Guide contents
Menu engineering is an analysis method, not a menu-design mood board. For every dish in one comparable category, it calculates contribution margin per sale and sales mix. Crossing those measures places the dish in one of four quadrants and forces a testable operating decision.
The method in 30 seconds
- Unit contribution margin = comparable selling price − portion food cost.
- Popularity = units of one dish ÷ total units in its category.
- A low food-cost percentage is not enough: the decision also depends on MAD left per sale, volume and kitchen constraints.
01 · InputsFreeze one period and one category before calculating.
Export at least the dish name, units sold, selling price and current portion food cost. Use a representative trading cycle, such as four weeks. Do not blend a Ramadan week, an opening promotion or a one-off event into an ordinary period without marking the distortion.
Analyse starters, mains, desserts and drinks separately. A drink served in seconds and a tagine occupying a station do not compete for the same order or capacity. Split dine-in, takeaway and delivery when prices, packaging or commissions change their economics.
Use the cost of the portion actually sold
Start from a current recipe card with yield and real serving quantity. A six-month-old theoretical cost produces a precise-looking but false matrix. Before the run, verify the five highest-volume dishes and the five most expensive to make. A yield or unit error there moves the most money.
Before opening, you can calculate margins and prepare measurement, but classification must wait for representative sales. The restaurant opening guide shows what to validate before those sales exist.
02 · Two axesCalculate margin in MAD and popularity as a share.
Unit contribution margin is selling price minus portion food cost. A dish sold for 88 MAD with 30 MAD of ingredients contributes 58 MAD toward labour, rent, energy and, only after those costs, operating profit. It is not the dish’s net profit.
Popularity is the dish’s share of sales within its category. If 260 kefta tagines sell out of 900 mains, the share is 28.9%. The classic method sets the popularity threshold at 70% of average share. Four dishes have a 25% average share, so the threshold is 17.5%.
Use weighted average contribution margin for the profitability threshold: total category contribution divided by total category units. In this example, 47,600 MAD divided by 900 dishes gives 52.9 MAD per sale.
Percentage monitors cost. Contribution in MAD funds the restaurant.Menu engineering principle · Kiwi Guides
03 · Worked exampleFour dishes, 900 sales and four different decisions.
| Dish | Calculation | Quadrant |
|---|---|---|
| Chicken tagine | 420 sold · 75 − 27 = 48 MAD · 46.7% mix | Plowhorse |
| Kefta tagine | 260 sold · 88 − 30 = 58 MAD · 28.9% mix | Star |
| Chicken pastilla | 100 sold · 110 − 38 = 72 MAD · 11.1% mix | Puzzle |
| Vegetable couscous | 120 sold · 65 − 22 = 43 MAD · 13.3% mix | Dog |
Kefta clears both thresholds. Chicken tagine sells heavily but contributes less than the weighted average. Pastilla contributes strongly but sells slowly. Couscous sits below both lines. These labels describe one period; they are not permanent judgements about the dishes.
Rebuild the control total: 420 × 48 + 260 × 58 + 100 × 72 + 120 × 43 = 47,600 MAD. If your sheet does not reproduce that total, stop before plotting anything.
04 · ActionsTreat every quadrant as a hypothesis, not an order.
Star: protect before optimising
Keep the recipe, availability and execution stable. Give it clear menu space and test whether a small price move damages demand. A star that regularly runs out or changes by shift loses the qualities that made it valuable.
Plowhorse: repair margin without punishing demand
Start with yield, portion, garnish and supplier terms. Then test a measured price change or a relevant add-on. Removing a loved item can reduce traffic, attached drink sales and regulars’ trust.
Puzzle: reduce the risk of choosing it
Check the name, description, image, placement and server recommendation. If guests do not understand the dish, clearer presentation may reveal demand. If it blocks the kitchen and stays slow after a fair test, its attractive unit margin is not enough.
Dog: find its role before removal
The dish may be seasonal, meet a dietary need, complete a table order or consume ingredients that would be wasted. Measure those effects. If it has no strategic role, demand or margin, rebuild or remove it and watch where its sales migrate.
05 · Operating vetoThe matrix cannot see the pass.
The classic method uses food cost and sales volume. It does not automatically see prep time, scarce skill, oven capacity, refunds, waste, delivery commission, drinks attached to a dish or guest satisfaction. Add those columns before an irreversible decision.
A dish contributing 72 MAD can still be poor if it delays fifteen tickets. A 43 MAD dish can be valuable when it shares mise en place and attracts tables that order drinks and dessert. The matrix closes one financial blind spot; it does not replace the chef or floor team.
06 · RoutineChange one variable, record the date, then measure a full cycle.
- Freeze the period, category and channel.
- Validate costs for the items carrying the most sales value.
- Calculate unit contribution, total contribution, sales mix and both thresholds.
- Add prep time, waste, returns and strategic role.
- Choose one action per dish: price, portion, buying, copy, placement or removal.
- Keep the before state and rerun after a comparable period.
If price, photo, recipe and placement all change together, the result cannot tell you what worked. Good menu engineering is an experiment log, not an annual meeting.
07 · Common questionsThe mistakes that corrupt the matrix.
Should the entire menu be analysed together?
No. Compare genuine substitutes within one category and, where economics differ, one channel. Mixing drinks, desserts and mains creates thresholds with no operating meaning.
Must a Star have the lowest food-cost percentage?
No. It must clear the popularity line and average contribution margin. A higher cost percentage can still leave more MAD when price and demand support it.
Should every Dog be deleted?
No. Check seasonality, strategic role, attached sales and kitchen load. The matrix starts an investigation; it does not automate removal.
Sources and methodSources and method
The formulas come from the references below. The MAD figures are worked Kiwi examples, not market averages.