Menu Engineering: Which Dishes Actually Make Money

A busy restaurant can still have a menu that quietly works against it. The room is full, the tickets keep coming, the section that sells hardest is the one everyone talks about — and the bank balance does not move the way the covers suggest it should.

Most of the time the reason is not dramatic. It is that “popular” and “good for the business” are not automatically the same thing. A dish can sell in volume and leave very little behind after what it costs to put on the plate. Another can leave a great deal behind and barely get ordered, because it sits at the bottom of a column where nobody reads.

Neither of those is visible from the pass. Both are visible from the numbers — and the numbers involved are simpler than most owners expect. By the end of this article you will know what data to gather, what to calculate from it, and what safe action to consider once you have an answer. Nothing here needs a new till system or a consultant in the building.

1. Start with the numbers you actually need

Menu analysis fails far more often from missing data than from bad arithmetic. Before anything else, you need six things for every dish you intend to look at.

Menu selling price. What the guest actually pays today, not what the menu said last spring.
Direct portion or ingredient cost. What the ingredients in one served portion cost you, at the prices you are paying now.
Units sold. How many of that dish left the kitchen.
A defined reporting period. A named window — last month, the last twelve weeks — used consistently for every dish in the comparison.
Menu section or category. Starter, main, side, dessert, hot drink, and so on.
Service channel, where it is relevant. Dine-in, takeaway and delivery are different businesses wearing the same menu.

The last three are the ones people skip, and they are the ones that decide whether the exercise is worth doing. A dish is only ever informative next to a genuinely comparable dish: starters with starters, mains with mains, coffees with coffees, and dine-in with dine-in. Comparing a £4 flat white against a £22 main tells you nothing except that they are different products. Comparing a delivery burger against a dine-in burger tells you even less, because the delivery version carries commission the dine-in version never sees.

The six figures to gather

Menu selling priceWhat the guest pays today
Direct portion costIngredients in one served portion
Units soldHow many left the kitchen
Reporting periodThe same window for every dish
Menu sectionStarter, main, dessert, drink
Service channelDine-in, takeaway, delivery

Compare like with like: mains with mains, coffees with coffees, and dine-in with dine-in.

2. Contribution margin: what each sale leaves after direct food cost

The single most useful number on a menu line is not a percentage. It is a cash figure: what one sale of that dish leaves behind once you have paid for the ingredients that went into it.

Contribution margin = selling price − direct food cost

That is the whole calculation. Here it is on an invented dish, with figures chosen to be easy to follow rather than to represent any real business:

Worked example — invented figures

Menu price£16.00

Direct ingredient cost£4.80

Contribution margin£11.20

Direct food cost £4.80 — 30% of the price
Contribution margin £11.20

Contribution margin is not profit. That £11.20 still has to contribute toward labour, rent, utilities, marketing, waste, delivery commissions, taxes and every other overhead before anything is left over.

That distinction matters more than any other point in this article, so it is worth repeating plainly. Contribution margin is not profit. It is the amount each sale contributes toward covering everything that is not the food itself. Whether the business makes money depends on how many of those contributions you collect in a week and how large the pile of overheads waiting for them is. A dish with a healthy contribution margin that sells twice a month is not carrying anything.

On VAT. Where VAT applies, work these calculations on sales excluding VAT, not on the gross till total — otherwise the selling price you are dividing by includes tax you never keep, and every percentage comes out flattering. VAT treatment depends on the supply and on your business circumstances, and this article is not tax advice. Confirm how it applies to you with your accountant before you rebuild a pricing model on it.

3. Why food-cost percentage alone can mislead you

Food-cost percentage is the same two numbers expressed differently: direct food cost divided by selling price. In the example above, £4.80 ÷ £16.00 gives 30%. It is genuinely useful — it is how you spot a portion that has drifted, a supplier price that has moved, or a dish that was priced years ago and never revisited.

What it cannot do is tell you, on its own, whether a dish is commercially strong. A percentage has no size. Two dishes can have identical percentages and leave wildly different amounts of cash behind, and a lower percentage can sit on a smaller cash contribution.

Two invented dishes, side by side

Dish A — lower food-cost percentage

Menu price: £12.00
Direct cost: £3.00
Food cost: 25%

25% food cost

Contribution margin

£9.00

Dish B — higher food-cost percentage Larger margin

Menu price: £18.00
Direct cost: £5.40
Food cost: 30%

30% food cost

Contribution margin

£12.60

Dish A looks better on the percentage and leaves £9.00. Dish B looks worse on the percentage and leaves £12.60 — £3.60 more, every single time it sells. Lower food cost does not automatically mean a stronger contribution margin.

Both dishes are invented for illustration. They are not drawn from any business.

So you need both measures, and they answer different questions. The percentage answers “is this dish costing what I think it costs?” — it is a control number, best read over time against its own history. The contribution margin answers “how much does this sale actually leave me?” — it is a comparison number, best read against the other dishes in the same section. Used together they tell you something. Used alone, either one will eventually send you the wrong way.

