where menu engineering comes from
The method was set out in 1982 by Michael L. Kasavana and Donald I. Smith, in their book Menu Engineering: A Practical Guide to Menu Analysis. Its core idea: judge a dish on the margin it brings in by value, not on its food cost percentage.
UK operators will recognise the debate as cash GP versus GP %. The difference is not academic. A dish at 45% food cost with a high price can leave more gross profit per plate than a dish at 25% with a low one. Managing on the percentage alone pushes you towards dishes that are cheap to make, not necessarily the ones that pay the bills. The worked example below shows how.
The two axes of the menu engineering matrix
Popularity: each dish’s share of sales
Count how many times each dish sold over a period, within a single category (mains against mains, desserts against desserts), then work out its share of the total. That is the sales mix.
The threshold is not the average, but 70% of the share each dish would have if sales were split evenly. With 8 dishes, an even share is 12.5%, so a dish counts as popular from 8.75% of sales. That allowance stops dishes that sell perfectly well from being labelled unpopular simply because sales are never spread evenly.
Margin: what each dish leaves after food cost
For each dish, the contribution margin is the selling price excluding VAT minus the food cost of one portion — its cash gross profit. The threshold is the sales-weighted average margin: the total margin the category generates, divided by the number of dishes sold. A dish is profitable if it does at least as well as that average.
That figure is only as reliable as the food cost behind it. The calculation itself is covered in how to calculate food cost; it also needs current prices. A cost based on prices from six months ago can be enough to move a dish to another quadrant, which is why a recipe costing sheet at real prices matters.
The four categories: stars, plowhorses, puzzles and dogs
| Category | Popularity | Margin | Typical decision |
|---|---|---|---|
| Star | High | High | Protect |
| Plowhorse | High | Low | Improve the margin |
| Puzzle | Low | High | Sell it better |
| Dog | Low | Low | Remove or replace |
Stars: protect what works
Popular and profitable, these are the dishes that carry the menu. Leave the recipe and the portion alone, and keep them in a prominent spot on the menu. A small price increase is sometimes possible. Above all, watch their ingredients: a supplier increase on a star costs more than anywhere else, because it repeats on every plate. That is what price drift is about.
Plowhorses: work on the margin without scaring guests off
Guests order them, but they earn less than the average. Taking them off would be a mistake. Improve their margin in small steps instead: trim the portion slightly, substitute an ingredient, renegotiate the main product or switch supplier, nudge the price up, or pair them with a more profitable side.
Puzzles: sell what pays
Profitable but rarely ordered. The obstacle is often visibility or understanding: a better spot on the menu, a more appealing description, a recommendation from front-of-house staff, a daily special. If nothing moves after one or two periods, the price may be too high for what guests perceive, or the dish may simply not belong on this menu.
Dogs: remove or replace
Low sales and low margin make them candidates for removal. Every item you drop simplifies mise en place, ordering and stock, and lowers the risk of waste on slow-moving products — a direct lever for reducing food waste. One fair exception: a dish that plays a role (vegetarian option, children’s menu, house signature) can stay, as long as you keep it knowingly.
Worked example: six mains over a month
Take six main courses, sold 1,300 times in total over the month. The figures are hypothetical and carry no currency.
| Dish | Sold | Share of sales | Price excl. VAT | Food cost | Margin per dish | Food cost % | Category |
|---|---|---|---|---|---|---|---|
| House burger | 420 | 32.3% | 18.00 | 6.30 | 11.70 | 35.0% | Plowhorse |
| Sea bass fillet | 180 | 13.8% | 26.00 | 8.10 | 17.90 | 31.2% | Star |
| Porcini risotto | 90 | 6.9% | 21.00 | 5.20 | 15.80 | 24.8% | Puzzle |
| Rib-eye steak | 300 | 23.1% | 29.00 | 13.50 | 15.50 | 46.6% | Star |
| Roast chicken | 260 | 20.0% | 17.00 | 4.80 | 12.20 | 28.2% | Plowhorse |
| Tuna tataki | 50 | 3.8% | 22.00 | 9.40 | 12.60 | 42.7% | Dog |
The two thresholds:
- Popularity: six dishes, so an even share is 16.7%; the threshold is 70% × 16.7% = 11.7% of sales.
- Margin: the category generates 18,010 of total margin over 1,300 dishes sold, a sales-weighted average margin of 13.85.
What the matrix tells you:
- The rib-eye has the highest food cost % on the menu (46.6%), yet it is a star: it leaves 15.50 per plate and sells well. “Fixing” it to bring its percentage down would be a mistake.
- The roast chicken shows a flattering 28.2% food cost (a 71.8% GP), but its margin is below average: it is a plowhorse. Given its volume, a small change pays off directly: at the same sales, 1.00 more on its price excluding VAT adds 260 of margin over the month.
- The risotto is a puzzle: 15.80 of margin per plate, but only 6.9% of sales. Promote it before drawing conclusions.
- The tuna tataki combines low sales with a below-average margin: first in line to be replaced.
The limits of the matrix
- Compare like with like. Analyse one category at a time (starters, mains, desserts), and ideally one service and one site at a time: a business lunch and a Saturday dinner do not have the same mix.
- Current costs or nothing. The matrix amplifies food cost errors. If recipe costs do not follow purchase prices, the ranking is wrong, and the gap between theoretical and actual cost skews it further (see actual vs theoretical food cost).
- Contribution margin is not the whole of profitability. Prep time, labour, an overloaded section in the kitchen: the method does not see them. A dish that looks very profitable on paper but takes ages to plate deserves a second look.
- A snapshot, not a film. Seasonality, a menu change, an ingredient price rise: the ranking moves. Re-running the analysis with every new menu, and whenever a major ingredient drifts, beats an annual exercise.
Menu engineering in Eqolux
The slow part of a menu engineering matrix is not the maths: it is gathering accurate food costs and reliable sales, dish by dish. Eqolux’s Menu Engineering module starts from your purchases:
- recipe sheets costed at the prices you actually paid, recalculated with every new invoice;
- the exact food cost, dish by dish, linked to the products you buy;
- sales and a dish performance matrix that crosses margin with popularity to decide what to push, rework or retire;
- dishes whose margin is eroding, flagged when an ingredient drifts, before the month-end close.
Key takeaways
- Menu engineering ranks each dish by popularity and by margin in value, within a single category.
- A dish is popular if it reaches 70% of an even share of sales; it is profitable if its margin reaches the sales-weighted average.
- Stars to protect, plowhorses to make more profitable, puzzles to sell better, dogs to remove or replace.
- Food cost % on its own can point at the wrong culprit: it is the cash margin per plate, multiplied by sales, that makes the result.
- The matrix is only as good as its costs: without recipe costs at real prices, it ranks your dishes on yesterday’s costs.
Want to see your own menu in the matrix, with your own purchase prices? Explore Eqolux menu engineering or book a demo.


