Where the method comes from
Menu engineering was proposed by Michael Kasavana and Donald Smith in 1982. The idea is simple: two things matter about each dish, how much it sells and how much cash margin it makes. Crossing the two gives four groups, and each one calls for a different decision.
The margin that counts here isn’t the percentage but the contribution: the price before VAT minus the ingredient cost. A dish at 40% food cost can leave you more pounds than one at 25%.
What data you need
- Units sold of each dish over a normal period, for example four weeks. They come from your till or EPOS system.
- Selling price before VAT (in the UK, the menu price divided by 1.20; in the US the menu price is already before sales tax).
- Ingredient cost per portion, in other words your recipe costing. If you don’t have it yet, start with how to price a menu and the recipe cost calculator.
Analyse each section separately: starters with starters and mains with mains. Comparing a dessert with a côte de boeuf tells you nothing.
How it’s calculated, step by step
- Margin per dish = price before VAT − cost.
- Weighted average margin = sum of (margin × units) ÷ total units in the section.
- Popularity threshold = (100% ÷ number of dishes) × 70%. With 5 dishes, 14%. That 70% is the original method’s rule: a dish is popular if it sells at least 70% of what it would if every dish sold equally.
- Classify: popular or not, and margin above or below the average.
An example with five mains and four weeks of sales:
| Dish | Sold | Share | Menu price | Cost | Margin | Group |
|---|---|---|---|---|---|---|
| Beef burger | 260 | 43% | 15.00 | 4.20 | 8.30 | Plowhorse |
| Sirloin steak | 180 | 30% | 28.00 | 10.50 | 12.83 | Star |
| Hake with clams | 90 | 15% | 18.00 | 4.48 | 10.52 | Star |
| Squid ink risotto | 50 | 8% | 17.00 | 3.20 | 10.97 | Puzzle |
| Vegetable lasagne | 20 | 3% | 13.00 | 3.10 | 7.73 | Dog |
The weighted average margin comes to £10.20. Look at the sirloin: its food cost is 45% of the price before VAT, higher than any other dish, and yet it’s a star because each one leaves more than £12. The lasagne, at 29% food cost, is the one that contributes least.
What to do with each group
| Group | Sells | Margin | What to do |
|---|---|---|---|
| Stars | A lot | High | Protect the quality and the portion, give them a good position and highlight them. If you touch the price, keep it small. |
| Plowhorses (or workhorses) | A lot | Low | Review the costing: garnish, portion, supplier. Raise the price gradually. Don’t highlight them: they already sell themselves. |
| Puzzles | Little | High | Better name and description, better position, have front of house recommend them. If they don’t take off, move them to the specials. |
| Dogs | Little | Low | Drop them or rework them, unless they play a clear role in the offer. |
In the example: the burger needs a cost review, not a highlight box; the squid ink risotto deserves a better description and the top spot in its section; and the lasagne either gets reworked or makes way for another vegetarian option.
Where customers look (and what we actually know)
It’s often said that the eye goes first to the centre of the page, then to the top right and then to the top left: the famous “golden triangle”. The idea has been around the trade for decades, but the evidence is thin.
An eye-tracking study published in 2012 by Sybil Yang, of San Francisco State University, watched 25 people choose a meal from a two-page menu. They read it like a book, left to right and top to bottom, with no spot where they lingered more. The bottom corners got the least attention.
These are small studies, so no golden rules. The prudent approach:
- Put stars and puzzles at the start of their section.
- Don’t hide what you want to sell in the bottom corners.
- Order the menu the way people eat, so nobody has to hunt.
Highlight without overdoing it
A box, a soft background or a “recommended” label makes a dish jump off the page. It works precisely because it’s the exception: one or two highlights per section at most. If everything is highlighted, nothing is.
In MenuDesign you select the dish and mark it as highlighted. Each design handles it in its own way; Brasserie, for example, uses a double-rule frame. You can also add the “Recommended” label with its icon.
Anchor dishes
A clearly more expensive dish in a section, such as a côte de boeuf for two or a seafood platter, shifts the customer’s reference point: next to it, everything else looks reasonable. It’s a well-known pricing device, with two conditions: it has to be a real dish you serve well, and its price has to make sense. If nobody ever orders it, at least make sure it doesn’t create waste.
Descriptions that sell
The description is the cheapest tool for moving a puzzle:
- Specific: ingredient, technique and accompaniment. “Squid ink risotto with cuttlefish and lemon aioli” says more than “Black risotto”.
- Origin or season if it adds something: “Isle of Wight tomatoes in season”.
- No empty adjectives: “delicious”, “exquisite” or “lovingly prepared” sell nothing.
- Honest: if you write “line-caught”, “wild” or “homemade”, it has to be. A misleading description is a problem with your customers and with trading standards.
Repeat the analysis
Redo the matrix when you change the menu, when you raise prices and, at the very least, every season. Compare it with the previous one: is the puzzle you moved selling more? Did the plowhorse survive the price rise?
Once you’ve made your decisions, take them to the menu. In MenuDesign you reorder dishes by dragging, mark the highlights and download the PDF in a couple of minutes. For the rest of the design, see restaurant menu design.
Frequently asked questions
What is menu engineering?
A method for analysing a menu by crossing two facts about each dish: how many you sell and how much cash margin each one makes. Michael Kasavana and Donald Smith proposed it in 1982, and it sorts dishes into four groups: stars, plowhorses, puzzles and dogs.
What sales data do I need?
The number of each dish sold over a representative period, for example four normal weeks, taken from your till or EPOS system. A single weekend, a bank holiday or a month of roadworks outside give misleading results.
Does the golden triangle work?
It's a much-repeated idea, but an eye-tracking study published in 2012 didn't find that hot spot: participants read the menu like a book. The studies are small, so the prudent thing is to highlight few dishes and put them at the start of their section.
Do I have to drop every dog?
Not necessarily. If a dish that sells little and makes little brings in a type of customer or rounds out the offer, such as a vegan option or a children's dish, it can stay. What makes no sense is keeping it just out of habit.