About the benchmark you will have been quoted

Somebody will have told you that food cost should be about 30% of the menu price. Published ranges vary more than that suggests, and it is worth knowing what the sources actually say before you treat any figure as a target. One widely-read industry guide puts the overall band at 25–40% and notes that most restaurants aim for around 30%, with quick service lower and fine dining higher.[1] A UK-focused source puts pub food and casual dining at 28–35% measured against sales excluding VAT.[2] A general full-service figure of roughly 25–30% of menu price is commonly cited.

Read those as context, not as a target for your business. What a sensible range looks like for you varies by concept, location, supplier costs, labour model, waste and menu positioning. A wine-led room with a small kitchen and a high-volume deli with three staff on the pass cannot sensibly be held to the same number. If your figure sits outside a published range, that is a prompt to go and look at portioning, supplier pricing and pricing itself — not evidence that anything is wrong.

4. Add popularity: what customers actually choose

Contribution margin tells you what a sale is worth. It says nothing about how often that sale happens. For that you need sales mix — the share of a comparison group that one dish takes.

Sales mix = units sold ÷ total units sold in the same comparison group

The phrase doing the work there is the same comparison group. A dish has to be judged against dishes in the same menu section, the same service channel and the same reporting period. Change any one of those and the share means something different. Twelve burgers a week is an impressive number in a wine bar and an alarming one in a burger restaurant; the figure only becomes information once you know what it is a share of.

Sales mix — invented example: mains, dine-in, last 30 days

Burger: 80 sold
Chicken dish: 55 sold
Pasta dish: 40 sold
Salad dish: 25 sold

Total comparable mains sold200 units

80 sold ÷ 200 comparable mains = a sales mix of 40%

Only compare dishes in the same section, channel and period. Change any one of the three and the share is answering a different question.

Invented figures for illustration. Not client data and not MenuMargin output.

One more caution before you act on any of this: incomplete data should not create false certainty. If half your delivery orders never reached the report, if a dish was off the menu for a fortnight, if the kitchen ran a special under the same button — the share you are looking at is describing something other than what you think. Check how complete the period is before you draw a conclusion from it. A partial month that looks decisive is more dangerous than a full month that looks ambiguous.

5. The menu-engineering view: four useful patterns

Put contribution margin on one axis and popularity on the other, and every dish lands in one of four positions. This is the classic menu-engineering matrix, and it is genuinely useful — provided you treat it as a way of grouping questions rather than a machine that issues instructions.

The four patterns

Read across for contribution margin, lower to stronger. Read up for popularity, lower to stronger.

Guests already want it, so the demand question is answered. The open questions are on the cost side: has the portion drifted, has a supplier price moved, is there waste in the prep, and is the price still right for what it now costs? A popular dish is also the one where a small, careful pricing change reaches the most guests — which cuts both ways.

The dish is working on both axes, so the job is not to improve it but to avoid breaking it. Keep the ingredients available, keep the specification consistent between shifts, and keep it where guests can find it on the menu. Stars are usually damaged by neglect and by well-meant tinkering, not by competition.

This is the quadrant where owners are most often told to delete something, and it is the quadrant where deleting something is most likely to be a mistake. Ask what the dish is for first. It may be the only option for a guest who would otherwise take the table elsewhere, it may carry the brand, it may be seasonal, or it may simply be badly written on the menu. Understand its role, then decide.

The economics are already good; the dish is not being chosen. That is usually a visibility or a confidence problem rather than a recipe problem. Where does it sit on the page, how is it described, does the floor team know what is in it and who it suits, and is there a pairing or a recommendation that would put it in front of the right guest?

This is a decision framework, not an automatic instruction. Consider guest expectations, brand, operations, seasonality and menu balance before changing a dish. Dish names are invented for illustration.

Two things this framework is not. It is not a verdict, and the fourth quadrant is not a bin. I have deliberately avoided the older labels for that group, because “dog” tells you what to do before you have asked why the dish is there. A dish in the weaker-on-both-axes position might need a price review, a smaller portion, a better description, a different position on the page, a word from the floor team, a new supplier, a rewritten recipe — or nothing at all, because it is doing a job the numbers do not capture.

Classification is where the conversation starts. It is not where it finishes.

6. What to do next — without making risky menu changes

The fastest way to damage a working menu is to act on a first analysis across every section at once. Here is the low-risk order.

