The lamb dish sells eleven a night and everybody mentions it. It carries a 58% gross profit, which is respectable, and it takes fourteen minutes of one chef's attention because of the garnish. The chicken sells nine a night, carries 74%, and takes four minutes. On a Saturday, the lamb is not your best dish. It is the reason the pass backs up at half past eight and the reason your labour cost looks the way it does.
Menu engineering is the exercise that makes this visible. It takes about a day, it needs no software, and most kitchens have never done it.
Two numbers per dish, and one of them is not a percentage
For every item on the menu you need the number sold over a defined period, ideally at least a month so that one bad week does not distort it, and the cash gross profit per portion.
Cash gross profit is the selling price excluding VAT, minus the food cost of the portion as you actually serve it. The exclusion of VAT matters: restaurant food eaten on the premises is standard rated at 20%, so a £24 dish is £20 of revenue. Working from the menu price inflates every margin by a fifth and makes the whole exercise useless.
Food cost means the full recipe, costed from current invoices, including the things people leave out: the oil, the butter that finishes the sauce, the garnish, the bread, the trim you cannot use, and a realistic yield allowance on anything you break down yourself. A whole fish yielding 45% costs nearly twice the per kilo price once it is on a plate.
Then the point nobody enjoys. Percentage gross profit is a useful control ratio, and it is not what you bank. A dish at 80% GP selling at £9 contributes £6 per portion. A dish at 65% selling at £26 contributes £14.08. You cannot pay rent with a percentage. Rank the menu by cash contribution as well as by percentage and the picture usually changes.
The four quadrants, and what each one is telling you
Plot every dish with popularity on one axis and cash contribution on the other, splitting each at the menu average. The resulting grid is the classic menu engineering matrix associated with Kasavana and Smith, and the four groups have names that have stuck.
High popularity, high contribution. The dishes that carry the business. Protect them absolutely: consistent recipe, consistent supplier, prominent placement, and no quiet price rise that risks the volume. These are also the dishes to watch for supplier price movement, because a four pence rise on an ingredient in your bestseller costs more than any change elsewhere on the menu.
High popularity, low contribution. Ordered constantly and earning little. The temptation is to raise the price. The better first move is to reduce the cost without the guest noticing: change the garnish, adjust the portion of the expensive component while increasing something cheap, or renegotiate that one ingredient specifically. A small price rise on a very popular dish is also the most effective single change available to you, precisely because the volume multiplies it.
Low popularity, high contribution. It earns well and nobody orders it. This is a description and positioning problem. Move it up the section, give it a better description, brief the floor to recommend it, or photograph it if you use a digital menu. Servers sell what they have tasted and what they can describe, which is a training decision rather than a menu decision.
Low popularity, low contribution. Rarely ordered and barely profitable, while still occupying stock, fridge space, prep time and staff memory. These come off, unless they are on the menu for a defensible reason such as a dietary requirement that keeps a table of four from going elsewhere.
Repositioning beats deleting
The instinct after this exercise is to cut. Resist most of it. Every dish removed is a guest whose favourite has gone, and a shorter menu only saves money if it also shortens the purchase list.
The cheaper interventions, in rough order of effect:
- Reprice the popular low-contribution dishes by a small amount. Volume does the work.
- Recost and reformulate rather than remove, adjusting the component that carries the cost.
- Change placement. Guests read menus in predictable patterns, spending most attention at the top of a section and around any item that is visually separated. Moving a high-contribution dish into that position changes its sales.
- Write better descriptions. Specificity sells: the farm, the cut, the method. Vagueness does not.
- Brief the floor weekly on two dishes to recommend, and check afterwards whether the mix moved.
- Cut the genuine dead weight, and cut the purchase lines that go with it.
A menu of a manageable size, where every item shares ingredients with several others, is both easier to run and more profitable than a long one. Cross-utilisation is the underrated lever: the same braised shoulder appearing in two dishes and a lunch special carries far less waste risk than three separate proteins.
Contribution per minute of kitchen time is the second lens
Cash contribution assumes every dish costs the same to produce. It does not. A kitchen on a Saturday is capacity constrained, and the binding constraint is chef minutes at the section, not food cost.
Take the cash contribution of each dish and divide it by the minutes of skilled labour it needs during service, excluding prep that happens in the morning. That number ranks dishes by what they earn per unit of your scarcest resource at your busiest time. Some elegant, well-margined dishes rank badly and are quietly capping the number of covers you can serve.
This does not mean removing them. It means knowing which dishes to push on a quiet Tuesday and which to feature when you expect a full room, and it means moving work into prep where you can. A dish that takes fourteen minutes on the pass at eight o'clock and four minutes if the element is prepped at eleven in the morning is a different dish commercially.
Run the whole exercise monthly once you have done it once. The inputs move: supplier prices change, the sales mix drifts, and a change you made six weeks ago either worked or did not. Menu engineering is not a project, it is a habit, and the kitchens that do it quarterly tend to discover a problem that has been running for a quarter.
Where this shows up in the game
Restaurants are part of Business Baron's full game, and a café or a pub is free to run. The game calculates gross margin from actual stock movements rather than list prices, and supplier costs move as the years pass, so a dish that made money when you priced it can quietly stop. The one everybody orders can be the one holding the business back.