Protect the strong dishes. Availability, consistency and visibility, in that order. If a dish is working on both axes, the most valuable thing you can do is make sure it is never off, never inconsistent, and never buried by a redesign.
Investigate the popular, lower-margin dishes. Check the portion against the specification, check what the supplier is actually charging now against what your costing assumes, check waste and yield on the prep, and only then consider a small pricing adjustment.
Improve visibility for stronger-margin dishes that sell less. Placement on the page, a description that makes the dish easy to picture, a photograph where the format allows it, floor-team knowledge, and a sensible pairing all move a dish that guests were simply not noticing.
Review weaker items carefully before changing anything. Guest expectations, brand identity, operational complexity, seasonality and menu balance are all real considerations that never appear in a margin column.
Test one change at a time, and give it a meaningful period. If you change three things in a week you will not know which one moved the figures — and normal week-to-week variation will look like a result.

None of this replaces your accountant, your head chef, your EPOS reporting or your own commercial judgment. It is a way of deciding where to point them.

7. A simple monthly menu review routine

Done once, this is an interesting afternoon. Done monthly, it becomes the thing that catches a supplier increase in week three instead of month seven. Five steps.

The monthly routine

Gather item-level salesExport from your EPOS, or count from the tickets. Same window every month.
Update direct portion costsAgainst current invoices, not last year's costing sheet.
Group comparable dishesBy section, by channel, by period.
Review contribution margin and sales mixBoth together. Neither on its own.
Test one small change and measure itOne change, one meaningful period, then compare.

The whole routine is an hour a month once the costing sheet exists. The first month is the long one.

Ready to review your own menu?

Scan a menu and see where the questions are.

Scan your menu with MenuMargin

Use MenuMargin to start the conversation

MenuMargin is the tool I built for the first part of this job: scanning and reviewing menu information without retyping it into a spreadsheet. You can add your real portion costs and sales figures to it, and it will help surface the areas worth attention — what to protect, what to investigate, where to look again at pricing, where to look again at portions, where visibility might be the issue, and where a small test is worth running.

Two honest limits. What it shows you depends entirely on the figures you give it: incomplete costs or a partial sales period produce a confident-looking output that is describing the wrong month. And it is an aid to a decision, not the decision — it belongs alongside your operational judgment, your kitchen's view and your accountant's, not instead of them.

Start with one section of your menu

Mains, or coffees. One section, one period, one comparison.

Scan your menu with MenuMargin

A dish nobody can find has a different problem to a dish that sells badly — if that is the question, start with how restaurants get found on Google, Maps and AI search. If you would rather work through the margin side with someone who has run the kitchens as well as the spreadsheets, that is what the menu and operational work is for, and you can get in touch here. For the numbers that sit around the menu — prime cost, labour, break-even — the companion piece is the financial metrics every restaurant owner should track. Everything else I have written is on the blog.

Frequently asked questions

What is a good food-cost percentage for a restaurant in the UK?

There is no single figure that fits every business. Published ranges differ: one widely-read industry guide puts the overall band at 25–40% with around 30% as a common aim, and a UK-focused source puts pub food and casual dining at 28–35% of sales excluding VAT. A general full-service figure of roughly 25–30% of menu price is commonly cited. Treat any of those as context, because what is sensible for you depends on your concept, location, supplier costs, labour model, waste and menu positioning. The more useful comparison is your own percentage against its own history, dish by dish.

What is contribution margin on a menu item?

Contribution margin is the selling price of a dish minus its direct food or ingredient cost. On a £16.00 dish costing £4.80 in ingredients, the contribution margin is £11.20. It is not profit: that £11.20 still has to contribute toward labour, rent, utilities, waste, marketing, any delivery commission, taxes and the rest of your overheads.

How do I know which dishes are most popular?

Count units sold per dish over a defined period, then express each as a share of the total units sold in the same comparison group — the same menu section, the same service channel, the same period. That share is the sales mix. Most EPOS systems will export item-level sales; if yours does not, a manual count over a representative period will do the same job. Check the period is complete before you read anything into it.

Should I remove a menu item that sells poorly?

Not automatically, and not as a first move. A dish that sells poorly may be badly described, badly placed on the page, unknown to the floor team, seasonal, or the only option that suits part of your audience. Look at pricing, portioning, description, placement, staff recommendation, menu design, the recipe and the supplier price before you look at removal — and consider what role the dish plays in guest expectations, brand and menu balance. Removal is a valid answer; it is rarely the first one.

How often should a restaurant review menu pricing?

Review the underlying costs monthly, because supplier prices move whether or not you are watching. Changing prices is a separate decision and usually a slower one — guests notice frequent movement, and you need a meaningful period after any change to see what it did. A practical rhythm is a monthly costing review, with pricing changes made deliberately and one section at a time.

Sources

TouchBistro, “Menu Pricing: How to Calculate Food Cost Percentage” — touchbistro.com. Checked 6 September 2026. ↑
Jelly, “How to Calculate Restaurant Food Cost Percentage” — blog.getjelly.co.uk. Checked 6 September 2026. ↑

Last checked: 6 September 2026. Every worked example, dish name and figure used to illustrate a method in this article is invented and labelled as such. Nothing on this page is tax, accounting or investment advice.